Thesis Maher e 2

Overcoming Controllable and Uncontrollable Work

Difficulties: Change Environment or Self?

Elise Maher, B.Sc. (Hons)

Submitted in fulfillment of the requirements for the degree of

Doctor of Philosophy

Deakin University, December 2002

DEAKIN UNIVERSITY

CANDIDATE DECLARATION

I certify that the thesis entitled “Overcoming Controllable and Uncontrollable Work

Difficulties: Change Environment or Self?”

submitted for the degree of Doctor of Philosophy is the result of my own work and

that where reference is made to the work of others, due acknowledgment is given.

I also certify that any material in the thesis which has been accepted for a degree or

diploma by any other university or institution is identified in the text.

Full Name Elise Catherine Maher

Signed ..................................................................................……………….

Date......................................................................................……………….

ii

Acknowledgements

First and foremost, I would like to acknowledge my supervisor, Professor

Robert Cummins. Bob offered endless support, guidance, and encouragement

throughout my studies. His dedication to research and his tremendous work ethic

motivated and inspired me. I am most grateful that he challenged me and allowed

me the freedom and respect to develop my own ideas and theories.

Second, I must thank all the organisations that allowed me to enter their

workplaces, and all the participants involved in the studies. I especially want to

thank staff at Australian Unity, members of the Australian Centre of Quality of Life,

and the hundreds of supermarket workers, teachers and academics that assisted me.

These people invested time and energy into completing my survey purely for the

benefit of helping others. I am so appreciative of their efforts and I am determined to

share the knowledge that I have gained from them.

Third, I would like to acknowledge the academic and administrative staff, and

fellow students at Deakin University. I especially want to acknowledge Rose-Anne

and Helen (my surrogate parents) for taking me under their wings. Their delightful,

vibrant personalities made work times pleasurable. Also, I would like to

acknowledge Carolyn and Catherine for always being there to listen and share.

Finally, I would like to acknowledge my family and friends. My parents,

John and Frances, and my brothers have provided endless support throughout the last

seven years. I also want to acknowledge Lauren, Taylah, Tyson and Buffy for

always lighting up my life. My partner, Tim Davis, has been my tower of strength,

and I am forever grateful for his love and support.

iii

Elise

iv

Table of Contents

ACKNOWLEDGEMENTS.......................................................................................................................III

LIST OF TABLES...............................................................................................................................XIII

LIST OF FIGURES..............................................................................................................................XVI

LIST OF APPENDICES.......................................................................................................................XVII

ABSTRACT......................................................................................................................................XVIII

CHAPTER 1 - LITERATURE REVIEW............................................................................................1

1.1 ABSTRACT.............................................................................................................................2

1.2 JOB SATISFACTION................................................................................................................3

1.2.1 Theories of Job Satisfaction: Environmental and Dispositional Predictors...................3

1.3 MASLOW’S (1954, 1970) HIERARCHY OF NEEDS..................................................................5

1.3.1 Need Hierarchy Theory....................................................................................................5

1.3.2 Applying Maslow’s (1954, 1970) Theory to Organisations.............................................6

1.3.3 Criticisms of Maslow’s (1954, 1970) Need Hierarchy Theory........................................7

1.3.4 Conclusion......................................................................................................................20

1.4 HERZBERG, MAUSNER AND SNYDERMAN’S (1959, 1993) TWO-FACTOR THEORY OF JOB

SATISFACTION....................................................................................................................................22

1.4.1 How the Two-Factor Theory has Contributed to our Understanding of Job

Satisfaction...................................................................................................................................22

1.4.2 Development of the Two-Factor Theory........................................................................22

1.4.3 Criticisms of Herzberg et al’s., (1959) Theory..............................................................24

1.4.4 Conclusion......................................................................................................................33

1.5 VROOM’S (1964) EXPECTANCY THEORY OF JOB SATISFACTION........................................34

1.5.1 How Expectancy Theory has Contributed to our Knowledge of Job Satisfaction.........34

1.5.2 Description of Expectancy Theory.................................................................................34

v

1.5.3 Applications of the Valence Model................................................................................36

1.5.4 Studies of the Valence Model.........................................................................................36

1.5.5 Methodological Limitations...........................................................................................37

1.5.6 Conclusion......................................................................................................................43

1.6 DISCREPANCY THEORIES.....................................................................................................44

1.6.1 How Discrepancy Theories have Contributed to our Knowledge of Job Satisfaction...44

1.6.2 Description of Discrepancy Theories.............................................................................44

1.6.3 Empirical Studies Investigating Discrepancy Theories.................................................44

1.6.4 Theoretical Problems with Discrepancy Theories.........................................................45

1.6.5 Conclusion......................................................................................................................46

1.7 JOB CHARACTERISTICS MODEL (JCM; HACKMAN & OLDHAM, 1976)..............................47

1.7.1 How the Job Characteristics Model has Contributed to our Knowledge of Job

Satisfaction...................................................................................................................................47

1.7.2 Description of the Job Characteristics Model...............................................................47

1.7.3 Empirical Studies of the Model......................................................................................50

1.7.4 Conclusion......................................................................................................................56

1.8 JOB DEMAND-CONTROL MODEL (KARASEK, 1979; KARASEK & THEORELL, 1990)..........57

1.8.1 How the Job Demand-Control Model Contributes to our Understanding of Job

Satisfaction...................................................................................................................................57

1.8.2 Description of the Job Demand-Control Model.............................................................57

1.8.3 Empirical Studies of the Job Demand-Control Model...................................................59

1.8.4 Conclusion......................................................................................................................62

1.8.5 Extensions on the Job Demand-Control Model.............................................................63

1.8.6 Addressing the “Gaps” in the Job Demand-Control Model..........................................63

vi

1.9 DEVELOPMENT OF A NEW EXPLANATION FOR THE RELATIONSHIP BETWEEN JOB

AUTONOMY AND JOB SATISFACTION: INFLUENCING EMPLOYEES’ RESPONSES TO WORK

DIFFICULTIES.....................................................................................................................................65

1.9.1 a) Primary Control Strategies and Secondary Control Strategies...............................66

1.9.2 b) Amounts of Primary Control and Secondary Control...............................................68

1.9.3 c) Which Control Strategies are more Adaptive for Employees?..................................69

1.9.4 Summary.........................................................................................................................75

1.10 EXPLAINING THE RELATIONSHIP BETWEEN JOB AUTONOMY AND JOB SATISFACTION: HOW

JOB AUTONOMY INFLUENCES PRIMARY AND SECONDARY CONTROL..............................................75

1.10.1 1) Use of Primary and Secondary Control................................................................76

1.10.2 2) Adaptiveness of Primary and Secondary Control................................................77

1.10.3 Summary....................................................................................................................86

1.11 OTHER MAJOR PREDICTORS OF JOB SATISFACTION...........................................................87

1.11.1 Personality................................................................................................................87

1.11.2 Life Satisfaction.........................................................................................................90

1.12 MODEL OF JOB SATISFACTION............................................................................................97

2 CHAPTER 2 - STUDY ONE..........................................................................................................100

2.1 ABSTRACT.........................................................................................................................101

2.2 PROPOSAL FOR STUDY ONE..............................................................................................102

2.2.1 Identifying Employees with Low/High Job Autonomy.................................................102

2.3 AIMS AND HYPOTHESES....................................................................................................104

2.4 METHOD............................................................................................................................108

2.4.1 Participants..................................................................................................................108

2.4.2 Materials......................................................................................................................108

2.4.3 Procedure.....................................................................................................................114

vii

2.5 RESULTS............................................................................................................................116

2.5.1 Data Screening and Checking of Assumptions............................................................116

2.5.2 Descriptive Statistics and Inter-Correlations..............................................................118

2.5.3 Factor Analyses............................................................................................................120

2.5.4 Factor Analysis of the Job Descriptive Index..............................................................120

2.5.5 Factor Analysis of the Primary and Secondary Control Scale....................................122

2.5.6 Factor Analysis of the Job Autonomy Scale.................................................................125

2.6 HYPOTHESIS TESTING........................................................................................................127

2.6.1 Hypothesis One- Assumption Testing...........................................................................127

2.6.2 Hypothesis Two- Occupational Differences in the Use of the Control Strategies......128

2.6.3 Hypothesis Three- Examining how Job Autonomy Relates to the Control Strategies. 130

2.6.4 Hypothesis Four- Examining how Job Autonomy Influences the Adaptiveness of the

Control Strategies.......................................................................................................................131

2.6.5 Hypothesis Five- Does Job Autonomy Moderate the Relationship Between the Control

Strategies and Job Satisfaction?................................................................................................132

2.6.6 Hypothesis Six- Do the Control Strategies Mediate the Relationship Between Job

Autonomy and Job Satisfaction?................................................................................................137

2.6.7 Hypothesis Seven- Occupational Differences in Job and Life Satisfaction.................142

2.6.8 Hypothesis Eight- Predictors of Job Satisfaction........................................................146

2.6.9 Conclusion....................................................................................................................148

2.7 DISCUSSION.......................................................................................................................150

2.7.1 Assumption Testing......................................................................................................150

2.7.2 Does Job Autonomy Influence the Use of the Control Strategies?..............................152

2.7.3 Does Job Autonomy Influence the Relationship Between the Control Strategies and Job

Satisfaction?...............................................................................................................................154

viii

2.7.4 Do the Control Strategies Mediate the Relationship Between Job Autonomy and Job

Satisfaction?...............................................................................................................................156

2.7.5 Examining Occupational Differences in Job Satisfaction...........................................158

2.7.6 Examining Occupational Differences in Life Satisfaction...........................................163

2.7.7 Predicting Job Satisfaction from Job Autonomy, Control Strategies, Personality, and

Life Satisfaction..........................................................................................................................164

2.7.8 Conclusion....................................................................................................................166

3 CHAPTER 3 - STUDY TWO.........................................................................................................167

3.1 ABSTRACT.........................................................................................................................168

3.2 PROPOSAL FOR STUDY TWO..............................................................................................169

3.2.1 a) The Primary and Secondary Control Scale.............................................................169

3.2.2 b) Job Autonomy Scale.................................................................................................181

3.2.3 c) Occupational Groups...............................................................................................182

3.2.4 d) Need for Job Autonomy............................................................................................184

3.2.5 e) Addition of Social Support.......................................................................................186

3.3 MODEL OF JOB SATISFACTION..........................................................................................189

3.4 AIMS AND HYPOTHESES....................................................................................................191

3.5 METHOD............................................................................................................................195

3.5.1 Participants..................................................................................................................195

3.5.2 Materials......................................................................................................................196

3.5.3 Procedure.....................................................................................................................203

3.6 RESULTS............................................................................................................................204



3.6.3 Preliminary Examination of the Primary Control and Secondary Control Scale.......206

ix


3.7.1 Hypothesis One: Levels of Job Autonomy and Job Satisfaction..................................211

3.7.2 Hypotheses Two and Three: Examining how Job Autonomy Influences the Amount of

Primary and Secondary Control Strategies...............................................................................212

3.7.3 Hypotheses Four and Five: Examining how Job Autonomy Influences the Relationship

Between the Control Strategies and Job Satisfaction.................................................................213

3.7.4 Hypothesis Six: Examining the Proposed Explanation for the Relationship Between Job

Autonomy and Job Satisfaction..................................................................................................216

3.7.5 Hypothesis Seven: Occupational Differences in Job Satisfaction and Life

Satisfaction.................................................................................................................................218

3.7.6 Hypothesis Eight: Examining how Social Support at Work Moderates the Relationship

between Difficulties at Work and Job Satisfaction.....................................................................222

3.7.7 Hypothesis Nine: The Moderating Role of Need for Autonomy on the Relationship

Between Job Autonomy and Job Satisfaction.............................................................................227

3.7.8 Hypothesis Ten: Major Predictors of Job Satisfaction................................................228

3.7.9 Conclusion....................................................................................................................231

3.8 DISCUSSION.......................................................................................................................232

3.8.1 Assumption- The Academics Represent a High Job Autonomy Group and the Teachers

Represent a Low Job Autonomy Group......................................................................................232

3.8.2 Hypothesis Testing.......................................................................................................235

3.8.3 Job Autonomy Influences the Amount of the Control Strategies..................................235

3.8.4 Job Autonomy Influences the Relationship Between the Control Strategies and Job

Satisfaction.................................................................................................................................238

3.8.5 Limitations in the Hypotheses Examining Job Autonomy and Control Strategies......239

3.8.6 Other Predictors of Job Satisfaction............................................................................241

3.8.7 Occupational Differences in Job Satisfaction and Life Satisfaction............................241

x

3.8.8 The Influence of Social Support at Work on the Relationship Between Work Difficulties

and Job Satisfaction...................................................................................................................245

3.8.9 The Influence that Need for Job Autonomy has on the Relationship Between Job


3.8.10 Major predictors of Job Satisfaction.......................................................................247

3.8.11 Conclusion...............................................................................................................248

4 CHAPTER 4 - STUDY THREE....................................................................................................249

4.1 ABSTRACT.........................................................................................................................250

4.2 PROPOSAL FOR STUDY THREE...........................................................................................251

4.2.1 Specificity of Hypotheses Testing the Proposal that Job Autonomy Influences the


4.2.2 Examining how the Controllability of a Difficulty Influences the Use of the Control

Strategies....................................................................................................................................252

4.2.3 Empirical Studies Examining if the Controllability of a Situation Influences the Use of


4.2.4 Examining how Controllability Influences the Adaptiveness of the Control

Strategies....................................................................................................................................257

4.2.5 Developing a Situation Specific Primary and Secondary Control Scale.....................261

4.2.6 Examining the Moderating Role of Primary and Secondary Control Strategies.........265

4.2.7 Examining the Moderating Role of Social Support at Work........................................267

4.3 REVISED MODEL OF JOB SATISFACTION...........................................................................269

4.4 HYPOTHESES.....................................................................................................................272

4.5 METHOD............................................................................................................................274

4.5.1 Participants..................................................................................................................274

4.5.2 Materials......................................................................................................................275

4.5.3 Procedure.....................................................................................................................280

xi

4.6 RESULTS............................................................................................................................281



4.6.3 Factor Analyses............................................................................................................284

4.6.4 Primary and Secondary Control Scale........................................................................284

4.6.5 Social Support at Work................................................................................................289

ITEMS WITH LOADINGS LESS THAN 0.30 ARE NOT SHOWN..............................................................290


4.7.1 Hypothesis One- Use of Control Strategies for Controllable and Uncontrollable

Difficulties..................................................................................................................................291

4.7.2 Hypothesis Two- Adaptiveness of the Control Strategies for Controllable and

Uncontrollable Difficulties.........................................................................................................295

4.7.3 Hypothesis Three- The Moderating Role of Primary and Secondary Control............298

4.7.4 Hypothesis Four - Moderating Role of Instrumental Support.....................................303

4.7.5 Hypothesis Five- Moderating Role of Emotional Support...........................................307

4.7.6 Hypothesis Six- Major Predictors of Job Satisfaction.................................................310

4.7.7 Conclusion....................................................................................................................311

4.8 DISCUSSION.......................................................................................................................313

4.9 HYPOTHESES TESTING.......................................................................................................313

4.9.1 Primary Control, Self-Protective Secondary Control, and Self-Affirmative

SecondaryControl.......................................................................................................................314

4.9.2 Proposal One: The Controllability of the Difficulty Influences the Amount and

Adaptiveness of the Control Strategies Used to Manage that Difficulty....................................317

4.9.3 Proposal Two: Moderators of Controllable and Uncontrollable Difficulties on Job

Satisfaction.................................................................................................................................329

4.9.4 Proposal Three: Predictors of Job Satisfaction...........................................................334

xii

4.9.5 Conclusion....................................................................................................................336

5 CHAPTER 5 - FINAL DISCUSSION...........................................................................................338

5.1 ABSTRACT.........................................................................................................................339

5.2 THE DEVELOPMENT OF A NEW MODEL OF JOB SATISFACTION.......................................340

1) Primary and Secondary Control Strategies Mediate the Relationship Between Job


5.2.2 Conclusion: Do the Control Strategies Mediate the Relationship Between Job

Autonomy and Job Satisfaction?................................................................................................350

5.2.3 2) Social Support at Work and Life Satisfaction Directly Predict Job Satisfaction....351

5.2.4 3) The Control Strategies and Social Support at Work Moderate the Relationship

Between Work Difficulties and Job Satisfaction........................................................................354

5.3 REVISED MODEL OF JOB SATISFACTION...........................................................................357

5.4 CONCLUSION.....................................................................................................................360

5.5 FINAL WORD.....................................................................................................................361

5.6 REFERENCES......................................................................................................................362

5.7 APPENDICES.......................................................................................................................399

List of Tables

TABLE 1- SOURCES OF GOOD/BAD TIMES FOR ACCOUNTANTS AND ENGINEERS (N =200)...................23

TABLE 2- SECONDARY CONTROL STRATEGIES......................................................................................74

TABLE 3- MEANS AND STANDARD DEVIATIONS OF MAJOR VARIABLES FOR ACADEMICS AND

SUPERMARKET WORKERS...........................................................................................................118

TABLE 4- INTER-CORRELATIONS FOR THE ACADEMICS AND THE SUPERMARKET WORKERS.............119

TABLE 5- FACTOR ANALYSIS OF JOB SATISFACTION SCALE...............................................................121

TABLE 6- TOTAL VARIANCE EXPLAINED BY A FIVE-FACTOR SOLUTION............................................123

xiii

TABLE 7- FACTOR ANALYSIS OF PRIMARY AND SECONDARY CONTROL SCALE.................................124

TABLE 8- FACTOR ANALYSIS OF JOB AUTONOMY SCALE...................................................................126

TABLE 9- MEANS AND STANDARD DEVIATIONS OF CONTROL MEASURES FOR ACADEMICS AND


TABLE 10- MULTIPLE REGRESSION OF PRIMARY AND SECONDARY CONTROL ON JOB SATISFACTION

FOR ACADEMICS AND SUPERMARKET WORKERS.......................................................................132

TABLE 11- MODERATING ROLE OF PRIMARY AND SECONDARY CONTROL ON THE RELATIONSHIP

BETWEEN JOB AUTONOMY AND JOB SATISFACTION..................................................................137

TABLE 12 -HIERARCHICAL MULTIPLE REGRESSION TESTING THE MEDIATING ROLE OF THE CONTROL

STRATEGIES................................................................................................................................140

TABLE 13- MEANS AND STANDARD DEVIATIONS OF JOB SATISFACTION SCALE FOR ACADEMICS AND


TABLE 14- MEANS AND STANDARD DEVIATIONS OF LIFE SATISFACTION FOR ACADEMICS AND


TABLE 15- MULTIPLE REGRESSION OF JOB AUTONOMY, CONTROL STRATEGIES, PERSONALITY, AND

LIFE SATISFACTION FOR ACADEMICS AND SUPERMARKET WORKERS.......................................148

TABLE 16- FACTOR ANALYSIS OF PRIMARY AND SECONDARY CONTROL SCALE...............................171

TABLE 17- ORIGINAL AND REVISED PRIMARY CONTROL ITEMS.........................................................175

TABLE 18- ORIGINAL AND REVISED SECONDARY CONTROL ITEMS....................................................179

TABLE 19- FUNCTIONS OF THE SECONDARY CONTROL STRATEGIES...................................................179

TABLE 20- DEMOGRAPHICS OF THE ACADEMICS AND TEACHERS.......................................................196

TABLE 21- FACTOR ANALYSIS OF THE NEED FOR JOB AUTONOMY SCALE.........................................199

TABLE 22- MEANS AND STANDARD DEVIATIONS OF MAJOR VARIABLES FOR ACADEMICS AND

TEACHERS...................................................................................................................................205

TABLE 23- INTER-CORRELATIONS FOR THE ACADEMICS AND TEACHERS...........................................206

TABLE 24- FREQUENCY OF PRIMARY AND SECONDARY CONTROL.....................................................207

TABLE 25- FACTOR ANALYSIS OF THE REVISED PRIMARY AND SECONDARY CONTROL SCALE.........209

xiv

TABLE 26- MULTIPLE REGRESSION OF PRIMARY AND SECONDARY CONTROL ON JOB SATISFACTION

FOR ACADEMICS AND TEACHERS................................................................................................214

TABLE 27- HIERARCHICAL MULTIPLE REGRESSION TESTING THE MODERATING ROLE OF THE CONTROL

STRATEGIES ON THE RELATIONSHIP BETWEEN JOB AUTONOMY AND JOB SATISFACTION........216

TABLE 28- HIERARCHICAL MULTIPLE REGRESSION TESTING THE MEDIATING ROLE OF THE CONTROL

STRATEGIES................................................................................................................................217

TABLE 29- MEANS AND STANDARD DEVIATIONS OF THE INTRINSIC AND EXTRINSIC JOB

SATISFACTION ITEMS FOR ACADEMICS AND TEACHERS............................................................220

TABLE 30- MEANS AND STANDARD DEVIATIONS OF THE DOMAINS OF LIFE SATISFACTION FOR

ACADEMICS AND TEACHERS.......................................................................................................221

TABLE 31- HIERARCHICAL MULTIPLE REGRESSION ANALYSIS EXAMINING IF SUPERVISOR SUPPORT

MODERATES THE RELATIONSHIP BETWEEN WORK DIFFICULTIES AND JOB SATISFACTION......223

TABLE 32- HIERARCHICAL REGRESSION ANALYSES EXAMINING WHETHER CO-WORKER SUPPORT

MODERATES THE RELATIONSHIP BETWEEN WORK DIFFICULTIES AND JOB SATISFACTION......226

TABLE 33- HIERARCHICAL REGRESSION ANALYSES EXAMINING WHETHER NEED FOR JOB AUTONOMY

MODERATES THE RELATIONSHIP BETWEEN JOB AUTONOMY AND JOB SATISFACTION..............228

TABLE 34- STANDARD MULTIPLE REGRESSION PREDICTING JOB SATISFACTION FOR EMPLOYEES WITH

LOW AUTONOMY AND EMPLOYEES WITH HIGH AUTONOMY.....................................................230

TABLE 35- NORMATIVE DATA FOR HACKMAN AND OLDHAM’S (1980) AUTONOMY SCALE..............234

TABLE 36- DEMOGRAPHIC CHARACTERISTICS OF THE SAMPLE..........................................................274

TABLE 37- MEANS AND STANDARD DEVIATIONS OF THE MAJOR VARIABLES...................................283

TABLE 38- INTER-CORRELATIONS AMONG MAJOR VARIABLES...........................................................283

TABLE 39- FACTOR ANALYSIS OF PRIMARY AND SECONDARY CONTROL ITEM IN CONTROLLABLE

SITUATIONS.................................................................................................................................286

TABLE 40- FACTOR ANALYSIS OF PRIMARY AND SECONDARY CONTROL ITEMS IN UNCONTROLLABLE

SITUATION...................................................................................................................................288

TABLE 41-SECONDARY CONTROL ITEMS INCLUDED IN ANALYSES.....................................................289

TABLE 42- FACTOR ANALYSIS OF THE SOCIAL SUPPORT AT WORK SCALE........................................290

xv

TABLE 43 - CONTROLLABLE AND UNCONTROLLABLE DIFFICULTIES REPORTED BY EMPLOYEES.......292

TABLE 44- EMPLOYEES USE OF PRIMARY AND SECONDARY CONTROL IN CONTROLLABLE AND

UNCONTROLLABLE SITUATIONS.................................................................................................293

TABLE 45- MEANS AND STANDARD DEVIATIONS OF INDIVIDUAL CONTROL STRATEGIES.................295

TABLE 46- STANDARD MULTIPLE REGRESSION ANALYSIS PREDICTING JOB SATISFACTION FROM

PRIMARY AND SECONDARY CONTROL........................................................................................297

TABLE 47- CORRELATIONS BETWEEN INDIVIDUAL CONTROL STRATEGIES AND JOB SATISFACTION FOR

CONTROLLABLE AND UNCONTROLLABLE DIFFICULTIES............................................................298

TABLE 48- HIERARCHICAL MULTIPLE REGRESSION TESTING THE MODERATING ROLE OF CONTROL

STRATEGIES ON THE RELATIONSHIP BETWEEN WORK DIFFICULTIES AND JOB SATISFACTION. 302

TABLE 49- HIERARCHICAL REGRESSION ANALYSES TESTING THE MODERATING ROLE OF

INSTRUMENTAL SUPPORT...........................................................................................................307

TABLE 50- HIERARCHICAL REGRESSION ANALYSES TESTING THE MODERATING ROLE OF EMOTIONAL

SUPPORT......................................................................................................................................309

Table 51- Standard Multiple Regression Predicting Job Satisfaction..................................................311

List of Figures

FIGURE 1-JOB CHARACTERISTICS MODEL..............................................................................................49

FIGURE 2- MODEL OF JOB SATISFACTION..............................................................................................99

FIGURE 3- EXPECTED MODERATED EFFECT OF JOB AUTONOMY ON A) PRIMARY CONTROL AND JOB

SATISFACTION AND B) SECONDARY CONTROL AND JOB SATISFACTION....................................133

FIGURE 4- JOB AUTONOMY MODERATES THE RELATIONSHIP BETWEEN A) PRIMARY CONTROL AND B)

SECONDARY CONTROL, AND JOB SATISFACTION.......................................................................135

FIGURE 5- MEDIATING ROLE OF CONTROL STRATEGIES ON THE RELATIONSHIP BETWEEN JOB

AUTONOMY AND JOB SATISFACTION..........................................................................................138

FIGURE 6 -REVISED MODEL OF JOB SATISFACTION FOR STUDY 2.......................................................190

xvi

FIGURE 7 - RELATIONSHIP BETWEEN WORK DIFFICULTIES AND JOB SATISFACTION FOR EMPLOYEES

WITH LOW/HIGH SUPERVISOR SUPPORT.....................................................................................225

FIGURE 8- REVISED MODEL OF JOB SATISFACTION.............................................................................271

FIGURE 9 – PRIMARY AND SECONDARY CONTROL MODERATE THE RELATIONSHIP BETWEEN WORK

DIFFICULTIES AND JOB SATISFACTION.......................................................................................300

FIGURE 10 - REGRESSION OF CONTROLLABLE WORK DIFFICULTIES ON JOB SATISFACTION FOR

EMPLOYEES WITH LOW INSTRUMENTAL CO-WORKER SUPPORT AND EMPLOYEES WITH HIGH

INSTRUMENTAL CO-WORKER SUPPORT......................................................................................305

Figure 11- Revised Model of Job Satisfaction.....................................................................................359

xvii

List of Appendices

APPENDIX A- PLAIN LANGUAGE STATEMENT FOR STUDY ONE..........................................................400

APPENDIX B- JOB AUTONOMY SCALE USED IN STUDY ONE (REVISION OF GANSTER, 1989, CITED IN

DWYER & GANSTER, 1991)........................................................................................................401

APPENDIX C- PRIMARY AND SECONDARY CONTROL SCALE USED IN STUDY ONE (REVISION OF HEEPS

ET AL., 2000)...............................................................................................................................403

APPENDIX D- JOB SATISFACTION SCALE USED IN STUDY ONE (REVISION OF ROZNOWSKI, 1989)....406

APPENDIX E- LIFE SATISFACTION SCALE USED IN STUDY ONE (CUMMINS, 1997).............................408

APPENDIX F- PERSONALITY SCALE USED IN STUDY ONE (COSTA & MCCRAE, 1992).......................409

APPENDIX G-LEVELS OF JOB SATISFACTION REPORTED BY VARIOUS OCCUPATIONAL GROUPS..........412

APPENDIX H- PRIMARY AND SECONDARY CONTROL SCALE FOR STUDY TWO (MAHER ET AL., 2001)

....................................................................................................................................................415

APPENDIX I- PLAIN LANGUAGE STATEMENT FOR STUDY TWO...........................................................419

APPENDIX J- JOB AUTONOMY SCALE FOR STUDY TWO (HACKMAN & OLDHAM, 1975)....................420

APPENDIX K- NEED FOR AUTONOMY SCALE FOR STUDY TWO (DE RIJK ET AL., 1998).....................421

APPENDIX L- JOB SATISFACTION SCALE FOR STUDY TWO (WEISS ET AL., 1967)..............................422

APPENDIX M- SOCIAL SUPPORT SCALE FOR STUDY TWO (REVISION OF KARASEK & THEORELL, 1990)

....................................................................................................................................................425

APPENDIX N-PLAIN LANGUAGE STATEMENT USED IN STUDY THREE.................................................426

APPENDIX O- PRIMARY AND SECONDARY CONTROL SCALE FOR STUDY THREE (MAHER & CUMMINS,

2002)...........................................................................................................................................427

APPENDIX P- LIFE SATISFACTION SCALE FOR STUDY 3 (CUMMINS ET AL., 2001)..............................433

APPENDIX Q- SOCIAL SUPPORT SCALE FOR STUDY 3 (DUCHARME & MARTIN, 2000).......................434

xviii

Abstract

Although theories of job satisfaction have been extensively studied, researchers are

yet to agree on the major predictors of job satisfaction. One theory, which is

particularly appealing to the workplace, is Karasek and Theorell’s (1990) job

demand-control model. Essentially, this model proposes that job autonomy can

reduce the effects of job demands on job satisfaction by allowing workers to redirect

the physiological arousal produced from job demands into an appropriate response.

This explanation is criticised however for being tautological, and a new explanation

is developed which incorporates the life span theory of control (Heckhausen &

Schulz, 1995) and the discrimination model (Thompson et al., 1998). Specifically it

is proposed that job autonomy influences the use, and the adaptiveness of primary

and secondary control strategies. This proposal is developed into a model of job

satisfaction that includes job autonomy, primary and secondary control, life

satisfaction, personality, and social support at work. This model of job satisfaction is

tested over three studies using university academic staff, secondary school teachers,

supermarket workers and general employees. Overall, the results demonstrated that

job autonomy did not influence the use or adaptiveness of the control strategies.

These results suggest that employees have trait control strategies, and they also

challenge the assumptions about primary control failure. The proposed model of job

satisfaction was revised to include job autonomy, primary and secondary control

strategies and their successfulness, life satisfaction, work difficulties, and

personality.

xix

Chapter 1 - Literature Review

1

1.1 Abstract

Although theories of job satisfaction have been extensively researched in the

organisational psychology literature, researchers are yet to agree on the major

predictors of job satisfaction. Several predictors have been investigated such as

needs, values, expectations and specific job characteristics such as job autonomy and

job demands. This chapter reviews such theories, focussing on the ones that have

made the greatest contribution to the understanding of job satisfaction. Although

these theories are well cited, many of them have theoretical and empirical problems

as well as having limited applicability to the workplace. One theory, which is less

problematic, and particularly appealing to the workplace, is the job demand-control

model. This model proposes that job autonomy can reduce the effect of job demands

on job satisfaction, and that the most satisfied workers are those with high job

demands and high job autonomy. According to the model, job autonomy influences

job satisfaction because it allows workers to redirect the physiological arousal

produced from job demands into an appropriate response. This explanation is

criticised however for being non-specific and tautological. A new explanation is

developed, where it is proposed that job autonomy influences how employees

respond to work difficulties. This explanation forms the basis of a model of job

satisfaction, which includes the following predictors: job autonomy; primary control

and secondary control; personality; and life satisfaction.

2

1.2 Job Satisfaction

Job satisfaction, the extent to which employees like their job and its

components (Spector, 1997), is one of the most extensively researched topic in the

industrial and organisational psychology literature (Highhouse & Becker, 1993).

The number of articles and books investigating this construct has increased from

over 3000 in 1976 (Locke, 1976), to over 5000 in 1992 (Harwood & Rice, 1992).

Today, a review of psychology and business databases demonstrates that over 10,000

publications on job satisfaction are available. Although this increasing interest in job

satisfaction is no doubt beneficial to the field of industrial and organisational

psychology, the amount of research has become overwhelming to both researchers

and practitioners. Nowhere is this more clearly evident than in the theories of job

satisfaction.

1.2.1 Theories of Job Satisfaction: Environmental and Dispositional

Predictors

Theories of job satisfaction include dispositional and environmental

predictors. The dispositional predictors of job satisfaction refer to characteristics of

the employee, such as needs, values, and expectations. The environmental predictors

refer to job characteristics, such as job control, workload, feedback, role ambiguity,

and role conflict. Some theorists focus on the dispositional predictors, whilst others

focus on the environmental predictors. More recent theorists recognise the

importance of both types of predictors.

3

Dispositional and environmental theories of job satisfaction have been

extensively researched, however researchers have still not reached consensus as to

the major predictors of job satisfaction. As a result, researchers continue to rely on

theories that have theoretical and empirical problems, or have limited applicability to

the workplace. In order to determine which theories are valid and useful, this review

will examine the theories that have made the greatest contribution to a shift in focus

of the determinants of job satisfaction. These include Maslow’s (1970) need

hierarchy theory, Herzberg, Mausner and Snyderman’s (1959) two-factor theory of

job satisfaction, Vroom’s (1964) expectancy theory, discrepancy theories, Hackman

and Oldham’s (1976) job characteristics model, and Karasek’s (1979) job demand-

control model.

4

1.3 Maslow’s (1954, 1970) Hierarchy of Needs

The need hierarchy theory was one of the first theories to focus on the

dispositional predictors of job satisfaction. It proposed that employees’ needs

determine their level of job satisfaction.

1.3.1 Need Hierarchy Theory

The need hierarchy theory (Maslow, 1954, 1970) posits that individuals are

born with a set of needs. There are five needs: physiological, safety, belongingness,

esteem, and self-actualisation. These are arranged in a hierarchy of relative

prepotency, meaning that lower-order needs are satisfied before higher-order needs

are activated.

The lowest need, physiological, refers to basic biological drives, such as

hunger, thirst and sex. These physiological needs are the most prepotent of all, as an

individual deprived of all needs would seek to gratify these needs first. They would

not be concerned with safety, belongingness, esteem, or self-actualisation. Once they

have gratified the physiological needs however, the strength of that need decreases,

and the next highest need, safety, becomes important.

The safety need refers to security, stability, dependency, protection, and need

for structure, order, law and limits. To gratify the safety need, an individual requires

a safe, orderly, predictable, lawful world. Once the safety need is gratified, its need

strength is reduced, and the strength of the belongingness need increases. The

individual will begin to hunger for affectionate relationships with people, and for a

place in their group or family. Once these belongingness needs are gratified, the

5

strength of the esteem need increases, and the individual will desire a high evaluation

of themselves, and others. Once an individual has gratified these four needs,

collectively known as deficiency needs (D-needs), they may begin to feel restless.

This restlessness is indicative of the need for self-actualisation.

The need for self-actualisation refers to the need for the individual to become

everything they are capable of becoming. When the strength of this need increases,

the individual strives for self-fulfilment. This fifth need is referred to as a being need

(B-need) because it sustains an individual’s interest without being driven by feelings

of deprivation. Unlike the previous four needs, when the need for self-actualisation is

gratified, it increases in need strength (Maslow, 1962). Growth is a continued

upward development, where the more that one gets, the more that one wants. This

growth is “endless, and can never be attained or satisfied” (Maslow, 1962, p. 31).

1.3.2 Applying Maslow’s (1954, 1970) Theory to Organisations

In terms of applying this theory to organisations, the theory proposes that the

lower-order needs must be gratified before the higher-order needs are activated. As

such, employers must ensure that their employees’ physiological, safety,

belongingness and esteem needs are satisfied. The employer can help the employee

to gratify each need. For example, to help them gratify their physiological and safety

needs, employers can increase their employees’ pay. Once these needs are satisfied,

the relationship between the employee and their supervisors and co-workers takes on

increased strength. The employer can help the employee to gratify this need through

increasing the amount of social interaction among employees. This process needs to

6

be continued until the employees have gratified all of the lower-order needs, and are

reaching for self-actualisation, should the nature of the job permit this level to be

attained.

1.3.3 Criticisms of Maslow’s (1954, 1970) Need Hierarchy Theory

Almost every aspect of Maslow’s (1954, 1970) work has been disputed on

both theoretical and empirical grounds (Neher, 1991; Wahba & Bridwell, 1976).

Five fundamental propositions of Maslow’s (1954, 1970) theory have been

questioned, including: 1) the higher the deprivation of a need, the higher its need

strength (i.e., deprivation/domination paradigm); 2) the higher the satisfaction with a

need, the higher the need strength of the need at the next level (i.e.,

gratification/activation paradigm); 3) the measurement of self-actualisation; 4) the

ability to achieve self-actualisation; and 5) the applicability of the theory to

organisations. Each of these will now be considered.

1.3.3.1 Criticism One: Deprivation/Domination Paradigm

The deprivation/domination paradigm postulates that the higher the

deprivation of a need, the higher its need strength. An early review concluded that

the deprivation/domination paradigm was only partially supported for self-

actualisation, and not supported for safety, belongingness and esteem needs (Wahba

& Bridwell, 1976). On the basis of this review, many researchers have assumed that

the proposition is not supported (Wicker, Brown, Wiehe, Hagen & Reed, 1993).

This assumption may be inaccurate however, as many of the studies included in the

7

review have methodological limitations. These limitations concern: a) the

operationalisation of need strength; and b) establishing causality.

1.3.3.2 a) Operationalising Need Strength

One of the main limitations in studies examining the deprivation/domination

paradigm concerns the operationalisation of need strength. Some researchers

measure need strength through desire, others through important or intention. Two

studies have measured need strength through desire. In Alderfer’s (1969) study,

subjects were asked to rate how much more of the following factors they would like

to have in their jobs; pay, fringe benefits, love, status, and growth. Similarly, in

Graham and Balloun’s (1973) study, subjects were asked how much improvement

they wanted in their physiological, security, social and self-actualisation needs.

These measures of need strength were then correlated with corresponding measures

of satisfaction.

Both studies provided some support for Maslow’s (1954) theory suggesting

that as satisfaction with a need increases, the strength of that need decreases. For

example, in Graham and Balloun’s (1973) study, the correlations between need

strength and satisfaction ranged from r = –0.42 to r = –0.72. Furthermore, in

Alderfer’s (1969) study, satisfaction and need strength were negatively correlated for

the relatedness need, which was composed of a respect from co-workers’ need, and a

respect from supervisors’ need. For the respect from co-workers’ need, the

correlations were all significant, ranging from r = -0.21 to r = -0.38. For the respect

from supervisors’ need, the correlations ranged from r = -0.06 to r = -0.49. Although

8

the correlations in Alderfer’s (1969) study were in the expected direction, they were

often small, and the correlations between satisfaction and need strength for the

belongingness need were insignificant (r = 0.02 to r = 0.07).

These two studies appear to provide some support for Maslow's (1954)

theory. Both of these studies assessed need strength ratings by desire, where

participants were asked how much more they wanted of a need. It must be

questioned however, if wanting or desiring more of a need is a measure of the

strength of the need. Wanting more of a need may actually be another way of

demonstrating dissatisfaction with the area covered by that need.

Other researchers have overcome this limitation by assessing need strength

using importance ratings, which may be less likely to measure satisfaction. For

example, Hall and Nougaim (1968) conducted a longitudinal study on managers,

interviewing them annually for five years. The participants rated the importance of,

and satisfaction with a number of needs including safety, affiliation, achievement and

esteem, and self-actualisation. Inconsistent with Maslow’s (1954) theory, the

correlations between the satisfaction of needs and the importance of needs were

positive. For safety, importance and satisfaction correlated r = 0.26, for affiliation

r = 0.16, for achievement and esteem r = 0.54, and for self-actualisation r = 0.29.

In addition, Hall and Nougaim (1968) also examined the longitudinal changes

in satisfaction and importance for each need. According to Maslow’s (1954) theory,

it would be expected that if satisfaction of a need increased from one year to the next,

importance of that need would decrease. However, they found that the importance of

9

a need in a given year was positively correlated with its own satisfaction in the

previous year. These correlations were moderate for safety (r = 0.25), affiliation

(r = 0.21), achievement and esteem (r = 0.53) and self-actualisation (r = 0.28).

Although Hall and Nougaim (1968) failed to discuss these correlations in detail, they

clearly contradict Maslow’s (1954) theory. Importance was positively related to

need satisfaction, suggesting that a satisfied need is an important need. This finding

does not support Maslow’s (1970, p. 393) proposal that “a satisfied need is not a

motivator.”

Although Hall and Nougaim’s (1968) findings are inconsistent with

Maslow’s (1954) theory, their validity has been questioned. Specifically, the study

relied on a small sample, and the interview was not designed to produce data relevant

to Maslow’s (1954) theory (Lawler & Suttle, 1972). Furthermore, the inter-rater

reliability of the coding of interviews was low (0.55 to 0.59).

A study designed to overcome the limitations identified in Hall and

Nougaim’s (1968) study was conducted by Lawler and Suttle (1972). They

employed a reasonably large sample of employees from government agencies and

retail stores. Their questionnaire, developed by Porter (1963), was designed to

measure Maslow’s (1954) needs. According to Maslow’s (1954) theory, the

importance of a need should be negatively correlated with satisfaction of that need.

Hence, as satisfaction with a need increases, the importance of that need decreases.

Lawler and Suttle’s (1972) results did not support this proposal for either the

government or retail organisations respectively, for social (r = -0.09, r = 0.07),

esteem (r = 0.06, r = -0.04), autonomy (r = 0.07, r =0.01), and self-actualisation

10

needs (r = 0.01, r = -0.10). There was however, some support for the security needs

(r = -0.34, r = -0.12).

As their study was longitudinal they also conducted change analyses. They

correlated the change in need importance with the change in need satisfaction. It was

expected that these correlations would be negative, indicating that increases in the

satisfaction of a need were associated with decreases in its importance. However,

these correlations were also positive ranging from r = 0.07 to r = 0.24. Hence, the

direction of the correlations were inconsistent with Maslow’s (1954) theory.

In summary, Hall and Nougaim’s (1968) and Lawler and Suttle’s (1972)

findings are inconsistent with those of Alderfer (1969) and Graham and Balloun

(1972). The major difference between these studies is that the latter two measured

need strength with desire or improvement, while the former two relied on measures

of importance. Although the desire and improvement measures were criticised

earlier for being too similar to measures of satisfaction, the use of importance as an

indicator of need strength has also been criticised (Wicker et al., 1993).

Although Maslow (1970) postulates that a need is important because of

deprivation, it has been suggested that a person may report that a need is important

because they have attained it and value it (Wicker et al., 1993). Indeed, Maslow

(1954, p. 148) proposed that “greater value is usually placed on higher-order needs

by persons who have gratified both kinds (i.e., lower and higher-order needs).”

Hence, people who are self-actualising may report that all the higher needs are

important because they value them. A person may thus report that a higher-order

need is important because they are deprived of it, or because they have attained it and

11

value it. If individuals report that a higher-order need is important because they have

attained it, it would be positively related to satisfaction (Wicker et al., 1993).

Although importance may be an ambiguous construct, the early studies

conducted by Hall and Nougaim (1968) and Lawler and Suttle (1972) should still be

valid. The majority of participants in these studies would not have gratified both

lower-order and higher-order needs. As such, they would only be expected to report

that a need was important if they were deprived of the need. Hence, although the

early studies tested the deprivation/domination paradigm using importance ratings,

this is not expected to reduce the validity of the findings, which are inconsistent with

Maslow’s (1954) theory.

More recent researchers have found some support for Maslow’s (1954)

theory using a different measure of need strength, namely intention. Wicker et al.,

(1993) examined how need strength relates to satisfaction when need strength is

operationalised in a number of different ways. They used, among others, ratings of

importance (i.e., “To what extent is it an important goal”) and ratings of intention

(i.e., “How much do you want to pursue it”). They correlated these variables with

attainment as a measure of deprivation (i.e., “To what extent do you already have

it”). According to Maslow’s (1970) theory, it would be expected that as attainment

of a need decreased, the intention of that need would increase. However, they found

the correlations of past attainment (deprivation) and intention were positive, ranging

from r = 0.39 to r = 0.96. This suggests that as attainment of a need increases, the

intention to pursue the need also increases.

12

Although Wicker et al’s., (1993) findings are inconsistent with Maslow’s

(1970) theory, they suggest that the correlations may have been inflated by

halo-effects or carryover rating bias. They postulate that the ratings may be affected

by a general motivation factor, and by earlier ratings. To control for such effects,

deviation scores were computed and correlated. Deviation scores are calculated by

subtracting the grand mean over all scales for a need from the mean of that need on

each particular scale. This removed a need-means factor from the data, “reducing

any biasing effect on correlations resulting from mean differences among needs”

(Wicker et al., 1993, p. 126). Using these deviation scores, the direction of the

correlations were reversed. For importance, two of the four correlations were in the

expected negative direction, however they were very small (r = -0.13 and r = –0.07).

For intention however, all four of the correlations were strong and negative, ranging

from r = –0.62 to r = –0.74. This suggests that if need strength is measured through

intention, and deviation scores are used, then it is negatively related to attainment.

On this basis, Wicker et al., (1993) postulate that it is too early to discard the

deprivation/domination paradigm. They propose that participants in earlier studies

(e.g., Hall & Nougaim, 1968; Lawler & Suttle, 1972) may have reported that a lower

order need was important because they had attained it and they valued it (high

satisfaction), or because they were deprived of it (low satisfaction). As a result, the

correlations between importance and satisfaction could be positive or negative,

depending on how need strength was operationalised. Despite this, it remains

concerning that the deprivation/domination paradigm is only supported when need

strength is operationalised as intention.

13

1.3.3.3 b) Establishing Causality

A second methodological problem, which may reduce the validity of the

studies examining the deprivation/domination paradigm is that although the

deprivation/domination paradigm is causal, the relationship is assessed through

correlational analyses (e.g., Alderfer, 1969; Graham & Balloun, 1972; Hall &

Nougaim, 1968; Lawler & Suttle, 1972). Only one study has attempted to establish

causality through experimentally manipulating deprivation and measuring

subsequent need strength. Wicker and Wiehe (1999) divided forty students into two

groups, where one group wrote about a past event where they felt especially close to

another person and the other group wrote about a time when they tried to get close to

someone, but felt unsuccessful. Both groups then rated their needs on each level of

the hierarchy on prior attainment (i.e., “To what extent to do you already have it”),

intention (i.e., “How much do you intend to pursue it”), and importance (i.e., “To

what extent is it an important goal”).

The interpersonal scenario was expected to affect their belongingness

responses, where the unsuccessful group would report lower attainment, and higher

need strength for the belongingness need. Inconsistently however, the two groups

did not report different levels of attainment on the belongingness need. The two

groups did report different levels of esteem attainment where the unsuccessful group

reported less past attainment of esteem needs than the successful group. The

unsuccessful group also reported higher intention on all levels of the hierarchy than

the successful group. The two groups did not however differ on importance ratings.

14

These data were interpreted as supporting Maslow’s (1970) theory, as when

the past attainment of esteem needs were low, intentions were higher. The results

must be interpreted with caution however as there were methodological limitations in

the study. Aside from each group having a small sample size (N = 20), need strength

was not assessed prior to the intervention. Hence, the differences in their intentions

may have been a pre-existing difference. Furthermore, although the groups were

asked to report a story relating to belongingness needs, the two groups did not report

different level of past attainment on belongingness needs. Hence, the belongingness

manipulation was not successful. In summary, although Wicker and Wiehe (1999)

present their study as supporting Maslow’s (1970) theory, the findings should be

viewed with caution.

1.3.3.4 Summary: Deprivation/Domination Paradigm

The deprivation/domination paradigm was rejected after several early studies

failed to find supportive correlations. Wicker et al., (1993) re-introduced the

proposition into the literature, attributing the inconsistent findings to the

operationalisation of need strength. They demonstrated that positive correlations

between attainment and need strength could be reversed if deviation scores were

used, and need strength was measured by intentions rather than importance. The

validity of these findings continues to be questioned however, as the relationship

between need strength and satisfaction, although causal has been assessed through

correlational analyses. In summary, the majority of research demonstrates that as

deprivation increases, need strength does not necessarily increase.

15

1.3.3.5 Criticism Two: Gratification/Activation Paradigm

The gratification/activation paradigm postulates that the higher the

satisfaction with a need, the higher the need strength of the need at the next level of

the hierarchy. The gratification/activation paradigm is different from the

deprivation/domination paradigm as the former examines the correlation between the

satisfaction of a need at one level with the importance of the need at the next level,

whereas the latter examines the correlation between satisfaction and need strength of

a need on the same level.

Two longitudinal studies have been conducted to evaluate the

gratification/activation paradigm. As previously mentioned, Hall and Nougaim

(1968) interviewed managers annually throughout a five-year period, coding their

responses on need strength and satisfaction. For each year, they correlated the

changes in need satisfaction from one year to the next with changes in need strength

at the next highest level during the same period of time. According to Maslow’s

(1954) theory, it was expected that high correlations would exist between the change

in satisfaction of a given need level and the change in strength of the next highest

level. The pooled correlations were low however, ranging from r = 0.05 to r = 0.22.

Hence, there was little evidence to suggest that the increasing satisfaction of a need

results in the increasing need strength of the next highest need. It must be noted

however that this study relied on a small sample size, and the interview used in the

study was not designed to produce data relevant to Maslow’s (1954) theory. These

limitations were addressed in Lawler and Suttle’s (1972) study.

16

As previously mentioned, Lawler and Suttle (1972) relied on Porter’s (1963)

questionnaire, which was specifically designed to measure Maslow’s (1954) needs.

According to Maslow’s (1954) theory, it was expected that the satisfaction of a need

would be positively correlated with the need strength of the need in the next highest

level. Lawler and Suttle’s (1972) results demonstrated that one correlation between

security satisfaction, and social importance was significant for the retail group

(r = 0.21), however the rest were all low ranging from r = -0.01 to r = 0.10. These

findings, as with Hall and Nougaim’s (1968) findings clearly raise questions

concerning the validity of the gratification/activation paradigm.

In summary, the gratification/activation paradigm proposes that as

satisfaction with a need increases, the need strength of the next highest need

increases. Studies investigating this paradigm generally demonstrate that the

correlations between need satisfaction and need strength of the next highest need are

low.

1.3.3.6 Criticism Three: Measurement of Self-Actualisation

There is a poor level of concordance between the definition of the need for

self-actualisation, and the measurement of the need for self-actualisation. Self-

actualisation is defined as “the full use of one’s talents, capacities, potentialities”

(Maslow, 1970, p. 150). It is the need for the individual to become everything they

are capable of becoming. Self-actualisers have a more efficient perception of reality,

accept others, are autonomous, do not need others, are less concerned with

themselves, and have deeper interpersonal relationships (Maslow, 1970). These

17

characteristics must be regarded with caution however as they were based on a social

discussion with a sample of 22 people whom Maslow (1954) believed to be self-

actualisers. These people were selected as they seemed to be fulfilling themselves,

and doing the best they were capable of. Perhaps as a consequence of this vague

definition, operational definitions of the need for self-actualisation vary extensively.

Several early studies measured self-actualisation using Porter’s (1963) need

scale (i.e., Lawler & Suttle, 1972; Roberts, Walter & Miles, 1971). This scale

includes three items which assess the opportunity for personal growth and

development in the job, the feelings of self-fulfilment a person gets from being in the

job, and the feelings of worthwhile accomplishment in the job. One problem with

these items however, is that they appear to assess how the person feels about their

work rather than whether they feel they are have reached their potential.

Although more recent scales tend to be more comprehensive, their validity is

still questioned. For example, Shoura and Singh (1999) assessed self-actualisation

through items measuring meaningfulness, self-sufficiency, effortlessness, creativity,

professional creativeness, self-understanding, independence, and harmony with the

universe. Examples of these items are “do you think you have enough talents and

capabilities to perform the job”, “does your work come as second nature to you” and

“do you feel your job is in harmony with the universe.” These items are criticised for

being vague, and it is questioned whether they measure if a person has become all

that they are capable of. Furthermore, these items only refer to self-actualisation on

the job, and in some cases, self-actualisation may occur off the job. In summary,

18

there seems to be a great deal of discrepancy between the definition and

measurement of self-actualisation.

1.3.3.7 Criticism Four: Ability to Achieve Self-Actualisation

The need for self-actualisation is the need for the individual to become

everything that they are capable of becoming. This suggests that anyone performing

their job to the best of their abilities is self-actualising. However, Maslow (1970)

screened 3000 college students and concluded that only one student was

self-actualising. Following this study, Maslow (1970) proposed that

self-actualisation of the sort he had found in older adults was not possible for

younger developing people. He proposed that young people lack many of the

experiences needed for self-actualisation such as identity, autonomy, and romantic

relationships. The proposal that younger people do not self-actualise has not

received empirical support. A study conducted on engineers demonstrated that the

junior engineers reported higher scores on self-actualisation than the senior engineers

(Shoura & Singh, 1999). Furthermore, in a study of academics, ranging in age from

30 to 68 years, age and self-actualisation were not related (Hawkins, Hawkins &

Ryan, 1989). It must be noted however that, as previously mentioned, these studies

relied on questionable measures of self-actualisation.

19

1.3.3.8 Criticism Five: Applicability of Maslow’s Hierarchy of Needs to

Organisations

Although some of the propositions in the need hierarchy theory have not

received empirical support, the theory has been extensively accepted in the

management literature (Roberts, 1982). Moreover, the general idea that the concepts

of love, safety, self-esteem, and growth contribute to motivation and satisfaction are

acceptable to both psychologists and management scientists (Shoura & Singh, 1999).

The fundamental problem in applying Maslow’s (1970) theory to work

organisations is that little is known about how to reach the ultimate goal of self-

actualisation. Maslow’s (1970, p.46) definition of self-actualisation as “what a man

can be, he must be” is extremely vague, and there is no agreed upon way of

operationalising the construct, or facilitating it in employees.

1.3.4 Conclusion

The need hierarchy theory proposes that individuals strive to gratify five needs,

namely physiological, safety, belongingness, esteem and self-actualisation needs.

The theory proposes that the higher the deprivation of a need, the higher its need

strength, and the higher the satisfaction with a need, the higher the need strength of

the next highest need. Although early studies tended to reject these propositions,

more supportive results were found when need strength was operationalised as

intentions rather than importance or desire. Even with some supportive findings, the

validity of the theory is still questioned as very little is known about the ultimate goal

for humans, the need for self-actualisation.

20

1.4 Herzberg, Mausner and Snyderman’s (1959, 1993) Two-Factor Theory of

Job Satisfaction

1.4.1 How the Two-Factor Theory has Contributed to our Understanding of

Job Satisfaction

The two-factor theory (Herzberg et al., 1959) questioned the assumption that

job satisfaction and job dissatisfaction lie on a single continuum. Rather, the theory

proposed that job satisfaction and job dissatisfaction are separate continua, and that

the factors which affect job satisfaction are different from the factors which affect

job dissatisfaction.

1.4.2 Development of the Two-Factor Theory

The two-factor theory is based on a study of accountants and engineers.

Through an interview, employees recalled experiences about times when they felt

especially good or bad about their jobs, and then rated how seriously their feelings

(good/bad) about their jobs had been affected by what happened. Using content

analysis, their responses were coded into 14 categories.

As demonstrated in Table 1, employees reporting the sources of good times

tended to recall events related to achievement, recognition, work itself,

responsibility, and advancement. These sources of satisfaction were termed

motivator factors. Employees reporting the sources of bad times tended to recall

events related to company policy and administration, supervision-technical, salary,

recognition, and interpersonal relations with supervisor. These sources of

21

dissatisfaction were termed hygiene factors. An obvious exception to this

classification is for the factor salary. Salary was reported a similar number of times

for employees reporting the source of good events and for those reporting the source

of bad events.

On the basis of these findings, Herzberg et al., (1959) proposed that paying

attention to motivator factors will increase job satisfaction, but will not affect job

dissatisfaction. Alternatively, paying attention to hygiene factors will decrease job

dissatisfaction but will not increase job satisfaction. For example, increasing status

is expected to reduce job dissatisfaction, but not increase job satisfaction.

Table 1- Sources of Good/Bad Times for Accountants and Engineers (N=200)

Factor Time felt especially good Time felt especially badAchievement 41** 7Recognition 33** 18Work Itself 26** 14Responsibility 23** 6Advancement 20** 11Salary 15 17Possibility of Growth 6 8IR-subordinate 6 3Status 4 4IR-Supervisor 4 15**IR-Peer 3 8**Supervision-technical 3 20**Company policy and administration

3 31**

Working conditions 1 11**Personal life 1 6**Job Security 1 1

**p<0.01; Motivator factors are boldedFrom Herzberg, F., Mausner, B., & Snyderman, B. (1959). The Motivation to Work. (p.60, 72). New York: Wiley.

22

1.4.3 Criticisms of Herzberg et al’s., (1959) Theory

The two-factor theory is criticised for deducing conclusions from a study that:

a) failed to test the main propositions; and b) was methodologically flawed. In

regards to the first criticism, there is insufficient evidence to demonstrate how

motivator and hygiene factors relate to job satisfaction. Although the study

demonstrated that employees recalling good times tended to recall motivator factors,

and employees recalling bad times tended to recall hygiene factors, there is no

empirical evidence for the proposal that motivator factors can only contribute to job

satisfaction and that hygiene factors can only contribute to job dissatisfaction. The

study did not measure job satisfaction, and as such, there is no basis for assuming

that the factors described in the incidents caused, or were even related to job

satisfaction (Ewen, 1964).

In regards to the second criticism of Herzberg et al’s., (1959) theory,

concerning the methodology of the study, several problems have been identified.

These include: 1) some of the findings contradict the theory; 2) the findings differ

depending on the method of data collection; and 3) the hypotheses and criterion

measures are ambiguous. These limitations will now be discussed more extensively.

1.4.3.1 Criticism One: Evaluation of Results

The results from Herzberg et al’s., (1959) study did not completely support

the theory. As can be seen in Table 1, employees often report motivator factors, such

as recognition when they are recalling a time when they felt bad. Although they

reported recognition significantly less for bad times than good times, recognition was

23

still the third highest source of a bad time. Furthermore, some of the hygiene factors

were reported only slightly more for bad events than good events (i.e., salary, status

and job security). Hence, some of the findings are not supportive of the two-factor

theory.

1.4.3.2 Criticism Two: The Interview Method

Replications of Herzberg et al’s., (1959) study have produced mixed results.

Some researchers have found support for the theory (i.e., Schmidt, 1976), whilst

others have contradicted the theory (e.g., Armstrong, 1971; Brenner, Carmack &

Weinstein, 1971; Hill, 1986; King 1970; Waters & Waters, 1969). A commonality

among the studies that have contradicted the theory is that they have departed from

the traditional interview method (Gardner, 1977; Salancik & Pfeffer, 1977). The

interview method is criticised for being retrospective, and selective (Gardner, 1977).

The employees are expected to more readily recall positive events which reflect upon

themselves, and negative events which can be attributed to external conditions

(Vroom, 1964). As a result, many researchers have tested Herzberg et al’s., (1959)

theory with rating scales.

1.4.3.3 Rating Scales

One example of such a study is Waters and Waters (1969) study of office

employees. Rather than using Herzberg et al’s., (1959) critical incidents interview,

employees completed a job satisfaction scale, a job dissatisfaction scale (as these are

proposed to be two separate dimensions), and a scale examining satisfaction with

24

specific facets of work. They correlated facet satisfaction with overall satisfaction

and overall dissatisfaction.

According to Herzberg et al’s., (1959) theory, it was expected that the

motivator factors (i.e., responsibility, work, sense of achievement etc.) would

correlate with overall satisfaction more than overall dissatisfaction. This finding was

not supported as the pattern of relationships with satisfaction and dissatisfaction were

similar (i.e., responsibility of job correlated with satisfaction r = 0.41 and with

dissatisfaction, r = -0.37). Similar results were obtained for the hygiene factor,

where for example, competent supervision correlated r = 0.44 with satisfaction and

r = -0.40 with dissatisfaction, and salary correlated r = 0.43 with satisfaction and

r = –0.28 with dissatisfaction. As motivator and hygiene factors acted as both

satisfiers and dissatisfiers, this study did not provide support for the two-factor

theory.

Other researchers who have relied on rating scales have also found that their

results fail to support the theory. For example, Brenner et al., (1971) conducted a

study on accountants, assessing “how much is there now” for each motivator and

hygiene factor. They correlated each of the items with a measure of overall job

satisfaction. Consistent with the two-factor theory, the motivator factors were

positively related to measures of job satisfaction, with the correlations ranging from

r = 0.39 to r = 0.62. Inconsistently however, the hygiene factors were also positively

related to job satisfaction, with the correlations ranging from r = 0.41 to r = 0.59.

These findings, fail to conform with Herzberg et al’s., (1959) theory, and suggest that

as motivator and hygiene factors increase, job satisfaction increases.

25

Although Waters and Waters (1969) and Brenner et al’s., (1971) studies

failed to support the two-factor theory using a rating scale, Hill’s (1986) study claims

to offer more support. Hill (1986) developed a 45-item questionnaire to measure

intrinsic and extrinsic factors of work in academia. The intrinsic factors

(i.e., teaching, convenience, recognition-support) were similar to motivator factors,

whilst the extrinsic factors (i.e., economic, administration, and collegial) were similar

to the hygiene factors. It was expected that the intrinsic factors would lead to job

satisfaction and that the extrinsic factors would lead to job dissatisfaction. To test

this proposal, Hill (1986) compared the mean level of satisfaction with each

dimension. The employees were more satisfied with the intrinsic dimension

(M = 4.43) than the extrinsic dimension (M = 4.18). Specifically, the following

means were observed where one is very dissatisfied and six is very satisfied: teaching

(M = 4.82), convenience (M = 4.52), recognition-support (M = 3.96), economic

(M = 4.24), administration (M = 4.00), and collegial (M = 4.23). From these results,

Hill (1986) concluded that the academics’ dissatisfaction with their work came from

extrinsic factors (i.e., hygiene factors), whilst their satisfaction came from intrinsic

factors (i.e., motivator factors).

The validity of this conclusion is questioned however, as the mean level of

satisfaction for the intrinsic and extrinsic factors were very similar. The difference

was significant, however this may be due, in part, to the large sample size

(N = 1000). More importantly however, it must be questioned whether Hill’s (1986)

study is even testing Herzberg et al’s., (1959) theory. The two-factor theory did not

propose that employees are more satisfied with the motivator factors than the

26

hygiene factors, but rather that the motivators serve to bring about job satisfaction,

and hygiene factors prevent job dissatisfaction. As such, although Hill’s (1986)

study claims to support the two-factor theory using a rating scale, the validity of the

findings are questioned.

In summary, it appears that studies testing the two-factor theory using rating

scales tend to be inconsistent with those using the interview method. The rating

scale may be superior to the interview method, however it is still problematic

(Herzberg, 1966; Silver, 1987; Whitsett & Winslow, 1967). Researchers propose

that the rating scales may induce respondents to indicate an attitude towards every

item, even on items that they have never thought about before (Herzberg, 1966).

Furthermore, there is pressure for the respondents to appear rational when they report

their satisfaction with the job facets and overall satisfaction, where they may attempt

to keep their responses consistent. As a result of these limitations, some researchers

have opted for free response scales (e.g., Silver, 1987).

1.4.3.4 Free Response Scales

Studies attempting to overcome the limits of both interview and ratings scales

have relied on free response scales. These scales are not retrospective, and allow the

employee to develop their own answers. For example, Friesen, Holdaway and Rice’s

(1983) study of school Principals relied on two questions including “which two

factors contribute most to your overall satisfaction with the principalship” and

“which two factors contribute most to your overall dissatisfaction with the

principalship.” They then calculated how often the Principals mentioned motivator

27

factors and hygiene factors when they referred to sources of their satisfaction and

dissatisfaction. These were converted into ratios, which included the number of

times each factor was mentioned as a satisfier, and the number of times each factor

was mentioned as a dissatisfier (satisfier: dissatisfier). For example, sense of

achievement was reported as a source of satisfaction 85 times, and a source of

dissatisfaction 5 times (i.e., 85: 5). Other factors that were reported as satisfiers

more than dissatisfiers included interpersonal relationship (77: 0), importance of the

work (24: 0), and relationship with central office (11: 0). These findings were

generally consistent with the two-factor theory, the exception being factors involving

relationships (i.e., interpersonal relationships and relationships with central office).

Relationship factors are hygiene factors, and as such, are expected to be reported as

dissatisfiers more than satisfiers.

The factors that were mentioned more as dissatisfiers than satisfiers include

amount of work (0: 68), overall constraints (0: 56), attitudes of society (0: 49), stress

(0: 21) and impact on home life (0: 14). These were also generally consistent with

the two-factor theory.

It must be noted however that many other factors were identified as sources

of both satisfaction and dissatisfaction, such as relationship with teachers (94: 42),

responsibility (81: 20), autonomy (70: 19), student attitudes (51: 25), challenge of

work (41: 36), relationships with parents (22: 51) and salary (6: 7). In fact, only

eight of the 20 factors occurred uniquely as either satisfiers or dissatisfiers and two

of these were in the wrong direction (i.e., interpersonal relationships, relationships

with central office). Hence, although researchers have proposed that this study

28

“represents a major step in resolving the controversy in favour of Herzberg’s

assertion” (Silver, 1987, p. 5), it provides at best, only partial support.

A similar study was conducted on educators by Silver (1987). The

participants were required to think of a time when they felt especially good/bad about

their jobs, and write a paragraph describing what happened. It was hypothesised that

the employees would cite motivator factors more often than hygiene factors when

describing positive events, and cite hygiene factors more often than motivator factors

when describing negative events. As hypothesised, the employees mentioned more

motivator factors (85) than hygiene factors (6), when recalling a positive event.

Inconsistently however, the employees reported more motivator factors (48) than

hygiene factors (40), when recalling a negative event. As such, Silver’s (1987) study

provides only partial support for the two-factor theory.

Silver (1987) conducted a second study using a questionnaire developed by

Wernimont (1966). The questionnaire contained two lists of statements, one positive

and one negative, each referring to one of Herzberg et al’s., (1959) 16 categories.

The participants were required to indicate whether an event had occurred, and then to

indicate whether it was a positive or negative event. For example, for the pay facet,

on the negative list was “the pay increase I got was insufficient for putting some

aside for the future” and on the positive list was “I received a substantial increase in

pay.” It was hypothesised that on the positive-feelings list, respondents would check

more motivator than hygiene items, and on the negative-feelings list, respondents

would check more hygiene than motivator items. On the positive list, the employees

checked 322 motivator factors and 259 hygiene factors, whilst on the negative list,

29

they checked 255 hygiene and 178 motivators factors. These results are assumed to

be supportive of the two-factor theory as respondents checked more motivator than

hygiene factors on the positive list and more hygiene than motivator factors on the

negative list. However, it is concerning that motivator and hygiene factors were

reported for both positive and negative events.

1.4.3.5 Summary

Studies that contradict the two-factor theory tend to depart from the

traditional interview method. These studies, relying on rating scales or free response

scales, claim to provide some support for the theory. Closer examination of the

results however, demonstrates that these studies provide at best, partial support of the

theory.

1.4.3.6 Criticism Three: Ambiguous Hypotheses and Criterion Measures

Researchers testing the two-factor theory have been criticised for employing

several different hypotheses and criterion measures (King, 1970). First, in regard to

the hypotheses, King (1970) cites several different ways that researchers test the

main propositions of the theory. Some researchers propose that all motivator factors

combined together should contribute more to job satisfaction than job dissatisfaction,

and that all hygiene factors combined should contribute more to job dissatisfaction

than job satisfaction. Other researchers examine each factor separately, proposing

that each motivator factor should contribute more to job satisfaction than job

dissatisfaction, and each hygiene factor should contribute more to job dissatisfaction

30

than job satisfaction. A more precise version of the theory proposes that only

motivators determine job satisfaction, and that only hygienes determine job

dissatisfaction. These examples serve to demonstrate that one researcher using a

broad hypothesis may report that their findings support the theory, whilst another

researcher using a specific hypothesis may report that their results are inconsistent

with the two-factor theory.

In regard to the criterion measures, researchers tend to evaluate their findings

differently (King, 1970). For example Sergiovanni (1967) conducted a study on

teachers using the critical incident technique. The results indicated that teachers

reported achievement as a source of a positive event (30) more than a source of a

negative event (9). Some researchers, including Sergiovanni (1967) propose that this

ratio is supportive of the two-factor theory as it is reported more in positive

experiences than negative experiences. However, other researchers (e.g., Friesen et

al., 1983) propose that it is not supportive as achievement was reported for some

negative experiences. Most researchers opt for the former, proposing that if one part

of the ratio is greater than the other part, the results are supportive of the two-factor

theory (i.e., Silver, 1987). Even so, these different criterion measures certainly

create confusion.

It must also be questioned whether a study can provide support for the two-

factor theory when some of the ratios are in the wrong direction (i.e., salary 20: 12).

Herzberg et al., (1993) did not comment on the issue, however they accepted results

that were not in the proposed direction in their study. King (1970) attempted to

specify some guidelines, proposing that failure to conform one item would not

31

contradict the whole theory unless that one item had a significant negative difference.

However, it still remains unclear how many items would need to be inconsistent for

the theory to be refuted.

1.4.4 Conclusion

The two-factor theory was notable for proposing that job satisfaction and job

dissatisfaction are separate continua, and that the factors which affect job satisfaction

are different to the factors which affect job dissatisfaction. The original study from

which the theory developed was methodologically flawed, and as such, it is not

surprising that empirical studies evaluating the two-factor theory often demonstrate

that motivator and hygiene factors affect both job satisfaction and job dissatisfaction.

Researchers that report supportive findings often rely on less stringent hypotheses

and criterion measures. In conclusion, the two-factor theory of job satisfaction has

received little empirical or theoretical support.

32

1.5 Vroom’s (1964) Expectancy Theory of Job Satisfaction

1.5.1 How Expectancy Theory has Contributed to our Knowledge of Job

Satisfaction

Expectancy theory (Vroom, 1964) was one of the first theories to focus on the

cognitive processes that underlie job satisfaction. It has received considerable

theoretical and empirical attention for over 30 years (Van Eerde & Thierry, 1996).

The number of studies examining expectancy theory has decreased recently however,

with only ten studies being conducted since the 1990’s (Ambrose & Kulik, 1999).

As such, this review will mainly be based on the earlier studies.

1.5.2 Description of Expectancy Theory

Expectancy theory describes its major constructs and propositions using its

own jargon. It refers to three major constructs, namely expectancy, valence, and

instrumentality. Expectancy refers to how much a person perceives that an action

will result in a certain outcome. For example, how much a person believes that if

they work harder, they will get a pay rise. Valence refers to the degree of anticipated

satisfaction or desirability of an outcome. Hence, in the previous example, the

valence would be a measure of how much the person desires a pay rise.

Instrumentality refers to the degree to which the person sees the outcome in question

as leading to the attainment of other outcomes. Hence, in our example,

instrumentality would be how much a person believes that a pay rise will result in

other outcomes, such as buying a house.

33

The way these constructs are combined depends on the variable that is being

predicted. Three dependent variables have been examined, namely job effort, job

performance and job satisfaction. This review will only examine the model

predicting job satisfaction, referred to as the valence model. This incorporates two of

the above-mentioned constructs, namely valence and instrumentality. It proposes

that job satisfaction can be predicted by multiplying the valence of an outcome by its

instrumentality. Hence, to predict job satisfaction, we would need to determine how

much a person likes or values an outcome of their job (i.e., being promoted) and

multiply this measure by how much they believe that this outcome will lead to other

outcomes (i.e., being offered a partnership in a business).

There is a great deal of ambiguity surrounding the measurement of the major

constructs in the expectancy theory (Van Eerde & Thierry, 1996). The

instrumentality construct has proved to be the most troublesome for researchers

(Wahba & House, 1978). Vroom (1964) referred to instrumentality as the

probability that an outcome will result in other outcomes (i.e., outcome-outcome

relationship), and expectancy as the probability that an action will result in an

outcome (i.e., action-outcome relationship). Researchers have confused these

variables however, and have measured instrumentality through examining the

probability that an action will result in an outcome (eg., Constantinople, 1967;

Pulakos & Schmitt, 1983; Reinharth & Wahba, 1976). These different

conceptualisations of instrumentality influence the application of the valence model

to the workplace.

34

1.5.3 Applications of the Valence Model

According to the valence model as defined by Vroom (1964), an employer

can increase their employees’ levels of job satisfaction through ensuring that

employees value the outcomes of their job (i.e., gaining admiration from other

workers, being promoted, feeling a sense of accomplishment, pay rise), and believe

that these outcomes will lead to other outcomes.

According to researchers who operationalise instrumentality as expectancy,

employers should ensure that their employees value the outcomes of their jobs, and

believe that their work will help them achieve those outcomes.

1.5.4 Studies of the Valence Model

Several early studies examined the relationship between job satisfaction and

the valence model (e.g., Constantinople, 1967; Ferris, 1977; Pulakos & Schmitt,

1983; Reinharth & Wahba, 1976; Sobel, 1971, Teas, 1981). A review of such studies

demonstrates that correlations between the valence model (valence x instrumentality)

and job satisfaction are generally positive, ranging from r = 0.03 to r = 0.57

(Mitchell, 1974). This demonstrates that together, valence and instrumentality

predict job satisfaction.

An example of a typical study conducted to assess how the valence model

influences satisfaction, is that conducted by Constantinople (1967). This study

examined how valence and instrumentality contributed to satisfaction in university

students. The students were given a list of 14 outcomes of university (e.g., learning

how to learn from books and teachers). Each outcome was rated in terms of its

35

importance (i.e., valence) and on the degree to which the university was helping the

students to achieve the outcome (i.e., instrumentality). The product of these two

ratings (i.e., instrumentality and valence) was obtained for each outcome, and the

products were summed across all 14 outcomes. This measure was then correlated

with a measure of satisfaction with college. According to the valence model, it was

expected that the valence times instrumentality interaction would be positively

related to satisfaction. The results were generally supportive of the model with the

correlations ranging from r = 0.34 to r = 0.49. It must be noted however that

Constantinople (1967) did not examine how much each component of the model

contributed to satisfaction.

1.5.5 Methodological Limitations

Although many studies testing Vroom’s (1964) valence model claim to

provide moderate support for Vroom’s (1964) expectancy theory (e.g., Ferris, 1977;

Pulakos & Schmitt, 1983; Reinharth & Wahba, 1976; Sobel, 1971, Teas, 1981), these

studies have some methodological limitations. Three such limitations have been

identified and will be discussed below as: 1) the finding that the components of the

valence model account for more of the variance in satisfaction on their own than

when combined; 2) violations of the assumptions of the multiplicative composite;

and 3) inflated correlations due to common method variance.

36

1.5.5.1 1) The Finding that the Components of the Valence Model Account for

more of the Variance in Satisfaction on their own than when Combined.

The valence model proposes that job satisfaction can be predicted by the

product of valence and instrumentality. However, many studies have demonstrated

that the components of expectancy theory account for more of the variance in

satisfaction on their own than when included in the expectancy model (e.g., Pulakos

& Schmitt, 1983; Reinharth & Wahba, 1976; Teas, 1981; Van Eerde & Thierry,

1996). In these studies, one of the components, either valence or instrumentality, has

predicted job satisfaction as well, or better than, the valence times instrumentality

interaction.

An example of such a study is that conducted by Reinharth and Wahba

(1976). They measured valence and expectancy in a sample of sales force

employees. Although instrumentality should have been included in the model, their

measure of expectancy was similar to a measure of instrumentality. They measured

expectancy by assessing the extent of agreement with the following items; “The

harder I work, the more I produce”, “there are no rewards for working hard in this

company” and “poor job performance may get me fired.” Their results demonstrated

that expectancy was as strongly correlated to job satisfaction (r = 0.43) as the

expectancy times valence interaction (r = 0.40).

Similar findings were reported in Pulakos and Schmitt’s (1983) study of

graduating students. Valence of work outcomes was assessed through rating the

importance of job facets (e.g., good pay, cooperative workers, opportunities for

personal growth), and instrumentality was assessed through rating the likelihood of

37

each facet. They correlated these measures with internal job satisfaction and external

job satisfaction. The results demonstrated that valence and instrumentality

considered separately correlated with job satisfaction as well or better than the

valence times instrumentality interaction. For example, in regard to the co-workers

facet, the correlations between the valence times instrumentality interaction (r = 0.04,

internal, r = 0.11, external) were lower than the correlation for instrumentality

considered on its own, (r = 0.11, internal, r = 0.12, external). Hence, in this example,

the valence model was not more strongly related to job satisfaction than the

components considered separately.

A recent meta-analysis of studies using the valence model to predict

occupational choice reached similar conclusions (Van Eerde & Thierry, 1996). The

results demonstrated that valence (r = 0.27) and instrumentality (r = 0.27) considered

separately correlated as well with choice as the valence times instrumentality model

(r = 0.28)

In conclusion, these studies suggest that the components of the valence model

often account for more of the variance in job satisfaction when considered separately

rather than when combined into the valence model. These results not only question

the usefulness of the two components of the valence model, but also how these

components are combined.

1.5.5.2 2) Violations of the Assumptions of the Multiplicative Composite

Although the valence model proposes that valence should be multiplied by

instrumentality, many assumptions underlying the multiplicative process may not be

38

met. First, although it is assumed that for a multiplication to be valid, the two

constructs are independent (Campbell & Pritchard, 1970), instrumentality and

valence are related to each other (e.g., r = 0.47; Pritchard & Sanders, 1973). Second,

although it is assumed that multiplicative composites are based on a ratio scales with

a true zero point (Evans, 1991), most researchers rely on interval scales (Mitchell,

1974). Some researchers have attempted to establish a zero-point on a likert scale by

having a scale that ranges from 0 to 10 (i.e., Pritchard & Sanders, 1973). This scale

does not have a true zero point, and rather, to establish a true zero point, a complex

and time-consuming process needs to be undertaken, that requires the scaling of

pairs, as well as individual outcomes or objects (Thurstone & Jones, 1957).

In summary, although the valence model proposes that the components of the

model should be multiplied, two major assumptions underlying multiplicative

composites may not be met.

1.5.5.3 3) Inflated Correlations due to Common Method Variance

Although the assumptions of the multiplicative composites are often ignored,

the correlations between the components, considered either on their own or in the

valence model, with job satisfaction, are still moderate. Critics suggest that these

moderate correlations occur as the measures of instrumentality, valence, and

satisfaction are all based on self-report (Schwab, Olian-Gottlieb & Heneman, 1979).

It has been proposed that when both the independent variables and dependent

variables are measured through self-report, they correlate higher than if one of the

variables is observed (Mitchell, 1974; Schwab et al., 1979). The problem with this

39

reasoning however is that self-report measures are expected to differ from objective

measures. Objective life satisfaction, for example, is poorly correlated with

subjective life satisfaction (r = 0.12; Cummins, 2000a). Thus, the subjective

measures cannot be verified through objective measures. Furthermore, it is the

subjective measures, which are important to the individuals’ levels of satisfaction.

As long as the employee perceives that by working hard, they will receive a pay rise

(instrumentality), and value a pay rise (valence), their satisfaction will be influenced.

As such, there is no evidence for the proposal that the correlations among variables

in the valence model are inappropriately inflated through common method variance.

Rather, the correlations used to make such claims are based on invalid comparisons

between objective and subjective variables.

Although common method variance is not deemed to be a problem in this

regard, researchers have tested the valence model using measures other than

self-report. Sobel (1971) conducted a study with students, experimentally

manipulating instrumentality. Two groups were formed; one with high

instrumentality and one with low instrumentality. Both groups were told that they

were required to complete a task of mental agility. Before completing this task, they

rated the valence of this task to themselves. They then completed the task, and their

score was calculated. They were given a table of norm probabilities which indicated

how likely it was that they would perform well on the next task. One group was

given a table of norms, which contained high probabilities (i.e., high instrumentality

group), whilst the other group was given a table of norms which contained low

probabilities (i.e., low instrumentality group). Both groups were then asked to rate

40

their satisfaction whilst considering the importance of the task, and the probability

that they would do well in the next task.

According to the valence model, it was expected that people with higher

instrumentality and higher valence would report higher satisfaction. In regard to

instrumentality, the high instrumentality group consistently reported higher

satisfaction (M = 19.5) than the low instrumentality group (M = 13.5), thus

supporting the model. In regard to the proposed interaction effect, it was expected

that for the high or low instrumentality group, people who reported high valence

would also report higher satisfaction than people who reported low valence.

Inconsistently however, the results demonstrated that for the high instrumentality

group, there was no difference in the level of satisfaction reported by the high

valence group (M = 20.3) and the low valence group (M = 18.7). Furthermore, for

the low instrumentality group, the low valence group reported significantly higher

satisfaction (M = 15.0) than the high valence group (M = 12.1).

Although these results are generally inconsistent with the valence model,

there was a major limitation in the study. The researchers failed to measure

instrumentality after the subjects had completed the intervention. As such, they

failed to demonstrate that their intervention altered levels of instrumentality.

In summary, researchers have suggested that instrumentality and valence

correlate well with job satisfaction because they are measured by self-report.

Although there is no evidence for this proposition, Sobel’s (1971) study suggests that

when the variables are experimentally manipulated, the results are inconsistent with

the valence model.

41

1.5.6 Conclusion

Expectancy theory proposes that job satisfaction can be predicted by

multiplying the valence of an outcome by its instrumentality. Reviews conducted on

the valence model have demonstrated that the two major components of the model

correlate well with job satisfaction. However, the individual components of the

model often account for more of the variance in job satisfaction than the

multiplicative composite. This has led researchers to not only question the validity

of the individual components of the model, but also the validity of the multiplicative

composite. In conclusion, while the valence model appears to be simple, it combines

a set of complex variables in a problematic manner.

42

1.6 Discrepancy Theories

1.6.1 How Discrepancy Theories have Contributed to our Knowledge of Job

Satisfaction

Discrepancy theories of job satisfaction focus on the cognitive processes that

underlie job satisfaction. These theories are particularly notable for proposing that

employees’ levels of job satisfaction are dependent on this source of comparison.

1.6.2 Description of Discrepancy Theories

Discrepancy theories propose that job satisfaction is a result of a comparison

between the perception of the current situation and some standard of comparison

(Lawler & Suttle, 1973; Locke, 1969; Michalos, 1985; Porter, 1961). Researchers

have defined this standard of comparison in various ways, including what they want,

what they feel they are entitled to, what they see others as getting, what they had in

the past, or what they expected to have (Harwood & Rice, 1992; Michalos, 1985). In

all of these theories however, the larger the difference between the perceptions of the

current situation and the standard of comparison, the lower the level of job

satisfaction.

1.6.3 Empirical Studies Investigating Discrepancy Theories

Although only a few empirical studies have examined the relationship

between discrepancy and job satisfaction, they have generally been supportive. For

example, Rice, McFarlin and Bennett (1989) measured how much employees have,

43

and want, thirteen job facets. They then calculated the amount of discrepancy

between what the employee has and what they want. They found that the perceived

have-want discrepancies were moderately negatively correlated with facet

satisfaction, where r = -0.48. Hence, as the have-want discrepancy increases,

satisfaction decreases.

Although Rice et al’s., (1989) study only examined have-want discrepancies,

similar results have also been found for other discrepancies. For example, Harwood

and Rice (1992) examined comparisons with: a) co-workers; b) what the person

believed that they should have; c) what they expected; and d) what they currently

expect. The correlations between these discrepancies and satisfaction, although all in

the predicted direction, varied depending on the comparison. The have-want

discrepancy was most highly correlated with satisfaction, where the average

correlation was r = -0.51. For the have-should have, r = -0.42, have-expected,

r = -0.33, have-expect, r = -0.25, and have-co-workers discrepancy, r = -0.22. In

summary, studies examining the discrepancies theories are generally supportive.

1.6.4 Theoretical Problems with Discrepancy Theories

Although the discrepancy between what a person has and some standard of

comparison correlates well with job satisfaction, there are difficulties in using

discrepancies to explain satisfaction (Cummins & Nistico, in press). When the

discrepancy theory is used to explain job satisfaction in the workplace, the

explanation becomes tautological. For example, the theory would propose that an

44

employee has a low level of job satisfaction because they want more from their job.

As such, these discrepancies may define job satisfaction rather than explain it.

1.6.5 Conclusion

Discrepancy theories propose that job satisfaction can be determined by

cognitive comparative processes. Empirical studies have demonstrated that the

discrepancy between what an employee has and some standard of comparison is

moderately correlated with job satisfaction. However, when discrepancies are used

as an explanation of job satisfaction, the explanation becomes tautological.

45

1.7 Job Characteristics Model (JCM; Hackman & Oldham, 1976)

1.7.1 How the Job Characteristics Model has Contributed to our Knowledge

of Job Satisfaction

The job characteristics model (Hackman & Oldham, 1976) was one of the

first theories to focus on the environmental determinants of job satisfaction.

1.7.2 Description of the Job Characteristics Model

The job characteristics model proposes that complex jobs are associated with

increased job satisfaction, motivation and performance. It postulates that five core

job characteristics are associated with positive outcomes (refer to Figure 1). These

include skill variety, task identity, task significance, autonomy, and feedback.

Skill variety is the degree to which the job requires employees to use different

skills. Task identity is the degree to which the job requires completion of a whole

piece of work. Task significance is the degree to which the job has an effect on other

peoples’ lives, and autonomy is the degree to which the job provides freedom.

Finally, feedback is the degree to which the job provides clear information about the

effectiveness of the employees’ performance.

These five variables do not directly relate to job satisfaction, rather the

relationship is mediated by three critical psychological states, including experienced

meaningfulness of the work, responsibility for outcomes, and knowledge of results.

Scores on these critical psychological states are determined by the five job

characteristics. Experienced meaningfulness of the work refers to the degree to

46

which the individual experiences the job as being meaningful and worthwhile. It is

determined by skill variety, task identity, and task significance. Experienced

responsibility, determined by autonomy, is the degree to which the individual feels

accountable for their work. Knowledge of results, determined by feedback, refers to

the degree to which the individual is aware of how they are performing the work.

These critical psychological states are expected to predict a number of

personal and work outcome measures including work motivation, work performance,

work satisfaction, absenteeism and turnover. However, the relationship between the

critical psychological states and outcomes is mediated by growth need strength.

Growth need strength is the need for personal growth and development. It is

proposed that individuals with high growth need strength will respond more

positively to their critical psychological states than those with low growth need

strength.

47

Figure 1-Job Characteristics Model

Skill VarietyTask Identity Experienced meaningfulness High internal Task Significance of work work motivation High quality Autonomy Experienced responsibility work performance for outcomes of the work High satisfaction Feedback Knowledge of the actual results with work of work activities Low absenteeism and turnover Employee Growth Need Strength

Source: Hackman, J.R., & Oldham, G.R. (1975). Development of the Job Diagnostic Survey. Journal of Applied Psychology, 60, p. 161.

Core job dimensions

Critical Psychological States

Personal and work outcomes

48

1.7.3 Empirical Studies of the Model

The job characteristics model has been extensively researched in over 200

studies (Renn & Vandenberg, 1995), and at least three reviews (Fried & Ferris, 1987;

Loher, Noe, Moeller & Fitzgerald, 1985; Roberts & Glick, 1981). These studies

have examined four major propositions of the model. These include that: 1) the five

core job characteristics contribute to the three critical psychological states; 2) the

critical psychological states will mediate the relationship between the job

characteristics and the outcome variables; 3) the model is moderated by growth need

strength; and 4) the model can be applied to the workplace. These propositions will

be examined for only one outcome variable, general satisfaction. General

satisfaction is an overall measure of the degree to which the employee is satisfied

and happy with their job (Hackman & Oldham, 1975).

1.7.3.1 Proposal One: The Five Core Job Characteristics Contribute to the

Three Critical Psychological States

As demonstrated in Figure 1, each job characteristic contributes to one

critical psychological state. The first three job characteristics (skill variety, task

identity and task significance) contribute to experienced meaningfulness. Autonomy

contributes to experienced responsibility, and feedback contributes to knowledge. It

is proposed that each job characteristic should only correlate with its designated

critical psychological state. However, studies examining the relationships among the

core job characteristics and the critical psychological states have provided only

moderate support for this proposal.

49

Generally, the job characteristics correlate with their designated critical

psychological state, but they also correlate with other critical psychological states.

For example, autonomy, as expected, correlates with experienced responsibility

(r = 0.40, Fox & Feldman, 1988; r = 0.41, Hackman & Oldham, 1976; r = 0.45, Wall,

Clegg & Jackson, 1978). However, autonomy also correlates with experienced

meaningfulness (r = 0.46, Hackman & Oldham, 1976; r = 0.37, Wall et al., 1978),

and knowledge of results (r = 0.25, Fox & Feldman, 1988; r = 0.26, Hackman &

Oldham, 1976; r = 0.32, Wall et al., 1978).

Another example of a job characteristic that correlates with more than one

critical psychological state is skill variety. Skill variety correlates with experienced

meaningfulness (r = 0.46, Fox & Feldman, 1988; r = 0.51, Hackman & Oldham,

1976; r = 0.30, Wall et al., 1978) and with experienced responsibility (r = 0.35, Fox

& Feldman, 1988; r = 0.40, Hackman & Oldham, 1976; r = 0.22, Wall et al., 1978).

These results suggest that, inconsistent with the job characteristics model, the core

job characteristics may predict several critical psychological states.

1.7.3.2 Proposal Two: The Degree to which the Critical Psychological States

Mediate the Relationship Between the Job Characteristics and the Outcome

Variables

Although the three critical psychological states are proposed to be the “causal

core of the model” (Hackman & Oldham, 1976, p. 8), only a few researchers have

examined the mediation hypothesis (e.g., Arnold & House, 1980; Fox & Feldman,

1988; Hackman & Oldham, 1976; Renn & Vandenberg, 1995; Wall et al., 1978).

50

This hypothesis proposes that the relationship between the five core job

characteristics and outcome variables is mediated by the three critical psychological

states.

This hypothesis has been tested by examining the correlations between the

job characteristics and the outcome variables before, and after, the relevant critical

psychological states have been controlled for. These results have provided some

support for the mediation hypothesis. For example, Hackman and Oldham (1976)

found that the correlations between the job characteristics and the outcome variables

were lower after controlling for the critical psychological state for skill variety and

task significance. However, for autonomy and feedback, the correlations remained

moderate (r = 0.29, r = 0.23). These results suggest that the critical psychological

states may be partial mediators for only some of the job characteristics.

The mediation hypothesis has also been tested using multiple regression

analyses. To support the mediation hypothesis, these analyses should demonstrate

that the critical psychological states account for sizeable proportions of the variance

in each of the dependent variable, and that the core job dimensions add little to this

when considered in the same analysis (Hackman & Oldham, 1976). Results have

demonstrated that the critical psychological states have accounted for sizeable

amounts of the variance in job satisfaction, where R = 0.68 (Hackman & Oldham,

1976), and R = 0.54 (Wall et al., 1978). When the five core job dimensions were

added to these analyses, the value of R increased by 0.01 in Hackman and Oldham’s

(1976) study and by 0.10 in Wall et al’s., (1978) study. This increase was significant

in Hackman and Oldham’s (1976) study, suggesting that the variance in the five core

51

job characteristics is explained by the three critical psychological states. It must be

noted however that the increase in R was small, and the significance may have

reflected that they employed a large sample size (N=658).

Although the above studies examined the five job characteristics together,

Renn and Vandenberg (1995) examined the job characteristics separately. They

examined the effects of the job characteristics before and after the relevant critical

psychological state were controlled for. They demonstrated that the effects of the job

characteristics were lower in magnitude when the critical psychological states were

controlled for, than when considered on their own. For example, when predicting

general satisfaction, the partial regression coefficient of task identity on its own was

0.27, and when meaningfulness of work was controlled for, the partial regression

coefficient was 0.20. However, the partial regression coefficient representing task

identity effects on general satisfaction after meaningfulness was controlled for was

still significant (0.20). This was the case for three of the five job characteristics.

Specifically, after the relevant critical psychological state was controlled for, the

partial regression coefficients for skill variety was 0.08, task identity was 0.20, task

significance was 0.15, autonomy was 0.53, and feedback was 0.11. These results

concur with the earlier studies that the critical psychological states are only partial

mediators of the relationship between job characteristics and outcomes.

In summary, although only a few studies have tested the mediation

hypothesis, they generally suggest that the critical psychological states are, at best,

only partial mediators of the relationship between the core job characteristics and

general satisfaction.

52

1.7.3.3 Proposal Three: The Degree to which the Model is Moderated by

Growth Need Strength.

Growth need strength is a need for personal growth and development. It is

postulated that people who have a high need for personal growth will respond more

positively to the critical psychological states than people who have a low need for

personal growth. Although an early study conducted by Hackman and Oldham

(1976) demonstrated that the relationship between the critical psychological states

and general satisfaction was significantly higher for employees with high growth

need strength than for those with low growth need strength, later studies have been

less supportive (Champoux, 1980; Fried & Ferris, 1987; Tiegs, Tetrick, & Fried,

1992).

For example, Tiegs et al., (1992) tested the moderating role of growth need

strength with over 6,000 subjects. Using univariate and multivariate hierarchical

moderated regression analyses, they demonstrated that growth need strength did not

moderate the relationships among job characteristics, critical psychological states,

and motivation and affective outcomes. In summary, more recent studies have

questioned the moderating role of growth need strength.

1.7.3.4 Proposal Four: Applying the Job Characteristics Model to Work

Organisations

According to the job characteristics model, an employer can increase job

satisfaction through increasing the five job characteristics (e.g., skill variety, task

identity etc). Through increasing these job characteristics, the employees’ critical

53

psychological states will increase, and job satisfaction will subsequently increase.

However, as tests of the theory have not examined changes in job characteristics, and

the theory does not specify how to make changes to the job characteristics, it may be

problematic to apply the job characteristics model to the workplace.

Tests of the theory tend focus on naturally occurring variations rather than

examining changes in job characteristics. However, the effects of changing the job

characteristics for an employee through job re-design may have different effects than

if the person was recruited into the already re-designed job (Kelly, 1992). This is

important because if the model were applied to a workplace, the five job

characteristics would be changed in an attempt to increase job satisfaction.

As researchers have not examined the effects of changing the job

characteristics, there is little research specifying how to change the job

characteristics (Roberts & Glick, 1981). Researchers have attempted to change them

using their own techniques, however these have not been particularly successful.

Kelly (1992) reviewed such studies, demonstrating that job re-design led to

improvements in job satisfaction in 17 out of 30 cases, a distribution that was not

significantly different from chance. This suggests that job re-design did not

consistently lead to increased job satisfaction. It must be noted however, that in

many of the studies, the employees did not alter their perceptions of the job after job

re-design. When this finding was taken into account, perceptions of job content and

job satisfaction were associated. The important finding from this review however is

that job re-design may not change employees’ perceptions of their jobs. This finding

has serious implications for employers intending to implement job re-design. It may

54

be costly and time-consuming to change the job characteristics, particularly if only a

few employees recognise and benefit from the changes.

1.7.4 Conclusion

The job characteristics model focuses on the environmental determinants of

job satisfaction. The model proposes that five job characteristics relate to job

satisfaction through influencing three critical psychological states. Empirical tests of

the model have provided partial support for the main propositions, however these

tests have also demonstrated that many of the relationships that exist between

variables were excluded from the model. Even if these relationships were added to

the model, practical difficulties in applying the findings to the workplace reduce the

usefulness of the theory.

55

1.8 Job Demand-Control Model (Karasek, 1979; Karasek & Theorell, 1990)

1.8.1 How the Job Demand-Control Model Contributes to our Understanding

of Job Satisfaction

The job demand-control model, developed by Karasek (1979) is one of the

most well known models in the occupational job stress literature (Fletcher & Jones,

1993). Like the job characteristics model (Hackman & Oldham, 1976), it focuses on

the characteristics of the job rather than the person. Unlike the job characteristics

model however, it proposes that job satisfaction can be increased without altering

work demands.

1.8.2 Description of the Job Demand-Control Model

The job demand-control model proposes that job satisfaction is a function of

the job demands placed on the worker (job demands), and the discretion permitted in

deciding how to address those demands (job decision latitude; Karasek & Theorell,

1990). Job demands are the psychological stressors in the work environment (i.e.,

high pressure of time, high working pace, difficult and mentally exacting work). Job

decision latitude is the worker’s potential control over his/her tasks and conduct

during the working day. Using the job demand and job decision latitude dimensions,

the job demand-control model predicts four outcomes. Two of these outcomes occur

when job demands are high (i.e., active model, high-strain model), whilst the other

two occur when job demands are low (i.e., low-strain model, and passive model).

56

The most positive outcomes, including learning and growth, result from

active jobs, where both job demands and job decision latitude are high. Although

high job demands increase physiological arousal (i.e., increase heart rate or

adrenaline), high job decision latitude allows this arousal to be reduced. Workers

with high job decision latitude redirect the arousal into an appropriate response.

They can choose how they deal with their demands. Through dealing with demands

in their own way, they can reduce the arousal.

A high strain job is one in which job demands are high and job decision

latitude is low. This type of job results in the most adverse reactions of

psychological strain (i.e., fatigue, anxiety, depression, physical illness). This is

because the arousal from the high job demands cannot be redirected. As the

employees have low job decision latitude, they cannot choose how to handle their

work demands. As a result, their arousal increases, producing a larger physiological

reaction.

The two other models are the low-strain model and the passive model. Low

strain jobs are those in which job demands are low and job decision latitude is high.

These low-strain jobs, although clearly not common, may characterise some self-

employed workers, who only have the occasional customer. Employees in these jobs

have a low risk of job strain as they have few demands that produce arousal. Even

when they do have the occasional demand, they can redirect the arousal into an

appropriate response. Finally, a passive job is one in which both job demands and

job decision latitude are low. Employees in these jobs face few challenges and are

57

unable to test ideas for improving the work environment. As a result, they often

suffer from reduced work motivation.

More recently, in addition to job demands and job decision latitude, Karasek

and Theorell (1990) added social support at work. Social support at work is defined

as the “overall levels of helpful social interaction available on the job from both co-

workers and supervisors” (Karasek & Theorell, 1990, p. 69). It is proposed that

social support at work is positively related to job satisfaction, and that job demands,

job decision latitude and social support at work interact to predict job satisfaction.

1.8.3 Empirical Studies of the Job Demand-Control Model

Initial tests of the job demand-control model demonstrated that both job

demands and job decision latitude predicted a number of dependent variables,

including exhaustion, depression, job dissatisfaction, life satisfaction, pill

consumption and sick days (Karasek, 1979). Job demands were positively related to

these variables, whilst job decision latitude was negatively related to these variables.

Replications of Karasek’s (1979) study have demonstrated that job demands and job

decision latitude separately predict the dependent variables (Dwyer & Ganster, 1991;

Fletcher & Jones, 1993; Payne & Fletcher, 1983; Spector, 1987; Warr, 1990).

Although these results are supportive of the model, the central proposition of

the job-demand control model is that job demands and job decision latitude interact

to predict job strain. This interaction effect was tested through regression analyses

where the interaction term was added (Karasek, 1979). These analyses demonstrated

that job demands and job decision latitude interacted to predict exhaustion, job

58

dissatisfaction, and life dissatisfaction. The following beta values were observed for

exhaustion, (decision latitude = -0.004, demands = 0.07, interaction = 0.11), job

dissatisfaction, (decision latitude = -0.22, demands = 0.001, interaction = 0.12), and

life dissatisfaction, (decision latitude = -0.13, demands = -0.03, interaction = 0.11).

Although these interaction terms were significant, the method of analysis was

subsequently criticised (Fletcher & Jones, 1993; Ganster & Fusilier, 1989).

Researchers propose that Karasek (1979) rejected the traditional tests of interaction

based on partialed product terms in regression analyses, and rather relied on variables

that reflected differences between demands and control (Fletcher & Jones, 1993;

Ganster & Fusilier, 1989).

When researchers have replicated these analyses using an appropriate test of

the interaction effect specified by Cohen and Cohen (1983), the interaction effect

tends to be insignificant (Fletcher & Jones, 1993; Payne & Fletcher, 1983; Warr,

1990). For example, Payne and Fletcher (1983) tested the job demand-control model

on secondary school teachers. Using multiple regression they demonstrated that the

interaction term did not predict the dependent variables, including depression,

anxiety, obsession, somatic symptoms, and cognitive failures.

Although these studies suggest that job demands and job decision latitude do

not interact to predict job satisfaction, a major problem has been identified in the

measurement of job decision latitude. Job decision latitude is defined as “the

working individual’s potential control over his tasks and his conduct during the

working day” (Karasek, 1979, p. 289-290). However, the most recent measure of job

decision latitude, developed by Karasek and Theorell (1990), includes items

59

reflecting decision latitude and decision authority. Decision latitude refers to

whether the job involves learning new things, and developing skills. Decision

authority refers to whether the person has the freedom to make their own decisions

and if they can choose how they perform their work. Although the decision authority

items are consistent with the definition of job decision latitude, the decision latitude

items have been criticised for measuring skill level, skill variety, and job complexity

(Ganster, 1989). Factor analyses of this scale have confirmed that two factors

emerge (Smith, Tisak, Hahn, & Schmeider, 1997), namely decision latitude and

decision authority. However, only the decision authority items are consistent with

the definition.

Many researchers have proposed that the definition of job decision latitude is

similar to the definition of job autonomy (de Jonge, Breukelen, Landeweerd &

Nijhuis, 1999; Ganster & Fusilier, 1989; Spector, 1986). Indeed, Ganster and

Fusilier (1989, p. 256) propose that the “definition of job decision latitude mirrors

job autonomy.” Job decision latitude is “the working individuals potential control

over his tasks and his conduct during the working day” (Karasek, 1979, p. 289-290),

whilst job autonomy is “the degree to which the job provides substantial freedom,

independence, and discretion to the individual in scheduling the work and in

determining the procedures to be used in carrying it out” (Hackman & Oldham,

1976, p.258). As a result of this similarity, researchers have tested the job demand-

control model using measures of job autonomy (de Jonge, Mulder & Nijhuis, 1999;

Dwyer & Ganster, 1991). It must be noted that these researchers may refer to their

60

scales as measuring job control, however job control and job autonomy appear to be

interchangeable (de Jonge et al., 1999b).

For example, the interaction effect of the job demand-control model was

tested using Ganster’s (1989, cited in Dwyer & Ganster, 1991) multidimensional

control scale. This scale examines the amount of choice employees have in several

areas of their work, such as their work tasks, work pacing, work scheduling, physical

environment, decision making, interaction and mobility.

Using regression analyses, Dwyer and Ganster (1991) demonstrated that the

interaction term predicted absenteeism, satisfaction with work, tardiness and sick

days. Specifically, the interaction term contributed an additional 15% to explaining

the variance in absences, 4% in satisfaction with work, 26% in tardiness, and 4% in

sick days. These findings suggest that further research on the job demand control

model is required using Ganster’s (1989, cited in Dwyer & Ganster, 1991) scale.

1.8.4 Conclusion

The job demand-control model is intuitively appealing, proposing that job

decision latitude can ameliorate job demands. This theory has received partial

support as job demands and job decision latitude have separately predicted the

dependent variables. Whether these two variables interact to predict job satisfaction

continues to be debated.

61

1.8.5 Extensions on the Job Demand-Control Model

Although the evidence for the job demand-control model has been equivocal,

there are two main reasons why this theory, over the other reviewed theories,

deserves further attention. First, the proposition that job autonomy can somehow

ameliorate job demands is certainly appealing to employers (Ganster & Fusilier,

1989). It suggests that employers can increase job satisfaction without altering work

demands. Second, although few researchers are continuing to investigate the other

theories, the job demand-control model continues to be the subject of many papers

(e.g., de Jonge et al., 1999a; Dollard, Winefield, Winefield, & de Jonge, 2000;

Hallqvist, Diderichsen, Theorell, Reuterwall, & Ahlbom, 1998; Lu, 1999; Parker &

Sprigg, 1999).

1.8.6 Addressing the “Gaps” in the Job Demand-Control Model

Although the job demand-control model deserves further attention, it must be

recognised that, in addition to the operationalisation of job decision latitude, there is

a major gap in the theory. This involves the explanation of how job decision latitude

results in positive outcomes.

1.8.6.1 The Explanation of how Job Decision Latitude Results in Positive

Outcomes

The model proposes that job decision latitude increases job satisfaction by

allowing employees to redirect the physiological arousal produced from job

demands. Specifically, Karasek and Theorell (1990) propose that employees with

62

high job decision latitude can translate the physiological arousal produced from job

demands into action through effective problem solving. They propose that workers

with high job autonomy are “given the freedom to decide what is the most effective

course of action in response to a stressor” (Karasek & Theorell, 1990, p. 36). Job

decision latitude gives employees the “freedom of action in accomplishing the formal

work task…and the freedom to engage in the informal rituals” (Karasek & Theorell,

1990, p. 34).

A major problem with this explanation however is that it is tautological. This

explanation proposes that job decision latitude, or the ability to choose at work, is

beneficial because it allows people to choose how they deal with their work

demands. Furthermore, the model is non-specific, failing to discuss how the

physiological arousal produced from job demands is redirected, and failing to define

the most effective course of action. As such, it is unknown how a person with low

job decision latitude handles a job demand, and how this is different from a person

with high job decision latitude. In response to this criticism, a new explanation for

the relationship between job decision latitude and job satisfaction is developed. This

explanation specifies how employees with low job autonomy differ from employees

with high job autonomy.

63

1.9 Development of a new Explanation for the Relationship Between Job

Autonomy and Job Satisfaction: Influencing Employees’ Responses to

Work Difficulties

The job demand-control model proposes that workers with higher job

autonomy have higher job satisfaction because they can channel the arousal produced

from job demands into an appropriate response. A new explanation is developed

which proposes that employees with low job autonomy respond differently to work

difficulties than employees with high job autonomy.

It must be noted that this explanation focuses on work difficulties rather than

job demands. Job demands are the psychological stressors in the work environment

(i.e., high pressure of time, high working pace, difficult and mentally exacting work;

Karasek & Theorell, 1990). It is expected that job autonomy will influence

employees’ responses to these job demands, but that the hypothesis can be extended

to any type of work difficulty. Thus, job autonomy is expected to influence

employees’ responses to their supervisors, co-workers, pay, opportunities for

promotion, and so forth.

In response to a work difficulty, employees can change the situation to suit

themselves, or they can change themselves to suit the situation (Heckhausen &

Schulz, 1995; Rothbaum, Weisz & Snyder, 1982). These two strategies are referred

to as primary control and secondary control strategies respectively. Before

discussing how job autonomy influences the control strategies that employees use,

the two strategies will firstly be examined. Specifically, the nature of the strategies

64

will be examined, followed by a discussion of the strategies that people generally use

and the most adaptive strategies.

1.9.1 a) Primary Control Strategies and Secondary Control Strategies

Two strategies implemented by employees when they face difficult situations

are primary control strategies and secondary control strategies (Rothbaum et al.,

1982). Primary control involves changing the work environment to suit one’s needs,

whilst secondary control strategies involve changing oneself to suit the work

environment. For example, if an employee felt they were being underpaid, they

could use a primary control strategy, such as confronting their employer, or they

could use a secondary control strategy and compare themselves to others who are

worse off.

This conceptualisation of primary and secondary control is similar to Lazarus

and Folkman’s (1984) conceptualisation of problem-focussed coping and emotion-

focussed coping. In this case, coping refers to the “constantly changing cognitive

and behavioral efforts to manage specific external and/ or internal demands that are

appraised as taxing or exceeding the resources of the person” (Lazarus & Folkman,

1984, p. 141). Coping strategies are employed to manage the problem causing the

distress (i.e., problem-focussed coping) and to regulate the accompanying emotions

(i.e., emotion-focussed coping; Folkman & Lazarus, 1980).

The theory underlying problem-focussed and emotion-focussed coping and

the questionnaire designed to assess these strategies (i.e., Ways of Coping

Questionnaire; WCQ; Folkman & Larazus, 1980) is shrouded in methodological

65

limitations (Edwards & O’Neill, 1998). First, the definition of coping focuses on

managing demands that tax or exceed personal resources. Thus, coping strategies

should manage or reduce demands and enhance personal resources to meet demands.

The Ways of Coping Questionnaire (Folkman & Larazus, 1980) examines how an

individual can cope with a situation by changing the environment or the self,

however it is not specified how these strategies manage demands or enhance

personal resources (Edwards & O’Neill, 1998).

Second, there is often a great deal of overlap among the coping dimensions,

where some problem-focussed coping strategies resemble emotion-focussed coping

strategies (Edwards & O’Neill, 1998). Problem-focussed coping is aimed at problem

solving, or doing something to alter the situation, however it also includes strategies

that alter the self. For example, the problem-focussed coping strategy of “shifting

one’s aspiration level” involves the person attempting to move one’s goals to be

more in line with the current situation (Lazarus & Folkman, 1984). Furthermore,

“reducing ego involvement” involves the person attempting to reduce the overall

significance of the situation to oneself (Lazarus & Folkman, 1984). These strategies

alter the self and are more consistent with emotion-focussed coping strategies, which

aim to reduce the emotional distress associated with the problem.

Third, and perhaps most concerning is that factor analyses of the Ways of

Coping Questionnaire are highly inconsistent. Researchers have found support for

three factors (Parkes, 1984), six factors (Vitaliano, Russo, Carr, Maiuro & Becker,

1985) and eight factors (Aldwin & Revenson, 1987; Folkman et al., 1986; Lazarus &

Folkman, 1984). Edwards and O’Neill (1998) used confirmatory factor analysis to

66

evaluate these alternative factor structures, concluding that these models yielded poor

fit.

The conceptualisation of primary control and secondary control is superior to

problem-focussed and emotion-focussed coping because it maintains the distinction

between changing the environment (i.e., primary control), and changing the self (i.e.,

secondary control). The control model not only addresses responses to threat and

negative events, but also behaviour directed at growth and potential (Schulz &

Heckhausen, 1996). Furthermore, the items on the scale are consistent with the

definitions of the control strategies.

1.9.2 b) Amounts of Primary Control and Secondary Control

The life span theory of control, developed by Heckhausen and Schulz (1995)

specifies which kind of strategies people rely on throughout their life. They propose

that adults implement both primary and secondary control strategies, however in

Western samples, primary control strategies tend to be implemented first, and are

generally preferred over secondary control strategies.

Research examining the frequency of primary control and secondary control-

type strategies in the work environment is generally supportive of Heckhausen and

Schulz's (1995) propositions (e.g., Boey, 1998; Koeske, Kirk & Koeske, 1993). For

example, Boey’s (1998) study on nurses demonstrated that, using a scale ranging

from 0 to 4, on average, problem-focussed strategies (M = 2.47) were reported more

than emotion-focussed strategies (M = 1.63). In general, theoretical and empirical

research suggests that people tend to rely on primary control more than secondary

67

control. The next step is to determine whether primary control strategies are also the

most adaptive strategies.

1.9.3 c) Which Control Strategies are more Adaptive for Employees?

To determine which control strategies are more adaptive for employees,

theoretical and empirical research is examined. The theoretical propositions are

based on the life span theory of control (Heckhausen & Schulz, 1995), and the

empirical studies specify the correlations between the control strategies and job

satisfaction.

1.9.3.1 Theoretical Propositions: The Life span Theory of Control

The life span theory of control (Heckhausen & Schulz, 1995) proposes that

primary control is more adaptive than secondary control as it allows individuals to

meet their own needs. If a person successfully changes their environment using

primary control, they overcome their difficulty and also enhance their general

perceptions of control.

Secondary control strategies are less adaptive than primary control strategies,

however they have two main benefits (Heckhausen & Schulz, 1995). They

compensate for primary control failure and they assist individuals to focus on goals

that expand primary control (Heckhausen & Schulz, 1995).

Secondary control compensates for primary control failure, which may

threaten self-esteem, self-efficacy, and general perceptions of control (Heckhausen,

Schulz & Wrosch, 1997). If an individual experiences repeated primary control

68

failure, they may become vulnerable to experiencing learned helplessness. However,

if they implement secondary control after primary control failure, they can protect

their self-esteem, and reduce the likelihood of experiencing repeated primary control

failure.

For example, an individual may face a difficulty at work, where a co-worker

is working at a slow pace. To handle this difficulty, they could use a primary control

strategy or a secondary control strategy. It is expected that they would firstly

implement a primary control strategy, where they may confront their co-worker.

They may discuss the problem with them, and the co-worker may agree to put in

more effort. If this primary control strategy is successful, they overcome their

difficulty. If the strategy fails however and the co-worker continues to work at the

same pace, the employee is likely to experience a loss in their general perception of

primary control. To avoid repeating this situation, they could implement a secondary

control strategy, such as wisdom control, where they think, “I can’t always get what I

want.” Through implementing this strategy, they avoid risking repeated primary

control failure.

Secondary control strategies are also beneficial in assisting individuals to

focus on goals that expand primary control. An individual may continue to persist to

solve a difficulty if they implement secondary control strategies such as focussing on

past success. Through such a focus, the individual may feel more confident in their

ability to overcome the problem. In summary, theoretically, primary control is more

adaptive than secondary control.

69

1.9.3.2 ii) Empirical Studies Examining the Adaptiveness of Primary and

Secondary Control

Empirical studies examining the relationship between primary control and

secondary control-type strategies and job satisfaction/job stress have provided some

support for the life span theory of control (Boey, 1998; Burke & Greenglass, 2000;

Koeske et al., 1993; Kohn, Hay & Legere, 1994; Norman, Collins, Conner, Martin &

Rance, 1995).

These studies generally demonstrate that primary control strategies are more

positively related to job satisfaction than secondary control strategies. For example,

Norman et al’s., (1995) study of teleworkers demonstrated that problem-focussed

coping was positively correlated with job satisfaction (r = 0.33) and

emotion-focussed coping was negatively related to job satisfaction (r = -0.22). In

Burke and Greenglass’s (2000) study of nurses, control coping was also positively

related to job satisfaction (r = 0.14) and escape coping was negatively related to job

satisfaction (r = -0.12). Furthermore, in Kohn et al’s., (1994) study of teachers,

task-oriented coping was negatively related to perceived stress (r = -0.38) and

emotion-oriented coping was positively related to stress (r = 0.63).

Although these results suggest that, consistent with the life span theory of

control, primary control is more adaptive than secondary control, it is important to

note however that these studies have conceptualised primary control and secondary-

type strategies using several different constructs and scales.

For primary control, many of the scales are poorly designed, including items

that do not appear to measure primary control-type strategies. For example, Burke

70

and Greenglass (1995) relied on Latack’s (1986) measure of control coping. Control

coping refers to actions and cognitive reappraisals that are proactive. Many of these

items refer to secondary control-type strategies that make the person feel better about

the problem. For example, the item “try to see the situations as an opportunity to

learn and develop new skills” is measuring a secondary control strategy known as

positive re-interpretation. Furthermore, the items “try to think of myself as a winner-

as someone who always comes through” and “tell myself that I can probably work

things out to my advantage” refers to another secondary control strategy known as

illusory optimism.

Another scale which confounds primary control strategies with secondary

control type strategies is the control coping scale implemented in Koeske et al’s.,

(1993) study. Many of the items included in this scale appear to measure secondary

control strategies. For example, “talked with spouse or other relative about the

problem”, “tried to see the positive side of the situation”, “got busy with other things

to keep my mind off the problem”, “told myself things that helped me feel better”,

“let my feelings out somehow”, and “exercising more.” These strategies attempt to

make the person feel better, rather than change a situation.

The secondary control scales have also been criticised for confounding

secondary control strategies with primary control strategies. For example, Burke and

Greenglass (2000) relied on Latack’s (1986) measure of escape coping. This scale

included primary control-type strategies, such as “delegate work to others” and “set

my own priorities based on what I like to do.” Many other scales focused on

negative responses, such as avoidance and denial. For example, Norman et al's.,

71

(1995) revised version of the COPE scale relied on only five emotion-focussed

coping items, such as “I use alcohol or drugs to make me feel better” and “I give up

the attempt to get what I want.” Furthermore, Boey (1998) measured avoidance

coping through items involving suppression of feelings, blaming others, and getting

mad at people (i.e., taking more tranquillising drugs, drinking more, avoided being

with people in general).

Although avoidance and denial are two types of secondary control, there are

many other ways that people can change the self to fit in with the environment.

Fourteen secondary control strategies have actually been identified in the Primary

and Secondary Scale (Heeps, Croft & Cummins, 2000). These strategies, displayed

in Table 2, concur with Rothbaum et al’s., (1982) and Heckhausen and Schulz's

(1995) definition of secondary control.

In summary, the empirical studies suggest that secondary control-type

strategies are negatively related to job satisfaction. However, these findings may not

be generalised to secondary control as conceptualised by Heckhausen and Schulz

(1995), since these authors recognise that there are many positive secondary control

strategies.

72

Table 2- Secondary Control Strategies

Item Secondary control strategyI can see that something good will come of it. Positive re-interpretationI remember you can't always get what you want. WisdomI know things will work out OK in the end. Illusory-optimismI remember I am better off than many other people. Downward social

comparisonI remember I have already accomplished a lot in life.

Past success

I remember the success of my family and friends. VicariousI think nice thoughts to take my mind off it. Positive approachI tell myself it doesn't matter. Goal disengagementI don't feel disappointed because I knew it mightHappen.

Predictive-negative

I can see it was not my fault. AttributionI ignore it by thinking about other things. Active avoidanceI realise I didn't need to control it anyway. Sour grapesI think about my success in other areas. Present success

Source: Heeps, L., Croft, C., & Cummins, R.A. (2000). Primary control and Secondary Control Scale (2nded.). Melbourne: Deakin University.

1.9.3.3 Comparing the Life Span Theory of Control and Empirical Studies

Examining the most Adaptive Control Strategy

The life span theory of control (Heckhausen & Schulz, 1995) proposes that

primary control strategies are more adaptive than secondary control strategies as they

allow individuals to meet their own needs, and they facilitate general perceptions of

control. Secondary control strategies are still useful however in compensating for

primary control failure and assisting individuals to focus on goals that expand

primary control. The empirical studies partly concur with these propositions,

demonstrating that primary control-type strategies are positively related to job

satisfaction. Inconsistently however, several studies demonstrate that secondary

73

control type strategies are negatively related to job satisfaction (Boey, 1998; Burke &

Greenglass, 2000; Friedman et al., 1992; Koeske et al., 1993; Kohn et al., 1994;

Norman et al., 1995). It must be noted however, that these empirical studies have

relied on many different scales, some of which are methodologically flawed.

1.9.4 Summary

Employees implement primary control and secondary control strategies when

they face a difficulty at work, however they tend to rely on primary control more

than secondary control. Primary control strategies allow individuals to meet their

own needs, and are positively related to job satisfaction. Secondary control

strategies are assumed to compensate for primary control failure, and assist

individuals to focus on goals that expand primary control. Although they have been

negatively related to job satisfaction in previous studies, the scales have been

criticised for focussing on negative strategies. It is expected that secondary control

strategies, as assessed through the Primary Control and Secondary Control Scale

(Heeps et al., 2000), will be beneficial for people after they have experienced

primary control failure.

1.10 Explaining the Relationship Between Job Autonomy and Job Satisfaction:

How Job Autonomy Influences Primary and Secondary Control

The explanation for the relationship between job autonomy and job

satisfaction proposes that job autonomy influences the way employees respond to

74

their work difficulties. It is expected that job autonomy will influence the use and

adaptiveness of primary and secondary control strategies.

Past research has tended to confuse job autonomy and primary control

(Thompson, Collins, Newcomb & Hunt, 1996) and as such, these two will be

differentiated. Job autonomy refers to whether employees perceive that they can

control aspects of their work environment, whereas primary control is a strategy that

employees use to change their work environment. An employee who has high job

autonomy perceives that they can choose, or control many aspects of their work. An

employee who has high primary control perceives that they change their environment

when they face a difficulty at work.

Although job autonomy and primary control are different, they are expected

to be related to each another. Specifically, job autonomy should influence: 1), which

control strategies employees rely on; and 2) the adaptiveness of the control strategies

(i.e., the relationship between the control strategies and job satisfaction).

1.10.1 1) Use of Primary and Secondary Control

It is expected that all individuals, with either low or high job autonomy, will

implement both primary control and secondary control strategies. Both groups will

implement primary control strategies first, and if they experience primary control

failure, they will then implement secondary control strategies. The difference

between the two groups lies in the amount of primary control failure that they

experience. Employees with low job autonomy have little influence over their work

environment are expected to experience more primary control failure than employees

75

with high job autonomy. As they need to compensate for this failure, it is expected

that these employees (i.e., low job autonomy) will implement more secondary

control than employees with high job autonomy. Hence, it is proposed that the

ability to choose is inversely related to the probability of primary control failure,

which in turn, influences the use of secondary control strategies.

These propositions are based on Heckhausen and Schulz’s (1995) life span

theory of control. This theory proposes that as people age, they experience reduced

autonomy, and they begin to experience primary control failure more often. To

compensate for this primary control failure, they need to increase their reliance on

secondary control strategies. For example, an older individual with restricted

mobility may experience primary control failure when working hard to maintain their

garden. To reduce the amount of primary control failure that they experience, they

can rely on secondary control strategies such as downward comparison (e.g. “I am

better off than others my age”). The proposal that older people rely on more

secondary control strategies than younger people has been confirmed in several

studies (i.e., Chipperfield, Perry & Menec, 1999; Maher & Cummins, 2001;

McConatha & Huba, 1999).

1.10.2 2) Adaptiveness of Primary and Secondary Control

In addition to influencing the relative use of primary and secondary control

strategies, job autonomy may also influence the adaptiveness of such strategies.

Although it was previously demonstrated that primary control-type strategies were

positively related to job satisfaction, and secondary control-type strategies were

76

negatively related to job satisfaction, it has been suggested that these relationships

may change if the person perceives that the situation is uncontrollable (Thompson et

al., 1996; Thompson, Nanni & Levine, 1994; Thompson, Sobolew-Shubin,

Galbraith, Schwankovsky & Cruzen, 1993; Thompson et al., 1998).

Two models have been developed to explain this relationship, namely the

discrimination model and the primacy/back-up model (Thompson et al., 1998). The

discrimination model proposes that primary control is more adaptive than secondary

control when the situation is controllable, and that secondary control is more

adaptive than primary control when the situation is uncontrollable. This model

underlies the philosophy of the serenity prayer; “Grant me the strength to change

what I can, the patience to accept what I cannot, and the wisdom to know the

difference” (Thompson et al., 1998, p. 587). In regard to job autonomy, this model

suggests that primary control strategies are more adaptive for employees with high

job autonomy, and secondary control strategies are more adaptive for employees

with low job autonomy.

The primacy/back-up model, on the other hand, proposes that primary control

is more adaptive than secondary control in controllable and relatively uncontrollable

situations. The role of secondary control is only to “compensate for low primary

control, and help increase feelings of overall control” (Thompson et al., 1998,

p. 587). Thus secondary control is only beneficial when primary control is low. In

regard to job autonomy, the primacy/back-up model proposes that primary control

strategies are the most adaptive strategy for employees with high job autonomy and

for employees with low autonomy, unless primary control is low.

77

It is difficult to differentiate the primacy/back-up model from the

discrimination model. The primacy/back-up model proposes that primary control is

more adaptive than secondary control unless primary control is low. If a person has

low primary control, they believe that they cannot change the environment using an

active strategy, such as working hard. However, this means that they perceive the

situation to be uncontrollable. Hence, the primacy/back-up model is proposing that

secondary control is only useful when primary control is low, however primary

control is low when the situation is perceived as being uncontrollable. This is indeed

similar to the discrimination model, which proposes that secondary control is best in

uncontrollable situations. As such, it appears that there may be some overlap in the

models.

In order to reduce the overlap in the models, the primacy/back-up model

should be revised to propose that primary control is the most adaptive strategy in

controllable and uncontrollable situations. The proposal that secondary control is

beneficial when primary control is low needs to be excluded as it overlaps with the

discrimination model. Researchers who have tested the primacy/back-up model

generally focus on the proposal that primary control is adaptive in low-control and

high-control situations.

1.10.2.1 Empirical Studies Examining the Discrimination Model and the

Primacy/Back-Up Model

Only a few studies have examined the most adaptive control strategies in

low-control situations (Thompson et al., 1996; 1994; 1993; 1998). In a review of

78

these studies, Thompson et al., (1998) concluded that they generally supported the

primacy/back-up model. As several serious methodological problems have been

identified in these studies, they will be reviewed.

1.10.2.2 Primacy/Back-Up Model

The first study that Thompson et al., (1998) cites as supporting the

primacy/back-up model is Thompson et al’s., (1993) study. According to Thompson

et al., (1998), this study demonstrated that cancer patients with higher levels of

primary control were less depressed than those with lower levels of primary control.

Control was negatively related to maladjustment (r = -0.46) and positively related to

physical functioning (r = 0.39) and marital satisfaction (r = 0.24). As such, it was

concluded that this study supported the primacy/back-up model (Thompson et al.,

1998).

However, this conclusion is inaccurate as the study did not measure primary

control, rather it measured perceived control. The participants were firstly asked

about how much control they had over various facets of their lives (i.e., perceived

control over emotions, physical symptoms, relationship with family). They were

then asked what type of things they have done to control their feelings over each

facet, and how effective these were. The items measuring amount of perceived

control were then added to the effectiveness item for each facet. The resulting scale

assessed perceived control, and the effectiveness of the control strategies, but clearly

failed to measure primary control.

79

When the findings are reinterpreted using perceived control rather than

primary control, they are intuitive. It is not surprising that it is beneficial for cancer

patients to believe that they can control areas of their lives. Indeed, a fundamental

belief about human nature is that we have a need to control events, people and

situations (DeCharms, 1968; White, 1959). However, perceived control is not the

same as primary control. Whereas autonomy or control refers to whether a person

perceives that can change the environment, primary control refers to the specific

strategies people use to change the environment to suit their needs. As such,

Thompson et al’s., (1993) study does not adequately test the primacy/back-up model.

Another study which claims to support the primacy/back-up model is

Thompson et al’s., (1996) study on HIV-positive men in prison. They examined the

relationship between primary and secondary control and distress. Regression

analyses demonstrated that primary control was negatively related to distress and

secondary control was positively related to distress. Although these findings suggest

that people in a low-control environment should rely on primary control, the

measurement of primary and secondary control in the study is questionable.

To measure primary control, the participants were asked how much control

they had over a variety of outcomes, such as their feelings, day-to-day activities,

nutrition, and HIV-related symptoms. This measure is criticised however, as the

items do not refer to primary control strategies, but rather refer to levels of perceived

control. As mentioned previously, primary control is not the same as perceived

control. As such, a person may report that they can control their relationship with

their cellmates and how their correctional officers treat them, however this does not

80

indicate that they use primary control strategies when they have a difficulty with

their cellmates or correctional officers.

A further problem with Thompson et al’s., (1996) study concerns the

measurement of secondary control. Secondary control was measured by the

following item; “How much do you feel okay about (an outcome) because you just

accept it and don’t try to change it?” Although secondary control generally involves

acceptance of the situation, this item is criticised as it fails to make respondents

aware of the different ways they can accept a situation. For example, they can

believe that it will work out okay in the end (i.e., illusory optimism), or they can

think that they can't always get what they want (i.e., wisdom). The respondents in

Thompson et al’s., (1996) study were not made aware of these different strategies,

and as such may have underestimated their use of secondary control. A further

problem with this measure of secondary control is that it does not just ask if the

person uses acceptance, rather it confounds acceptance with feeling okay.

One final study which claims to support the primacy/back-up model is

Thompson et al’s., (1998). They examined whether adults (young, middle, and

older) use primary or secondary control to handle their appearance-related changes

due to aging. The youngest group was expected to have the most perceived control

over age-related changes, whilst the oldest group was expected to have the least.

Averaging over all age groups, primary control (r = 0.46) and secondary

control (r = 0.42) were positively related to satisfaction with physical appearance,

and primary control (r = -0.20) and secondary control (r = -0.24) were negatively

related to emotional distress. Although these correlations suggest that secondary

81

control is adaptive, multiple regression analyses indicated that secondary control was

only beneficial when primary control was low. There was no relationship between

secondary control and distress for those with high primary control, but for those with

low primary control, secondary control was negatively related to distress.

Although these results appear to provide support for the primacy/back-up

model, the items measuring primary and secondary control were poorly constructed.

Primary control was measured by the following five items rated on a scale from

strongly agree to strongly disagree: 1) “I feel that I have some control over the

effects of aging on my appearance”; 2) “I dread the thought of aging, but there is not

much I can do about it” (reversed coded); 3) “As long as I put the effort in I can keep

looking attractive”; 4) “I can stay attractive and youthful as long as possible if I just

work at it"; and 5) “I get depressed when I think about what’s coming as I get older”

(reverse coded).

These items are criticised as some of them are based on the assumption that

aging is a negative process (items two and five). For example, although Thompson

et al., (1998) proposed that a person who disagreed with the item “I dread the thought

of aging, but there is not much I can do about it” has high primary control, it may be

that they do not dread the thought of aging. Furthermore, the item “I get depressed

when I think about what’s coming as I get older” does not refer to a secondary

control strategy, and simply refers to the persons attitude towards aging. Other items

are based on the assumption that people perceive themselves as being attractive

(items three and four). For example, a person may disagree with the item “I can stay

82

attractive and youthful as long as possible if I just work at it”, not because they have

low primary control, but because they do not believe that they are attractive.

The items in the secondary control scale are also criticised for being based on

the assumption that aging is a negative process. The scale includes items such as “as

long as I know what’s coming, it doesn’t bother me too much to get older” and “I am

not worried about getting older, because I trust that God will take care of me.” These

items confound the perceptions of aging with the secondary control strategy. As

such, it is impossible to tell if the person is referring to the part of the question

referring to aging or the part referring to the strategy. For example, a respondent

may report that they strongly agree they are "not worried about getting older, because

they trust that God will take care of them” because they are not worried about getting

older, or because they trust that God will take care of them. In summary, as with the

other reviewed studies, several measurement issues limit the validity of Thompson et

al’s., (1998) findings.

1.10.2.3 Discrimination Model

One study conducted by Thompson et al., (1994) supported the

discrimination model. This study examined the relationship between primary and

secondary control and depression for men with a diagnosis of HIV. Both strategies

appeared to be adaptive for people who presumably were in a low-control situation.

Primary control (r = -0.36) and secondary control (r = -0.41) were negatively related

to depression. Furthermore, for the group that was low in primary control, secondary

control was negatively related to depression. For those high in primary control,

83

secondary control was weakly related to depression. The values of these correlations

cannot be discussed however, as the authors only reported them in graphical form.

This study provided some support for the discrimination model, however the

measurement of primary control and secondary control strategies was once again

limited. The items measuring the control strategies were the same as Thompson et

al’s., (1996) study, where primary control strategies were measured by perceived

control and secondary control strategies were measured by acceptance.

1.10.2.4 Conclusion: Does Research Support the Primacy/Back-Up Model or

the Discrimination Model?

Most of the studies reviewed thus far have concluded that their findings

support the primacy/back-up model. However as these studies have often failed to

validly measure primary control and secondary control strategies, further research is

required to determine whether primary control is adaptive in low-control situations.

This research must rely on a measure of primary and secondary control that concurs

with Heckhausen and Schulz's (1995) conceptualisation of control.

It is expected that employees with low job autonomy will rely on less primary

control and more secondary control than employees with high job autonomy.

Employees with low job autonomy are expected to have a higher probability of

failing when implementing primary control. To compensate for this primary control

failure, they can rely on secondary control.

84

1.10.3 Summary

Job autonomy refers to the perceived ability to exert choice in the work

environment. It may influence employees use of primary and secondary control, and

the adaptiveness of the control strategies. In regard to the use of the control

strategies, it is expected that the ability to choose facilitates the probability of

primary control failure, which in turn, influences the use of secondary control

strategies. In regard to the adaptiveness, empirical studies suggest that primary

control strategies are more positively related to job satisfaction than secondary

control strategies in low-control situations. These studies are criticised however for

their measurement of primary and secondary control, and it is expected that when

they are measured validly, primary control is more adaptive in controllable

situations, and secondary control is more adaptive in uncontrollable situations.

85

1.11 Other Major Predictors of Job Satisfaction

In addition to job autonomy and the control strategies, two other major

predictors of job satisfaction are examined. They are personality and life

satisfaction.

1.11.1 Personality

Researchers have recently paid considerable attention to the role of

personality in predicting job satisfaction. The most common taxonomy of

personality, the five-factor model (Costa & McCrae, 1985) includes neuroticism,

extroversion, conscientiousness, agreeableness and openness to experience.

Researchers have examined how some of these personality variables, namely

neuroticism and extroversion, influence levels of job satisfaction.

1.11.1.1 Personality and Job Satisfaction

Personality may directly influence job satisfaction. As evidence for this

proposal, researchers have demonstrated that job satisfaction is consistent over time

and across situations. For example, Staw and Ross (1985) demonstrated that job

attitudes remained consistent over time, even if the person changed employer, and/or

occupation. They conducted a longitudinal survey, administering a one-item

measure of job satisfaction to over 5000 men in 1966, 1969 and 1971. They

correlated the scores on this measure of job satisfaction over time. The correlation

between satisfaction scores when the employer and occupation were the same, were

moderate (r = 0.37 to r = 0.48). When the employer or the occupation had changed,

86

the correlations were only slightly lower (r = 0.19 to r = 0.34). These correlations

provide support for the stability of job satisfaction, however more supportive results

were provided by the regression analyses.

In the regression analyses, the authors used the 1966 and 1969 job

satisfaction scores to predict the 1971 job satisfaction scores. The prior job

satisfaction scores (i.e., 1966, 1969 data) were almost always a better predictor of the

1971 job satisfaction scores than situational variables, such as changes in pay and job

status. This was even the case when the sample had changed employers but had the

same occupation, and when the sample had changed occupation but still had the

same employer. Situational variables, including change in pay was a significant

predictor of the 1971 job satisfaction score only when the employer and occupation

had changed. However, the strength of the relationship was considerably less than

prior job attitudes. Hence, Staw and Ross’s (1985) study demonstrated that job

satisfaction scores could be predicted five years later by earlier job satisfaction

scores, even if the individual had changed their employers or changed their

occupation.

Although Staw and Ross’s (1985) study demonstrated that job satisfaction

remained stable, the authors did not specifically examine the relationship between

personality and job satisfaction. However, Staw, Bell and Clausen (1986) used

measures of childhood personality to predict adulthood levels of job satisfaction.

They combined three longitudinal surveys, and compared the subjects at early

adolescence (12-14 years), late adolescence (15-18 years) and adulthood. They

correlated childhood measures of personality with facet job satisfaction, and an

87

overall one-item measure of career satisfaction. The correlations were all positive,

ranging from r = 0.04 to r = 0.45. Hence, these results suggest that childhood

personality is related to job satisfaction in adulthood.

To add support to the proposal that job satisfaction is influenced by

dispositional variables such as personality, researchers have more recently tested

whether there is a genetic component to job satisfaction. Arvey, Bouchard, Segal

and Abraham’s (1989) studied monozygotic twins who were reared apart. They

completed the Minnesota Satisfaction Questionnaire (Weiss, Dawis, England &

Lofquist, 1967), which consists of an intrinsic satisfaction scale, an extrinsic

satisfaction scale, and a general satisfaction scale. Intraclass correlations, adjusted

for age and sex, were significant for intrinsic satisfaction (r = 0.32) and for general

satisfaction (r = 0.31). Similar findings were found by Arvey, McCall, Bouchard,

Taubman and Cavanaugh (1994) where r = 0.27, and Lykken and Tellegen (1995)

where r = 0.44 to r = 0.52.

In summary, these findings suggest that job attitudes are consistent over time,

that personality measured in adolescence predicts job satisfaction in adulthood, and

that there is a genetic component to job satisfaction. The next step is to examine the

relationship between specific personality characteristics and levels of job

satisfaction.

88

1.11.1.2 The Relationship Between Neuroticism and Extroversion and Job

Satisfaction

Neuroticism tends to be negatively related to job satisfaction, where r = -0.29

(Judge, Bono & Locke, 2000), r = -0.18 (Tokar & Subich, 1997), r = -0.25 (Terry,

Nielsen & Perchard, 1993), r = -0.21 (Smith, Organ & Near, 1983), and r = -0.40,

r = -0.26, r = -0.34 (Hart, 1999). The relationship between extroversion and job

satisfaction tends to be much weaker than that of neuroticism. The following

correlations between extroversion and job satisfaction have been reported; r = 0.25,

r = 0.08, r = 0.18 (Hart, 1999), and r = 0.16 (Tokar & Subich, 1997). In summary,

people reporting higher extroversion and lower neuroticism tend to report higher job

satisfaction.

1.11.1.3 Summary

Personality appears to be an important predictor of job satisfaction. Research

has demonstrated that job attitudes are consistent over time, and that personality

measured in adolescence predicts job satisfaction in adulthood. People high on

extroversion and low on neuroticism tend to report higher job satisfaction.

1.11.2 Life Satisfaction

Researchers have long been interested in the relationship between life

satisfaction and job satisfaction (Judge & Watanabe, 1994). Although varying

definitions and theories of life satisfaction have been proposed, theoretical and

empirical support has been provided for seven domains of life satisfaction. These

89

include material well-being, emotional well-being, productivity, health, intimacy,

safety, and community (Cummins, 1996; Felce & Perry, 1995). Before the

relationship between life satisfaction and job satisfaction is examined, it will be

demonstrated that life satisfaction, like job satisfaction, is influenced by personality.

1.11.2.1 Personality and Life Satisfaction

In addition to job satisfaction, neuroticism and extroversion also predict life

satisfaction (DeNeve, 1999). Neuroticism is negatively related to life satisfaction

where r = -0.29 to r = -0.37 (McCrae & Costa, 1991), r = -0.42 (Costa & McCrae,

1989), and r = -0.46 (Judge et al., 2000). Extroversion is positively related to life

satisfaction, (e.g., r = 0.19 to r = 0.22 (McCrae & Costa, 1991), and r = 0.17 to

r = 0.20 (Costa & McCrae, 1989). On the basis of these low to moderate

correlations, extroversion and neuroticism have been proposed as the key to the

relationship between personality and life satisfaction (DeNeve, 1999; Diener, Suh,

Lucas & Smith, 1999).

1.11.2.2 Life Satisfaction and Job Satisfaction

Job satisfaction is expected to be related to life satisfaction, as work is a

significant and central aspect of many peoples’ lives. Two models have been

developed to explain the linkage between job satisfaction and life satisfaction,

namely the spillover model, and the compensatory model (Wilensky, 1960). The

spillover model assumes that satisfaction in one domain of an individual’s life

extends into other areas. Life satisfaction may spillover into job satisfaction or job

90

satisfaction may spillover into life satisfaction. Either way, life satisfaction and job

satisfaction would be positively related. Alternatively, the compensatory model

proposes that job satisfaction and life satisfaction would be negatively related. An

employee with low job satisfaction would be expected to compensate for this by

engaging in satisfying non-work activities.

A meta-analysis of 34 studies examining the relationship between job

satisfaction and life satisfaction demonstrated that the two variables were positively

correlated, with an average correlation of r = 0.44 (Tait, Padgett & Baldwin, 1989).

Several more recent studies found correlations of similar magnitudes. For example,

Iverson and Maguire (2000) found a correlation of r = 0.23, and Beutell and

Wittig-Berman (1999) reported a correlation of r = 0.39. Judge, Locke, Durham and

Kluger (1998) found that r = 0.68 and r = 0.42, and Landry (2000) found that

r = 0.44. These findings demonstrate that job satisfaction and life satisfaction are

moderately related, and as such, support the spillover model.

Researchers have also examined how life satisfaction relates to the specific

facets of job satisfaction (Wright, Bennett & Dun, 1999; Judge & Locke, 1993).

Judge and Locke’s (1993) study of clerical workers demonstrated that life

satisfaction was positively related to all facets of job satisfaction, including nature of

work (r = 0.39), co-workers (r = 0.17), supervision (r = 0.26), pay (r = 0.35), and

promotion (r = 0.24). In Wright et al’s., (1999) study of professional card dealers,

only satisfaction with pay (r = 0.33) and satisfaction with the work itself (r = 0.28),

were related to life satisfaction. Satisfaction with supervision (r = 0.20), satisfaction

with promotional opportunities (r = 0.16), and satisfaction with co-workers

91

(r = -0.04) were not significantly related to job satisfaction.

Studies generally provide support for the spillover model, and most

researchers tend to rely on this model (Rain, Lane & Steiner, 1991). Although it has

been suggested that this model may not be appropriate for everyone, Judge and

Watanabe (1994) concluded that job satisfaction and life satisfaction were positively

related for approximately 80% of the participants in their study.

Although these studies have supported the spillover model, the methodology

has been criticised. First, common method variance has been identified as a problem

as both job satisfaction and life satisfaction are measured by self-report (Rain et al.,

1991). This issue is extremely difficult to avoid however as there is no acceptable

way to measure attitudes other than self-report. Objective measures of life

satisfaction correlate poorly with self-reported life satisfaction (Cummins, 2000a),

and behavioural measures of job satisfaction correlate only weakly with self-reported

measures of job satisfaction (Iaffaldano & Muchinsky, 1985).

The second methodological limitation concerns the cross-sectional study

designs, which cannot determine the direction of causality between two variables.

Cramer (1995) used cross-lagged correlations to examine a time-related relationship

between job satisfaction and life satisfaction over 13 months. Job satisfaction and

life satisfaction were positively related at the initial testing and also 13 months later,

suggesting that the two variables may be causally related.

In summary, the direction of the relationship between job satisfaction and life

satisfaction continues to be debated (Iverson & Maguire, 2000). It is generally

assumed that job satisfaction contributes to life satisfaction, but it is possible that life

92

satisfaction influences job satisfaction, or that the relationship is reciprocal. It is

clear however, that life satisfaction and job satisfaction are positively related. This

suggests that people with high job satisfaction will also have high life satisfaction,

and that people with low job satisfaction will also have low life satisfaction.

However, the relationship may not be quite so straightforward as life satisfaction is

held under homeostatic control.

1.11.2.3 Consistency of Life Satisfaction

Recent publications have proposed a model for the homeostatic maintenance

of life satisfaction (Cummins, 2000b). The basis of this model is the finding that life

satisfaction, when measured either by a single question about “satisfaction with life

as a whole” or by satisfaction averaged across a number of domains, is remarkably

predictable. The demonstration of this phenomenon has rested on a statistic called a

percentage of scale maximum (%SM) which converts Likert scale data into a range

from 0 to 100. Applying this statistic to the combined mean values from large

population surveys has revealed that they average 75 + 2.5%SM. In other words,

using two standard deviations to define the normative range, it can be predicted that

the mean level of life satisfaction of Western population samples will lie within the

range 70-80%SM (Cummins, 1995).

The consistency of these data provides a basis for the proposal that life

satisfaction is held under homeostatic control. The model that describes how such

homeostasis can be achieved proposes two levels of influence. The first involves an

affective “set-point range” which is determined by personality. The second level

93

involves a buffering system comprising the three processes of perceived control,

optimism, and self-esteem (Cummins, 2000b). Thus, it is proposed, through the

interaction of these mechanisms, the average life satisfaction for normative

population samples is held within the range 70-80%SM.

This model of homeostasis can be used to make predictions about the life

satisfaction of employees with low job autonomy and employees with high job

autonomy. Provided that their homeostatic systems are operating normally, their life

satisfaction is predicted to lie within the normal range. However, the homeostatic

system can be defeated by a substantial source of negative input, and the low job

autonomy group may have an increased probability of encountering such

circumstances. This may be, for example, through exposure to circumstances of

reduced personal control. Thus it is predicted that a sample of employees with low

job autonomy will contain more people experiencing homeostatic defeat than a

sample of employees with high job autonomy. The employees experiencing such

defeat are expected to report an average level of life satisfaction that approximates

the lower boundary of the normative range (70%SM) or even falls below this level.

1.11.2.4 Summary

As postulated by the spillover model, life satisfaction and job satisfaction are

positively related. Although they are expected to co-vary, life satisfaction is held

under homeostatic control and may not be free to vary. The average level of life

satisfaction reported by employees is expected to lie within 70-80%SM. They may

94

report a lower level of job satisfaction however, if they are experiencing homeostatic

defeat.

95

1.12 Model of Job Satisfaction

This review has identified five main predictors of job satisfaction. As

demonstrated in Figure 2, these include job autonomy, primary control, secondary

control, personality, and life satisfaction. The major proposal of the model is that

primary and secondary control mediate the relationship between job autonomy and

job satisfaction. This is represented by the arrows from job autonomy, through

primary and secondary control, to job satisfaction. Primary and secondary control

may not account for all of the variance in job autonomy, and thus job autonomy is

also directly related to job satisfaction.

It is expected that job autonomy influences the use and adaptiveness of the

control strategies. In terms of the use of the control strategies, employees with high

job autonomy are expected to rely on more primary control and less secondary

control than employees with low job autonomy. Employees with high job autonomy

are expected to be more successful when implementing primary control, and thus

have less need for secondary control, which serves to compensate for primary control

failure. In Figure 2, this relationship is represented by the arrow from job autonomy

to the control strategies.

In regard to the adaptiveness of the control strategies, it is proposed that

employees who match their level of job autonomy with their control strategies will

be most satisfied with their jobs. It is expected that primary control will be more

adaptive for employees with high job autonomy and that secondary control will be

more adaptive for employees who cannot control their environment. It is thus

expected that job autonomy moderates the relationship between the control strategies

96

and job satisfaction. In Figure 2, this moderation effect is represented by the

interaction terms (i.e., job autonomy x primary control, job autonomy x secondary

control). These interaction terms are expected to predict job satisfaction.

In addition to the control strategies, personality and life satisfaction are

expected to directly influence job satisfaction. People higher on extroversion and

lower on neuroticism are expected to report a higher level of job satisfaction and life

satisfaction. Life satisfaction and job satisfaction are also proposed to influence one

another.

In summary, the model proposes that job satisfaction can be predicted from

job autonomy, primary and secondary control, personality and life satisfaction. This

model will be tested in study one, with employees that are low in job autonomy and

employees that are high in job autonomy.

97

Figure 2- Model of Job Satisfaction

Job Autonomy

Primary Control

Job Satisfaction

Personality Life Satisfaction

Secondary Control

Job Autonomy x Secondary Control

Job Autonomy x Primary Control

98

2 Chapter 2 - Study One

99

2.1 Abstract

This study tests the model of job satisfaction developed in chapter 1. The major

proposal of this model is that job autonomy influences the use of the control

strategies and the relationship between the control strategies and job satisfaction

(refer to Figure 2). Employees with high job autonomy are expected to rely on more

primary control strategies and less secondary control strategies than employees with

low job autonomy. Furthermore, primary control is expected to be the most adaptive

strategy for employees with high job autonomy, whilst secondary control is expected

to be the most adaptive strategy for workers with low job autonomy. These

propositions were tested by comparing a sample of high job autonomy employees

(university academic staff) with a sample of low job autonomy employees

(supermarket register operators). As hypothesised, the academics reported higher job

autonomy and lower secondary control than the supermarket workers, however the

two groups did not report different levels of primary control. Additionally, primary

control appeared to be the most adaptive strategy for both occupational groups, and

secondary control was not related to job satisfaction. These findings are discussed in

relation to the life span theory of control and the discrimination model.

100

2.2 Proposal for Study One

The model of job satisfaction developed in chapter 1 proposes that job

autonomy relates to job satisfaction through influencing the use of the control

strategies, and the relationship between the control strategies and job satisfaction.

The model also proposes that personality and life satisfaction predict job satisfaction

(refer to Figure 2). In order to test these propositions, study one will compare

workers with low job autonomy with workers with high job autonomy. The first step

is to identify what type of employees fit into these two groups.

2.2.1 Identifying Employees with Low/High Job Autonomy

According to Ganster’s (1989, cited in Dwyer & Ganster, 1991) scale,

employees with high job autonomy can exert choice in several domains of their

work, such as in the scheduling of their rest breaks, procedures and policies, and in

the variety of tasks they perform. An occupational group that appears to exemplify

high job autonomy, is university academic staff. Their level of job autonomy has

rarely been assessed (Leung, Siu, & Spector, 2000), however academics have

traditionally had flexibility in their work, and freedom to pursue their own research

interests (Winefield, 2000). They can often choose among a variety of tasks,

including research, teaching, and administration (Fisher, 1994).

It is particularly important to study academics’ level of job autonomy as

researchers have recently suggested that “although in theory, the freedom indicative

of high control still exists, in practice, there has been a steady erosion of job control”

(Fisher, 1994, p. 61). This has been attributed to the increasing demands placed on

101

academics, where their workloads have increased and there is increasing pressure to

attract external funding (Winefield, 2000). However, the current study proposes that

even if their level of job autonomy is decreasing, they should still be in the upper

range.

Employees in the lower range of job autonomy are those that have little

opportunity to exert choice in their work. They tend to have “routinised” jobs and

have few tasks from which to choose. Supermarket register operators were selected

as representing such low autonomy workers. These workers are expected to have

little control over many aspects of their job, such as their rest breaks, the tasks they

work on, and their working pace.

102

2.3 Aims and Hypotheses

This study aims to compare the levels of job autonomy, control strategies,

and job satisfaction of supermarket register operators with academics. A number of

hypotheses have been developed as follows:

1) Job autonomy will be positively related to job satisfaction, and academics will

report higher job autonomy than supermarket workers.

This hypothesis tests the basic assumptions of the study. The study aims to

extend Karasek’s (1979) job demand-control model, proposing that employees with

high job autonomy have high job satisfaction because they rely on different control

strategies. As such, it needs to be demonstrated that job autonomy is related to job

satisfaction, and that the two groups selected for this study differ in their levels of job

autonomy as expected.

2) The academic group will report less secondary control and more primary control

than the supermarket workers.

This hypothesis examines how job autonomy influences the use of the control

strategies. As the academics are expected to have more control over their working

environment than the supermarket workers, they are more likely to successfully

change it using primary control. As secondary control is used to compensate for, and

avoid future primary control failure, it is expected that the supermarket workers will

report more secondary control than the academics.

103

3) Job autonomy is positively related to primary control and negatively related to

secondary control.

This hypothesis also tests whether job autonomy influences the use of the

control strategies, however, unlike hypothesis two, it is based on the measured level

of job autonomy rather than the assumed level.

4) Primary control will be more positively related to job satisfaction than secondary

control for the academics, and secondary control will be more positively related to

job satisfaction than primary control for the supermarket workers.

This hypothesis tests whether job autonomy influences the relationship

between the control strategies and job satisfaction. According to the discrimination

model (Thompson et al., 1998), primary control is most adaptive in controllable

situations and secondary control is most adaptive in uncontrollable situations.

Although empirical studies have generally failed to support this model, the studies

have been criticised for their measurement of primary and secondary control

strategies.

5) The relationship between primary and secondary control and job satisfaction will

be moderated by job autonomy.

It is expected that the relationship between primary and secondary control

and job satisfaction will change depending on the level of measured job autonomy.

This hypothesis is similar to hypothesis four, however, rather than being based on the

assumed level of job autonomy, it is based on the measured level of job autonomy.

104

6) The relationship between job autonomy and job satisfaction is mediated by

primary and secondary control strategies.

This hypothesis tests the proposed explanation for the relationship between

job autonomy and job satisfaction. It offers an alternative to Karasek and Theorell’s

(1990) explanation of the job demand-control model. They propose that job decision

latitude (i.e., similar to job autonomy) is positively related to job satisfaction because

it gives workers the freedom to choose how they complete their work and thereby

reduces the arousal produced from job demands. An alternative explanation, to be

tested here, is that workers with high job autonomy mostly rely on the preferred

control strategies, namely primary control.

7) Academics will report higher job satisfaction and higher life satisfaction than the

supermarket workers.

As the academics are expected to report higher job autonomy and more

primary control than the supermarket workers, they are expected to report a higher

level of job satisfaction. This level of job satisfaction is expected to be positively

related to life satisfaction.

8) Primary control, secondary control, job autonomy, personality, and life

satisfaction will predict job satisfaction.

These variables are assumed to be the major predictors of job satisfaction, as

depicted in Figure 2.

105

2.4 Method

2.4.1 Participants

The sample consisted of 104 university academic staff, and 96 supermarket

register operators. The academic group was obtained from seven Schools within

Deakin University. The response rate was 32%. The supermarket workers group

was obtained from two supermarket chains, with 16 stores being involved. As these

employees only collected a questionnaire if they were interested in participating in

the study, a response rate could not be calculated.

2.4.2 Materials

Both the academics and the supermarket workers received a plain language

statement (refer to Appendix A) and an anonymous questionnaire. The questionnaire

consisted of several scales, which measured job autonomy, job related primary

control and secondary control, job satisfaction, life satisfaction and personality.

2.4.2.1 Job Autonomy

Although this study is examining the job demand-control model, Karasek and

Theorell’s (1990) scale of job decision latitude was not used. This scale is criticised

for confounding job control with skill level, skill variety, and job complexity

(Ganster, 1989). In response to this criticism, Ganster (1989, cited in Dwyer &

Ganster, 1991) developed an work control scale which examined the amount of

choice an employee has in several areas of their work, such as their work tasks, work

106

pacing, work scheduling, physical environment, decision making, interaction, and

mobility.

Although the scale has good reliability (Fox, Dwyer & Ganster, 1993;

Ganster, Dwyer & Fox, 2001; Schaubroeck & Merritt, 1997), a factor analysis

demonstrated that two factors emerged (Smith et al., 1997). One factor included

items on job autonomy (16 items), while the other factor included items on

predictability (5 items). The predictability items include “how much can you

generally predict the amount of work you will have to do on a given day” and “how

much are you able to predict what the results of decisions you make on the job will

be.” As these predictability items load on a different factor from the job autonomy

items, they should be excluded from the scale. Hence, for the purpose of the present

study, only the former items were used.

A further potential problem with this scale is that some of the items directly

refer to control. In an attempt to disguise the purpose of the scale, these items were

changed from “control” to “choice.” For example, the item “how much control do

you have over the quality of your work” was changed to “In my job, I can choose the

quality of my work.”

Furthermore, to reduce the number of items in the scale, two repetitive items

were deleted. The items “how much control do you have over when you come to

work and leave” and “how much control do you have over when you take vacations

or days off” were deemed to be too similar to the following item; “how much control

do you have over the scheduling and duration of your rest breaks.” All items refer to

the timing and scheduling of rest breaks, and as such, only the latter item was

107

retained. This revised job autonomy scale consisted of 14 items (refer to Appendix

B). They were rated on a 10-point scale ranging from 1 (do not agree at all) to 10

(agree completely).

2.4.2.2 Primary Control and Secondary Control

The control strategies were assessed by the Primary and Secondary Control

Scale, developed by Heeps et al., (2000). This includes five items assessing primary

control strategies and 14 items assessing secondary control strategies. All of these

items were revised to be relevant to the workplace (refer to Appendix C). They were

measured on a 10-point scale, ranging from 1 (do not agree at all) to 10 (agree

completely). Although this scale was only developed recently, early factor analyses

suggest that two factors emerge (Maher & Cummins, 2001; Misajon & Cummins, in

press).

This scale was deemed to be superior to Heckhausen et al’s., (1997)

Optimisation in Primary and Secondary Control Scale (OPS). The reason this scale

was not selected merits discussion, as the proposed model of job satisfaction is partly

based on the life span theory of control. In this theory, Heckhausen and Schulz

(1995) propose that humans face two challenges in life; the need to be selective and

the need to compensate for failure. On this basis, the Optimisation in Primary and

Secondary Control Scale (Heckhausen et al., 1997) measures two types of primary

and secondary control; selective and compensatory.

Selective primary control is the investment of resources to reach goals, whilst

compensatory primary control is used when internal resources are insufficient

108

(i.e., others help, technical aids). Selective secondary control refers to self-

management directed at enhancing commitment to goals, and compensatory

secondary control serves to buffer the negative effects of failure.

An alternative simpler explanation for the use of primary and secondary

control is offered. Rather than assisting with the need to be selective and the need to

compensate for failure, it is proposed that people use control strategies whenever

they risk losing control. Examples of such situations are when people are unable to

solve a problem, or when something bad happens to them. Primary control provides

a sense of control derived from changing one’s realities, whereas secondary control

provides a sense of control derived from accepting or adjusting to one’s realities

(Halliday & Graham, 2000; Thompson et al., 1994).

As this new explanation is not consistent with the Optimisation in Primary

and Secondary Control Scale (Heckhausen et al., 1997), this scale was not

appropriate for this study. The scale includes some situations, which do not appear

to prompt the use of control strategies, such as “when I have decided on something.”

The scale also includes statements that are assessing general beliefs rather than

strategies. For example, “I invest my time in developing broad skills that can be

used in many areas”, “I stay active and involved in several different areas of life”,

and “many life goals become important to me because it is the right time for them.”

As a new explanation for the use of control strategies has been developed, the

Optimisation in Primary and Secondary Control Scale (Heckhausen et al., 1997) is

no longer appropriate. As such, the Primary and Secondary Control Scale (Heeps et

109

al., 2000) will be used in this study. This scale examines how people react to

situations where they risk losing control.

2.4.2.3 Job Satisfaction

Two scales of job satisfaction were administered; a facet scale and a global

scale. The facet scale is a revision of the Job Descriptive Index (Smith, Kendall &

Hulin, 1969). This scale, reported to be the most frequently used measure of job

satisfaction (Ironson, Smith, Brannick, Gibson & Paul, 1989), assesses five facets of

job satisfaction. The scale contains 72 items assessing nature of work, supervision,

pay, co-workers, and opportunity for promotion. This scale is reliable and

convergent and discriminant validity has been demonstrated (Gillet & Schwab, 1975;

Johnson, Smith & Tucker, 1982). This scale has been criticised however, as the

items have not been revised since the scale was developed.

In response to this criticism, Roznowski (1989) developed a revised scale by

calculating the discriminating power of the existing items, as well as some new

items. This revised scale had higher reliability with the alpha coefficient ranging

from 0.81 to 0.91. Although this revised scale may be more relevant to today’s

workforce, it still contains 72 items. To reduce the number of items for the current

study, a further revision was made. Only three items were selected to measure each

facet (refer to Appendix D). These items were selected as they had the highest

discrimination power.

This facet measure of job satisfaction is useful to diagnose the strengths and

weaknesses of organisations, however it cannot be summed to produce an overall

110

measure of job satisfaction (Ironson et al., 1989). Many researchers continue to use

facet scales to obtain an overall measure of job satisfaction (e.g., O’Driscoll &

Beehr, 2000; Schappe, 1998), however facet scales have been criticised as they may

exclude areas that are important to the respondent, or include areas that are

unimportant to the respondent. Therefore, in addition to the facet measure, a global

item of job satisfaction was also used.

The global measure of job satisfaction is a one-item measure. The item is

“taking into consideration all the things about your job, how satisfied are you with

it?” This item was rated on a 10-point scale ranging from 1 (completely dissatisfied)

to 10 (completely satisfied). This global scale requires the respondent to combine

their reactions to various aspects of the job into a single response. When answering

this question, the respondent may incorporate aspects of their job not included in the

facet scale. Although internal consistency cannot be established, a meta-analysis of

single-item measures of job satisfaction has demonstrated that single-item measures

correlate with other measures, such as the Job Diagnostic Survey (Hackman &

Oldham, 1976), the Job in General Scale (Ironson et al., 1989), and the Minnesota

Satisfaction Questionnaire (Weiss et al., 1967). On average, the correlation between

other scales and single item scales was r = 0.63 (Wanous, Reichers & Hudy, 1997).

2.4.2.4 Life Satisfaction

The subjective dimension of the Comprehensive Quality of Life Scale (Com-

QOL) developed by Cummins (1997) assesses satisfaction with seven domains of

life, including material well-being, health, productivity, intimacy, safety, community,

111

and emotional well-being (refer to Appendix E). An 11-point scale was utilised,

ranging from 0 (completely dissatisfied) to 10 (completely satisfied). The scale is

psychometrically sound, with internal reliability ranging from 0.5 to 0.8 (Cummins,

1997) and validity has been established using data from a review of the QOL

domains (Cummins, 1996).

2.4.2.5 Personality

The neuroticism and extroversion subscales of the NEO Five Factor

Inventory (short form; Costa & McCrae, 1992) was used to measure personality.

This scale contains 12 items to measure extroversion and 12 items to measure

neuroticism (refer to Appendix F). Six facet scales are measured in each factor.

Neuroticism is the sum of scales measuring anxiety, angry hostility, depression, self-

consciousness, impulsiveness, and vulnerability. Extroversion is the sum of warmth,

gregariousness, assertiveness, excitement-seeking, and positive emotions.

Convergent and discriminant validity of both of these factors has been established

(Costa & McCrae, 1992, Leong & Dollinger, 1991; Tinsley, 1994).

2.4.3 Procedure

Ethics approval was obtained from Deakin University, and consent was

obtained from the Heads of School for the academics, and the Human Resource

Managers for the supermarket workers. The recruitment procedure differed

depending on the group. The academics were sent the questionnaire via the internal

mail system. They returned the questionnaire by post. The supermarket workers

112

questionnaires were left in the staff room. As they were expected to complete the

questionnaire outside of work time, a $5 lottery ticket was given to each participant

to thank them for their time. The lottery tickets were given to the managers of the

store. The supermarket workers that returned the questionnaire collected their lottery

ticket from the service desk. At the conclusion of the study, the participating Heads

of School and the Human Resource Managers received a summary of the aggregated

results for all academics and supermarket workers.

113

2.5 Results

2.5.1 Data Screening and Checking of Assumptions

Procedures for data screening, and checking the procedural analytic

assumptions for all dependent variables followed those appropriate for group data.

The data set was initially examined for missing values, acquiescence, outliers,

normality and linearity. In regard to missing values, less than 4% of the values for

academics, and less than 5% of the values for supermarket workers were missing for

any one item. As there was no pattern to these missing values, they were replaced

with the group mean. Although this reduces the variance of the variables and

bivariate correlations (Tabachnick & Fidell, 1996), the replacement is a conservative

estimate.

Once the missing values were replaced, the data set was examined for

participants consistently reporting extreme scores (i.e., 1 or 10), in an attempt to

reduce the influence of acquiescence. One participant was omitted from the entire

sample for consistently reporting extreme scores on every scale. Other participants

reporting extreme scores on just one scale were excluded from that particular

analysis. Specifically, seven participants’ (all supermarket workers) responses were

deleted from the life satisfaction analyses, and nine participants’ responses were

omitted from the primary control analyses (three academics, six supermarket

workers).

Univariate outliers were identified on the facet job satisfaction scale, the life

satisfaction scale, and the control scales. Specifically, five cases of job satisfaction,

114

12 cases of life satisfaction, three cases of primary control, and nine cases of job

autonomy, lay outside three standard deviations from the mean. As these cases are

from the intended population, yet have more extreme values than the normal

distribution, they were recoded to three standard deviations from the mean.

On completion of the screening process, normality was assessed using the

skew/standard error <3, Kolmogorov-Smirnof values, frequency histograms, and

normal probability plots. In the academic group, overall life satisfaction (-3.91) and

the one-item measure of job satisfaction (-5.60) were mildly negatively skewed. In

the supermarket workers group, primary control was negatively skewed (-3.27). As

transformations are not recommended for data that are mildly and naturally skewed

(Tabachnick & Fidell, 1996), these data were not transformed. Finally,

homoscedasticity and linearity were assessed through bivariate scatterplots and these

appeared to demonstrate reasonable linear relationships between the variables.

115

2.5.2 Descriptive Statistics and Inter-Correlations

Table 3 contains the means and standard deviations for the major variables in

the study for each occupational group. In this table, and the tables thereafter, all

mean scores are converted to a percentage of scale maximum (%SM) which ranges

from 0-100. The formula is:

%SM = (mean score for the original domain – 1) x 100/ (number of scale points – 1).

Table 3 demonstrates that the academics report slightly higher job

satisfaction, primary control and job autonomy, and lower secondary control than the

supermarket workers. Table 4 displays the correlations among all of the major

variables for the academics and the supermarket workers. This demonstrates that job

autonomy and primary control are positively related to job satisfaction for both

occupational groups.

Table 3 - Means and Standard Deviations of Major Variables for Academics

and Supermarket Workers

Variable Academics Supermarket WorkersM SD M SD

Job Satisfaction - 1 item 66.05 21.09 59.71 25.69Job autonomy 51.94 14.63 34.50 20.24Primary Control 71.56 11.95 67.06 18.62Secondary Control 36.63 15.64 46.74 19.77Life Satisfaction 78.22 10.96 73.30 15.97Neuroticism 36.60 15.78 39.25 17.88Extroversion 61.71 12.41 65.28 13.99

116

Table 4 - Inter-Correlations for the Academics and the Supermarket Workers

JS JA PC SC LS Neu Ext

JS 0.25* 0.38** 0.14 0.07 -0.23* 0.17JA 0.41** 0.43** 0.03 0.06 -0.16 0.17PC 0.44** 0.43** 0.07 -0.02 -0.03 0.42**SC 0.04 0.08 -0.02 -0.07 0.04 0.02LS 0.20* 0.15 0.11 -0.01 -0.50** 0.29**Neu -0.27 -0.19* -0.14 0.05 -0.59** -0.30**Ext 0.09 0.09 0.25** -0.12 0.25** -0.17*

* p<0.05 , ** p>0.01; Correlations for supermarket workers are bolded.

JS = Job satisfaction; JA = Job autonomy; PC = Primary control; SC = Secondary control; LS = Life satisfaction; Neu = Neuroticism; Ext = Extroversion

117

2.5.3 Factor Analyses

Prior to testing the hypotheses, factor analyses were conducted on the revised

scales of job satisfaction, primary and secondary control, and job autonomy.

2.5.4 Factor Analysis of the Job Descriptive Index

To ensure the 15 job satisfaction items represented each of the five facets, a

principle components analysis with direct oblimin rotation was conducted. The

assumptions of sample size, normality, outliers, linearity, and the factorability of the

correlation matrix were initially examined.

Factor analysis requires a minimum of five subjects per variable (5 x 15 = 75)

(Coakes & Steed, 1999), hence the sample size of 199 was adequate. Some of the

job satisfaction items were not normally distributed however the solution is still

worthwhile if normality is not met (Tabachnick & Fidell, 1996). Five outlying cases

were recoded to three standard deviations from the mean. Reasonably linear

relationships existed among the variables. In regard to the factorability of the

correlation matrix, all of the correlations exceeded 0.30. The measures of sampling

adequacy (MSA) were > 0.50. Bartlett’s test of Sphericity was large and significant

(1574.89), and Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy

exceeded 0.60.

A principal components analysis, with direct oblimin rotation, yielded four

eigenvalues over 1. With this four factor solution 22% of the nonredundant residuals

had absolute values > 0.05, suggesting the presence of another factor. When a

principal components analysis with direct oblimin rotation was conducted with five

118

factors, only 7% of the nonredundant residuals had absolute values > 0.05. As such,

a five-factor model was deemed to be most appropriate. The loadings of the items on

each of the five factors are presented in Table 5. These loadings demonstrate that

Factor 1 refers to promotion, Factor 2 to nature of work, Factor 3 to pay, Factor 4 to

supervisors, and Factor 5 to co-workers.

Table 5- Factor Analysis of Job Satisfaction Scale

Item F1 F2 F3 F4 F5There is a good chance for promotion inmy job.

0.94

There is a fairly good chance for promotion in my job.

0.94

There are good opportunities for advancement in my job.

0.90

My work is dull. 0.97My work is boring. 0.90My work is interesting. 0.81My pay is bad. 0.91I am well-paid. 0.82My pay is unfair. 0.81My supervisors know how to supervise. 0.81My supervisors are bad. 0.79My supervisors are annoying. 0.78My co-workers are stupid. 0.79My co-workers are responsible. 0.73My co-workers are a waste of time. 0.79

Eigenvalues 4.32 2.78 1.87 1.65 0.94% of variance 28.77 18.56 12.46 10.99 6.27Cumulative % 28.77 47.32 59.79 70.77 77.04

Cronbach's Alpha (total scale) 0.82

Items with loadings less than 0.30 are not shown.

2.5.5 Factor Analysis of the Primary and Secondary Control Scale

To ensure that the primary and secondary control items loaded on two

separate factors, a factor analysis was conducted on the primary and secondary

119

control scale. The sample size was adequate (N = 190), and the secondary control

items were normally distributed for both groups. The primary control items were

mildly negatively skewed for both the academics (pc4 = -4.41, pc5 = -3.17) and the

supermarket workers (pc1 = -3.5, pc2 = -3.77, pc4 = -4.26, pc5 = -3.17; refer to

Table 7 for items). However, as factor analysis is robust to violations of normality,

the resulting solution was still deemed to be worthwhile (Tabachnick & Fidell,

1996). Linearity among the variables as assessed through scatterplots was

reasonable. The correlation matrix was factorable with all correlations exceeding

0.30. The measures of sampling adequacy exceeded 0.50 for all variables. Bartlett’s

test of sphericity was large and significant (1429.34), and Kaiser-Meyer-Olkin

(KMO) Measure of Sampling Adequacy exceeded 0.60.

A principal components analysis with direct oblimin rotation yielded 5

factors. The total variance explained by these five factors is demonstrated in Table

6.

120

Table 6- Total Variance Explained by a Five-Factor Solution

Initial Eigenvalues

Factor Total % of variance Cumulative %1 5.125 26.975 26.9712 2.846 14.980 41.9563 1.615 8.498 50.4544 1.476 7.766 64.1205 1.121 5.901 69.269

This five-factor solution demonstrated that four of the five primary control items

loaded on one factor, and that the rest of the secondary control items loaded on the

other four factors. However, as there was no clear pattern in the other four factors, a

four-factor and three-factor solution were also requested. In both of these analyses

however, many of the items loaded on more than one factor.

To investigate the hypothesised two-factor solution, a principal components

analysis with direct oblimin rotation was requested. More than two factors are

present however, as 67% of the nonredundant residuals had absolute values > 0.05.

As demonstrated in Table 7, all of the primary control items loaded on Factor 2.

Seven of the 14 secondary control items loaded on Factor 1, and the remaining seven

secondary control items loaded only on Factor 2 or on both factors. As such, in

subsequent analyses, the scale will include all five primary control items and only the

seven non-complex secondary control items. With only these items, a factor analysis

reveals that the primary control factor accounts for 19.84% of the variance, and the

secondary control factor accounts for 28.30% of the variance.

121

Table 7- Factor Analysis of Primary and Secondary Control Scale

No. Item F1 F2pc1 When I have a goal at work that is difficult to reach, I

think about different ways to achieve it.0.70

pc2 When I want something at work to change, I think Ican make it happen.

0.39

pc3 When a work task really matters to me, I think aboutit a lot.

0.51

pc4 When I really want to reach a goal at work, I believe Ican achieve it.

0.71

pc5 When faced with a difficult work situation, I believe Ican overcome it.

0.51

sc1* I can see that something good will come of it. 0.32 0.65sc2 I remember you can't always get what you want. 0.32 0.45sc3 I know things will work out okay in the end. 0.44 0.67sc4 I remember I am better off than many other people. 0.36 0.51sc5 I remember I have already accomplished a lot in life. 0.53 0.39sc6 I remember the success of my family or friends. 0.55 0.34sc7 I think nice thoughts to take my mind of it. 0.74sc8 I remind myself the situation will change if I am just

patient.0.65 0.38

sc9 I tell myself it doesn’t matter. 0.72sc10 I think about my success in other areas. 0.74sc11 I don’t feel disappointed because I knew it might

happen.0.61

sc12 I can see it was not my fault. 0.40sc13 I ignore it by thinking about other things. 0.69sc14 I realise I didn’t need to control it anyway. 0.61

Eigenvalues 5.13 2.85 % of variance 14.98 26.98 Cumulative variance 14.98 41.96 Cronbach's Alpha (for revised scale) 0.82 0.70

Items with loadings less than 0.30 are not shown.pc= primary control; sc=secondary control; Bolded items are included in the scale. *All secondary control items preceded by “When something bad happens that I cannot change”

122

2.5.6 Factor Analysis of the Job Autonomy Scale

To ensure the items on the job autonomy scale were measuring a single

construct, a factor analysis was conducted. The assumption of normality was

violated with items 1, 2, 4, and 10 being mildly negatively skewed for academics.

Items 7, 8, 9, and 11 were mildly positively skewed for the supermarket workers. As

before, these variables were not transformed. Nine univariate outliers were recoded

to three standard deviations from the mean. All correlations exceeded 0.30 and all of

the variable MSA exceeded 0.50. Barlett's test of sphericity was significant

(1077.97), and Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy exceeded

0.60.

A principal components analysis with direct oblimin rotation demonstrated

that the job autonomy items loaded on three factors. There was no meaningful

pattern within these factors however, and all but three items loaded on more than one

factor. In an attempt to find a pattern among the items, a two-factor principle

components analysis with direct oblimin rotation was conducted. This analysis,

presented in Table 8, demonstrated that six items loaded only on the first factor, one

item loaded only on the second factor, and the remaining seven loaded on both

factors. Factor 1 contained items that were directly related to the nature of the work

(i.e., tasks, order of work, working pace), whereas the items that loaded on Factor 2

related to organisational structure (i.e., pay, evaluation). Although this factor

analysis demonstrates that two factors emerge, all items will be retained in this scale

as the overall measure of job autonomy should be based on the nature of the work

and the organisational structure.

123

Table 8 - Factor Analysis of Job Autonomy Scale

No Item F1 F21 In my job I can choose among a variety of tasks or

projects to do.0.76 0.33

2 In my job I can choose the order in which I do my work. 0.843 In my job I can choose how quickly I do my work. 0.714 In my job I can choose how I schedule my rest breaks. 0.715 In my job I can choose the physical conditions of my

workstation.0.56 0.46

6 In my job I can choose when I interact with others. 0.747 In my job I can choose the amount I earn. 0.668 In my job I can choose the number of times I am

interrupted at work.0.42 0.58

9 In my job I can choose how my work is evaluated. 0.7610 In my job I can choose the quality of my work. 0.5611 In my job I can choose the policies and procedures in my

work unit.0.54 0.59

12 In my job I can choose among a variety of methods tocomplete my work.

0.82 0.36

13 In my job I can choose how much work I get done. 0.67 0.4114 In general, how much are you able to influence work and

work-related matters.0.64 0.50

Eigenvalues 0.58 1.27 % of variance 41.37 9.04 Cumulative variance 41.37 50.41 Cronbach's Alpha (revised scale) 0.86


124

2.6 Hypothesis Testing

In order to test the proposed model of job satisfaction, multivariate analyses

of variance were conducted to investigate how the academics and the supermarket

workers differed in their levels of job autonomy, control strategies, and job

satisfaction. Multiple regression analyses were also conducted to predict job

satisfaction from job autonomy, control strategies, personality, and life satisfaction.

In order to test these hypotheses, 22 p-values must be computed, and as such,

familywise error rate must be considered. Familywise error rate is the probability of

making at least one Type I error in a set of analyses (Keppel, 1991). Increasing the

number of statistical tests can potentially increase the familywise error. The formula

for familywise error is FW= (alpha level) x (number of comparisons). In this study,

the familywise error rate is (0.05) x (22) = 1.1. One solution to reduce this

familywise error rate is adjust the alpha level using the Bonferroni test (Keppel,

1991). The desired alpha level (0.05) is divided by the number of tests (22), yielding

a recommended alpha level of 0.002. Although reducing the alpha level decreases

the probability of Type I errors, it also increases the probability of Type II errors

(Keppel, 1991). The solution therefore is to strike a balance between the two errors.

Thus, the alpha level will be reduced to 0.01.

2.6.1 Hypothesis One- Assumption Testing

In order to test the first part of hypothesis one, proposing that job autonomy

and job satisfaction are positively related, the correlation coefficients for each

occupational group were examined. Consistently, job autonomy was positively

125

related to job satisfaction for both the academics (r = 0.41) and the supermarket

workers (r = 0.25).

In order to test the second part of hypothesis one, proposing that the

academics would report higher job autonomy than the supermarket workers, an

analysis of variance was employed. The assumption of univariate homogeneity of

variance, assessed using Levene’s test was not met, F (1, 197) = 12.77, p = 0.00,

however as this assumption is of little concern when the sample sizes are similar

(Tabachnick & Fidell, 1997), the analysis proceeded with caution using an alpha

level of 0.01. The univariate test of significance demonstrated that, consistent with

hypothesis one, the academics (M = 51.94, SD = 14.63) reported significantly higher

job autonomy than the supermarket workers, (M = 34.50, SD = 20.24),

F (1, 197) = 49.51, p = 0.00.

2.6.2 Hypothesis Two- Occupational Differences in the Use of the Control

Strategies

In order to examine hypothesis two proposing that the academics will report

less secondary control and more primary control than the supermarket workers, a

multivariate analysis of variance was performed. The assumptions of normality,

linearity, multicollinearity, and homogeneity of variance-covariance were examined

for the variables.

All of the variables were normally distributed, and reasonably linear

relationships were evident. Two univariate outliers were recoded to three standard

126

deviations from the mean and there were no multivariate outliers. There was no

evidence of multicollinearity, as the determinant of the within-cell correlation

was > 0.0001 (i.e., 0.798).

The assumption of univariate homogeneity of variance, as assessed by

Levene’s test, was met for secondary control, F (1, 188) = 2.84, p > 0.05. Equality of

error variance was not found however for primary control, F (1, 188) = 17.07,

p < 0.05. Levene’s test is sensitive to non-normality however, and this can lead to

overly conservative rejection (Tabachnick & Fidell, 1997). As such, the analyses

will proceed with caution using an alpha level of 0.01. The assumption of

multivariate homogeneity of variance-covariance, assessed through Box’s M test was

also violated. Box’s M test is a notoriously sensitive test of homogeneity of

variance-covariance, and it is recommended that if the test is violated, the

multivariate tests be examined by Pillai’s criterion rather than Wilk’s lamba.

The multivariate test of significance, using Pillai’s criterion, demonstrated

that occupational differences existed, F (2, 187) = 10.03, p = 0.00. As demonstrated

in Table 9, the supermarket workers reported significantly higher secondary control

than the academics, F (1, 188) = 15.50, p = 0.00. The two groups did not report

significantly different levels of primary control, F (1, 188) = 3.99, p = 0.04. It must

be noted however that the difference in primary control was significant at 0.05, but

not at the more stringent alpha level of 0.01. Hence, only partial support was

provided for the second hypothesis.

127

Table 9 - Means and Standard Deviations of Control Measures for Academics


Variable Academic SupermarketM SD M SD

Primary Control 71.56 11.95 67.06 18.62Secondary Control 36.63 15.64 46.74 19.77

Bolded constructs demonstrate significant occupational differences.

2.6.2.1 Summary

Multivariate analyses of variance demonstrated that the academics report

higher job autonomy, and lower secondary control than the supermarket workers.

The two groups did not report significantly different levels of primary control.

2.6.3 Hypothesis Three- Examining how Job Autonomy Relates to the

Control Strategies

To examine hypothesis three, proposing that job autonomy will be positively

related to primary control and negatively related to secondary control, the correlation

coefficients were examined. It was necessary to examine whether job autonomy

influences the control strategies using the measured level of job autonomy because

there was some variability in the level of job autonomy reported within occupational

groups. A median split was conducted on job autonomy and the employees were

split into two groups. The majority of academics were in the high job autonomy

group (66%), however 34% were in the low job autonomy group. Similarly, 70% of

128

the supermarket workers were in the low job autonomy group, however 30% were in

the high job autonomy group.

Job autonomy was positively related to primary control (r = 0.46), but not

related to secondary control (r = - 0.18). These results provide partial support for

hypothesis three, suggesting that job autonomy influences primary, but not secondary

control.

2.6.4 Hypothesis Four- Examining how Job Autonomy Influences the

Adaptiveness of the Control Strategies

To examine hypothesis four, proposing that i) primary control will be more

positively related to job satisfaction than secondary control for the academics, and ii)

secondary control will be more positively related to job satisfaction than primary

control for the supermarket workers, a standard multiple regression analysis was

conducted on each occupational group. The assumptions of normality, linearity and

homoscedasticity of residuals were assessed through examination of the residual

scatterplots. These assumptions were met, and there was no evidence of

multicollinearity.

As demonstrated in Table 10, R was significantly different from zero for both

the academics, R = 0.44, F (2, 102) = 12.53, p = 0.00, and the supermarket workers,

R = 0.31, F (2, 82) = 4.37, p = 0.01. Primary control predicted job satisfaction for

both groups, accounting for 20% of the variance in job satisfaction for the academics

and almost 10% for the supermarket workers. Secondary control did not predict job

satisfaction for either group. Hence, consistent with hypothesis four, primary control

129

was more positively related to job satisfaction than secondary control for the

academics. Inconsistently however, secondary control was not related to job

satisfaction for the supermarket workers.

Table 10 - Multiple Regression of Primary and Secondary Control on Job

Satisfaction for Academics and Supermarket Workers

Group Variable JS PC B sr2 (unique)Acad

PC 0.44 0.78 0.44 19.62**SC 0.04 -0.02 0.006 0.05

R =0.44** R2=0.20 Adj R2=0.18Super

PC 0.31 0.43 0.31 9.61**SC -0.02 0.08 -0.05 -0.04

R =0.31* R2=0.10 Adj R2=0.07

**p<0.01; Acad – Academics; Super – Supermarket workers; JS – Job satisfaction; PC – Primary control; SC – Secondary control

For academics, R2 composed of shared variance (1.9%) and unique variance (98.1%)For supermarket workers, R2 composed of shared variance (3.9%) and unique variance (96.1%).

2.6.5 Hypothesis Five- Does Job Autonomy Moderate the Relationship

Between the Control Strategies and Job Satisfaction?

In order to examine hypothesis five, proposing that the relationship between

the control strategies and job satisfaction is moderated by job autonomy, two

hierarchical multiple regression analyses were conducted. Job autonomy is proposed

to be a moderator, which means that it affects the direction and/or the strength of the

relationship between the control strategies and job satisfaction. Specifically, the

130

Low Job Autonomy

High Job Autonomy

relationship between primary control and job satisfaction is expected to be positive

when job autonomy is high and negative when job autonomy is low. Furthermore,

the relationship between secondary control and job satisfaction is expected to be

positive when job autonomy is low, and negative when job autonomy is high. These

expected relationships are demonstrated below in Figure 3.

Figure 3- Expected Moderated Effect of Job Autonomy on a) Primary Control

and Job Satisfaction and b) Secondary Control and Job Satisfaction

a)

Primary Control

b)

Secondary Control

A moderation effect can be tested in a number of ways depending on whether

the variables are continuous or discrete (Baron & Kenny, 1986). In this hypothesis,

the moderator variable and the independent variable are both continuous. When both

High Job Autonomy

JS

Low Job Autonomy

JS

131

variables are continuous, and when the effect of the independent variable on the

dependent variable varies linearly with respect to the moderator, a hierarchical

multiple regression analysis is conducted to test the presumed relationship (Baron &

Kenny, 1986). As demonstrated in Figure 4, the dependent variable is regressed on

the independent variable, the moderator variable, and the product of the independent

variable and the moderator (Baron & Kenny, 1986). Moderator effects are

demonstrated if the interaction term is significant when the independent variable and

the moderator variables are controlled (Baron & Kenny, 1986).

132

Primary control

Job autonomy

Primary control x Job autonomy

Job Satisfaction

1.1.1.1 Job

Secondary control

Job autonomy

Secondary control x Job autonomy

Job Satisfaction

Figure 4- Job autonomy Moderates the Relationship between a) Primary

control and b) Secondary Control, and Job Satisfaction.

Order of Variable Entry

a)

Step 1

Step 2

Step 3

b)

Step 1

Step 2

Step 3

In order to test the moderating effect of job autonomy on primary control and

secondary control, two hierarchical multiple regression analyses were conducted on

the combined sample. By using the combined sample, there was more range in the

levels of job autonomy. In these analyses the control strategies were entered first,

then job autonomy, and then the interaction term. For primary control, R was

significantly different from zero after the first step (i.e., primary control), R = 0.37,

F (1, 188) = 30.34, p = 0.00, and the second step (i.e., job autonomy), R = 0.40,

Finc (1, 187) = 4.89, p = 0.03. However, the addition of the interaction term was not

significant, R =0.40, Finc (1,186)= 0.33, p = 0.57.

133

For secondary control, R was not significantly different from zero after the

first step, R = 0.01, F (1, 197) = 0.04, p = 0.85. After job autonomy was entered, the

value of R increased, R = 0.33, F (1, 196) = 24.01, p = 0.00. There was no further

increase however when the interaction term was entered in step three, R = 0.34,

F (1, 195) = 1.58, p = 0.21. These analyses, displayed in Table 11, demonstrate that

inconsistent with hypothesis five, job autonomy did not moderate the relationship

between the control strategies and job satisfaction.

134

Table 11- Moderating Role of Primary and Secondary Control on the

Relationship Between Job Autonomy and Job Satisfaction

Step IV DV B sr2 (unique)

1 Primary control JS 0.57 0.37 13.91**2 Primary control 0.46 0.30 6.96**

Job Autonomy 0.20 0.17 2.19*3 Primary control 0.55 0.36 3.46**

Job autonomy 0.41 0.34Primary control x job autonomy

-0.003 -0.21

R =0.40 R2=0.16 AdjR2=0.15

1 Secondary control JS 0.02 0.012 Secondary control 0.02 0.02

Job Autonomy 0.40 0.33 10.89**

3 Secondary control 0.26 0.21Job autonomy 0.63 0.52 4.54*Secondary control x job autonomy

-0.005 -0.28

R =0.34 R2=0.12 AdjR2=0.10

**p<0.01, *p<0.05; JS – Job satisfaction

2.6.6 Hypothesis Six- Do the Control Strategies Mediate the Relationship

Between Job Autonomy and Job Satisfaction?

Hypothesis six proposes that the relationship between job autonomy and job

satisfaction is mediated by the control strategies. In this hypothesis, the control

strategies are acting as mediators because they are explaining why job autonomy is

related to job satisfaction. That is, employees with high job autonomy are expected

to rely on more primary control and less secondary control than employees with low

job autonomy. As primary control strategies are more positively related to job

135

satisfaction than secondary control strategies, employees with higher job autonomy

report higher job satisfaction.

It must be noted that although secondary control strategies are less positively

related to job satisfaction than primary control, it is proposed that for workers with

low job autonomy, secondary control strategies are superior to primary control

strategies. If these workers use primary control, they are expected to experience


According to Baron and Kenny (1986), in order to establish mediation, three

standard regression analyses must demonstrate that: a) job autonomy predicts

primary and secondary control; b) primary and secondary control and job autonomy

together predict job satisfaction; and c) job autonomy predicts job satisfaction. For a

mediation effect to be significant, all three regression equations must be significant,

and the effect of the independent variable on the dependent variable must be less in

b) than in c) (Baron & Kenny, 1986). This mediation analysis is demonstrated in

Figure 5.

Figure 5- Mediating Role of Control Strategies on the Relationship Between Job

Autonomy and Job Satisfaction

a b

c

Primary/Secondary Control

Job autonomy Job Satisfaction

136

This method will not be used however as there is an easier way to test the

mediating role of the control strategies. Rather than conducting three regression

analyses, only one hierarchical regression analysis is needed (M. Stokes, personal

communication, August 16, 2002). In this analysis, primary and secondary control

strategies are entered first, followed by job autonomy. It is expected that once

primary and secondary control strategies have been entered, there would be no

relationship between job autonomy and job satisfaction. As such, primary and

secondary control would explain the relationship between job autonomy and job

satisfaction.

The assumptions of normality, linearity and homoscedasticity of residuals

were met, and there was no evidence of multicollinearity. As demonstrated in Table

12, R was significantly different from zero after primary and secondary control were

entered, R = 0.37, F (2, 187) = 15.23, p = 0.00. Primary control accounted for 13%

of the variance in job satisfaction, and secondary control was not significant. R did

not significantly increase after job autonomy was added to the equation, R = 0.40,

Finc (3, 186) = 11.92, p = 0.03. Even if the less stringent alpha level of 0.05 was

used, job autonomy only accounts for 2% of the unique variance in job satisfaction.

As such, it appears that when primary and secondary control are entered first, there is

no relationship between job autonomy and job satisfaction. This suggests that partial

support is provided for hypothesis six as primary control, but not secondary control,

mediates the relationship between job autonomy and job satisfaction.

137

Table 12 -Hierarchical Multiple Regression Testing the Mediating Role of the

Control Strategies

Step IV DV B sr2 (unique)1 Primary Control JS 0.57 0.37 13.91**

Secondary Control -0.04 -0.03

R =0.37** R2=0.14 AdjR2=0.13

2 Primary Control JS 0.46 0.30 6.96**Secondary Control -0.02 -0.02Job Autonomy 0.20 0.17 2.01*

R =0.40** R2=0.16 AdjR2=0.15

**p>0.01 *p>0.05; JS - Job satisfaction

It must be noted that although the results demonstrate that primary control is

a partial mediator of the relationship between job autonomy and job satisfaction, the

use of multiple regression to estimate a mediational model is based on the

assumption that there is no measurement error in the mediator. This assumption is

particularly concerning as the mediator is likely to be measured with error. The

presence of such error tends to produce “an underestimation of the effects of the

mediator, and an overestimation of the effects of the independent variable on the

dependent variable” (Baron & Kenny, 1986, p. 1177).

One statistical method that models the measurement error is structural

equation modeling. Structural equation modeling is based on the analysis of sample

variances and covariances rather than individual cases. This approach is particularly

useful for latent variables, which are hypothetical constructs that cannot be directly

measured, such as job satisfaction.

138

Although structural equation modelling has some advantages over multiple

regression, it will not be used in this thesis for a number of reasons. First, unlike

hierarchical multiple regression, structural equation modelling is a confirmatory

technique. The current study, although grounded in theory, is exploratory,

attempting to combine the propositions of the job demand-control model (Karasek &

Theorell, 1990) with propositions of the life span theory of control (Heckhausen &

Schulz, 1995). As the theory is exploratory, there are a variety of different models

that can be examined. If numerous modifications of a model were tested, the

analysis would be exploratory, and there would be an increased risk of Type I errors

(Ullman, 1996). As this thesis is attempting to develop and explore the proposed

model of job satisfaction and search for unexpected relationships, structural equation

modelling may be problematic. Once the model is more established however,

structural equation modelling may be required.

A further problem with using structural equation modelling is that it requires

large sample sizes. The issue of an adequate sample size continues to be debated,

however Boomsma (1983) suggested that as a general rule, samples of 200 are

required to give parameter estimates with any degree of confidence. As the

relationship between the variables is expected to be different for academics and

supermarket workers, two models would need to be conducted, thus there would

need to be 200 in each occupational group.

A sample size of 200 is problematic due to time constraints, but also because

of the particular workers selected for this study. The data collection process

undertaken in study one demonstrated that workplaces, particularly those employing

139

low autonomy workers, such as call centres, factories, or supermarkets, were

reluctant to become involved in any research. The employers refused to participate

in the surveys for a variety of reasons. Some mentioned that the majority of their

employees were from Non-English speaking backgrounds and as such would be

unable to understand the survey. Others admitted that work motivation was very

low, and as such, the response rate would be poor. Still others were concerned that

the employees would expect changes to be made to the workplace on the basis of

their responses. These employers’ reactions indicate that is difficult to obtain a

sample size of 400.

2.6.6.1 Summary

In summary, it appears that primary control mediates the relationship between

job autonomy and job satisfaction. This finding was based on multiple regression

analyses however, which assumes that there is no measurement error in the mediator.

Although this measurement error can be accounted for in structural equation

modeling, it is concluded that such a method is not appropriate whilst the proposed

model of job satisfaction is in an exploratory stage.

2.6.7 Hypothesis Seven- Occupational Differences in Job and Life

Satisfaction

Hypothesis seven proposes that the academics will report higher job

satisfaction and higher life satisfaction than the supermarket workers. A univariate

analysis of variance was conducted on the global one-item measure of job

140

satisfaction. The assumptions of normality, linearity, and homogeneity of variance

were examined. Job satisfaction was normally distributed for the supermarket

workers, however it was negatively skewed for the academics (-5.60). The

assumption of homogeneity of variance was violated, F (1, 197) = 6.06, p < 0.05, and

as such, the analysis will proceed with caution using an alpha level of 0.01.

Inconsistent with hypothesis seven, there were no occupational differences in the

one-item measure of job satisfaction, F (1, 197) = 3.66, p = 0.04. The levels of job

satisfaction reported by the two groups are provided in Table 13.

To examine whether the two groups differed on the facets of job satisfaction,

a multivariate analysis of variance was conducted on the five facets of job

satisfaction, namely nature of work, co-workers, pay, supervisors, and opportunities

for promotion. Normality was assessed using skew/standard error < 3,

Kolmogorov-Smirnof values, and normal probability plots. Although the nature of

work facet (-3.49) and the co-workers facet (-4.87) were negatively skewed for the

academic group, the remainder of the variables were normally distributed for both

groups. Five univariate outliers were recoded to three standard deviations from the

mean, and no multivariate outliers were identified. Examination of bivariate

scatterplots, and correlations revealed reasonably linear relationships. There was no

evidence of multicollinearity as the determinant of the within-cell correlation

was >0.0001. Univariate homogeneity of variance, assessed by Levene’s test,

demonstrated that equality of error variance was evident for the supervision facet,

F (1, 197) = 0.40, p > 0.05. Equality of error variance was not found however for

pay, F (1, 197) = 5.53, p < 0.05, nature of work, F (1, 197) = 27.14, p < 0.05,

141

co-workers, F (1, 197) = 5.82, p < 0.05, and promotion, F (1, 197) = 4.20, p < 0.05.

As such, the univariate tests will be examined with caution. The assumption of

multivariate homogeneity of variance-covariance, as assessed through Box’s M test,

was also violated, F (15, 146782) = 4.90, p < 0.001.

The multivariate tests were examined using Pillai’s criterion. The job facets

were affected by occupation, F (5, 193) = 35.10, p = 0.00. As demonstrated in Table

13, academics reported significantly higher satisfaction with nature of work,

F (1, 197)= 95.59, p = 0.00, and co-workers, F (1, 197) = 32.51, p = 0.00, than

supermarket workers. However, the supermarket workers reported higher

opportunity for promotion than the academics, F (1, 197) = 9.21, p = 0.00.

Table 13- Means and Standard Deviations of Job Satisfaction Scale for

Academics and Supermarket Workers

Variable Academic SupermarketM SD M SD

Nature 85.97 13.74 56.98 26.95Co-Workers 82.30 15.75 67.85 20.01Pay 51.20 27.55 54.17 23.56Supervisors 60.29 24.47 67.03 24.49Promotion 38.31 27.40 51.12 32.18One-item measure 66.05 21.09 59.71 25.69

Bolded variables indicate significant occupational differences

It was expected that levels of job satisfaction would be related to levels of life

satisfaction, and that the academics reporting higher job satisfaction than the

supermarket workers would also report higher life satisfaction. To examine this

hypothesis, a univariate analysis of variance was conducted to examine overall life

142

satisfaction, and a multivariate analysis of variance was conducted to examine on

which domains the groups differed.

To compare their overall life satisfaction, a univariate analysis of variance

was conducted. Life satisfaction was normally distributed for the supermarket

workers however it was mildly negatively skewed for the academics (-3.94). The

assumption of homogeneity of variance, as assessed through Levene’s test of

equality of error variance was violated, F (1, 190) = 15.24, p < 0.05, and as such, the

analysis proceeded with caution using an alpha level of 0.01. Consistent with

hypothesis seven, the academics reported higher life satisfaction than the

supermarket workers, F (1, 190) = 6.38, p = 0.01.

A multivariate analysis of variance was conducted on the seven domains of

life satisfaction to examine where these differences lay. The assumptions of

normality, linearity, multicollinearity and homogeneity of variance-covariance were

examined for the seven domains. The emotional well-being domain was mildly

negatively skewed for the academics, and the intimacy domain was mildly negatively

skewed for the supermarket workers. 12 univariate outliers were recoded to three

standard deviations from the mean. Four multivariate outliers were examined and

recoded to the next less extreme score. The assumption of linearity, examined

through bivariate scatterplots, was met. Equality of error variance was demonstrated

only for satisfaction with health, F (1, 190) = 2.28, p > 0.05, and as such, the analysis

will proceed with caution. As the assumption of multivariate homogeneity of

variance, examined through Box’s M test was also violated, F (28, 110823) = 2.54,

p < 0.001, Pillai’s criterion was used to examine the multivariate test.

143

Pillai’s criterion was significant, F (7, 184) = 1039.56, p < 0.01. Academics

reported significantly higher productivity satisfaction, F (1, 190) = 7.63, p = 0.006,

and safety satisfaction, F (1, 190) = 12.62, p = 0.00, than supermarket workers. The

means and standard deviations for the satisfaction domains are provided in Table 14.

Table 14- Means and Standard Deviations of Life Satisfaction for Academics


Occupation Academic SupermarketM SD M SD

Material Satisfaction 80.56 13.18 78.56 19.43Health Satisfaction 75.03 20.07 69.16 23.85Productivity Satisfaction 76.00 13.91 69.05 21.67Intimacy Satisfaction 81.08 18.44 76.63 24.21Community satisfaction 75.71 16.22 70.07 20.13Safety Satisfaction 83.52 15.13 74.46 20.85Emotional Satisfaction 78.86 15.27 76.63 22.10OVERALL SATISFACTION 78.22 10.96 73.30 15.97

Bolded variables indicates occupational differences

2.6.8 Hypothesis Eight- Predictors of Job Satisfaction

In order to evaluate hypothesis eight, which proposes that primary control,

secondary control, job autonomy, personality and life satisfaction predict job

satisfaction, a multiple regression analysis was conducted on both occupational

groups. The correlations among the variables are displayed in Table 4.

For both groups, the assumptions of normality, linearity and homoscedasticity

of residuals were met, and there was no evidence of multicollinearity. R was

significantly different from zero after all the variables had been added for both the

academics, R = 0.54, F (6,98) = 6.69, p = 0.00, and the supermarket workers,

144

R =0.46, F (6, 73) = 3.25, p = 0.00. For the academics, the unique predictors of job

satisfaction were job autonomy and primary control. As demonstrated in Table 15,

primary control and job autonomy accounted for 4% and 8% of the variance in job

satisfaction respectively. It must be noted however that job autonomy was not

significant at the more stringent alpha level of 0.01. For the supermarket workers,

there was only one unique predictor of job satisfaction, namely primary control.

Primary control accounted for 8% of the variance in job satisfaction.

These results suggest that hypothesis eight is partially supported as primary

control and job autonomy predicted job satisfaction. However, secondary control,

personality and life satisfaction were not unique predictors of job satisfaction.

Furthermore, even when all the variables were included in the equation, R2 was small

(R2 = 0.29, R2 = 0.21).

145

Table 15- Multiple Regression of Job Autonomy, Control Strategies,

Personality, and Life Satisfaction for Academics and Supermarket Workers

Group IV DV B sr2 (unique)

Acad Job autonomy JS 0.34 0.24 4.41*Primary Control 0.58 0.33 8.12**Secondary Control 0.03 0.03Neuroticism -2.30 -0.16Extroversion -1.02 -0.05Life Satisfaction 0.10 0.05

R =0.54** R2=0.29 Adj2=0.25

Super Job autonomy JS 0.08 0.06Primary Control 0.46 0.36 8.35**Secondary Control 0.15 0.12Neuroticism -3.63 -0.25Extroversion -1.14 -0.06Life Satisfaction -0.04 -0.03

R =0.46** R2=0.21 Adj2=0.15

** p<0.01, * p<0.05; Acad – Academics; Super- Supermarket workers; JS – Job satisfaction

2.6.8.1 Summary

The academics reported significantly higher life satisfaction than the

supermarket workers, but similar levels of job satisfaction. The major predictors of

job satisfaction were job autonomy and primary control strategies.

2.6.9 Conclusion

The major propositions of this study were that job autonomy influences the

use of the control strategies and the relationship between the control strategies and

146

job satisfaction. As hypothesised, the academics reported higher job autonomy,

higher life satisfaction and lower secondary control than the supermarket workers.

Inconsistent with the hypotheses, the two groups reported similar levels of primary

control and job satisfaction. However, job autonomy was positively correlated with

primary control and not correlated with secondary control.

In regard to the proposal that job autonomy influences the relationship

between the control strategies and job satisfaction, the findings were less supportive.

Primary control was the most adaptive strategy for both groups, and secondary

control was not related to job satisfaction for either group. The implications of these

findings will now be discussed.

147

2.7 Discussion

This study tested a new explanation for the relationship between job

autonomy and job satisfaction, namely that job autonomy influences the use and

adaptiveness of primary and secondary control strategies. In regard to the use, the

findings demonstrated that the supermarket workers reported more secondary control

than the academics, but that only primary control was related to job autonomy. In

regard to the adaptiveness, primary control was the most adaptive strategy for

academics and supermarket workers. These findings are discussed in terms of the

life span theory of control (Heckhausen & Schulz, 1995) and the discrimination

model (Thompson et al., 1998). Before these propositions are explained in detail, the

basic assumptions of the study will be examined.

2.7.1 Assumption Testing

The basic assumptions of the study were that job autonomy was positively

related to job satisfaction, and that the study used two occupational groups that

differed in their level of job autonomy. Consistently, job autonomy was positively

related to job satisfaction for the academics and the supermarket workers

(r = 0.41, r = 0.25, respectively). These correlations are slightly lower than those

reported in other studies using Ganster’s (1989, cited in Dwyer & Ganster, 1991)

scale. For example, Munro, Rodwell and Harding (1998) demonstrated that the

correlation between job autonomy and job satisfaction was r = 0.69, whilst Fox et al.,

(1993) demonstrated that r = 0.46. It must be noted however that these studies relied

on the original version of the scale, which included items on predictability.

148

Consistent with the second part of hypothesis one, the academics’ levels of

job autonomy (M = 52%SM) were significantly higher than the supermarket workers

(M = 32%SM). For the purpose of this study, this difference in job autonomy should

be sufficient to examine the differences in primary and secondary control. It is

expected that the relationship between job autonomy and the control strategies is

linear, and that with increasing job autonomy, the use of primary control will

increase, and the use of secondary control will decrease. As such, even if the

difference between the group is not extremely large, the differences in the use of the

control strategies should still exist, however they may be less extreme.

In order to understand the meaning of these levels of job autonomy, it is

useful to compare them with other studies. Although normative data on Ganster’s

(1989, cited in Dwyer & Ganster, 1991) scale are not available, a few studies have

relied on this scale. They have shown that nurses scored 46%SM (Ganster et al.,

2001), and 57%SM (Munro et al., 1998). Furthermore, manufacturing employees

scored 57%SM (Dwyer & Ganster, 1991). It is difficult to make comparisons with

past studies however, as these studies have generally altered the scale in some way

(e.g., Ganster et al., 2001; Munro et al., 1998). Indeed the current study made an

important change to the scale, as the items on predictability were excluded.

2.7.2 Does Job Autonomy Influence the Use of the Control Strategies?

Partial support was provided for the second hypothesis as there was a

significant occupational difference in secondary control (M = 36%SM academics;

149

M = 46%SM supermarket workers), but not primary control (M = 71%SM

academics; M = 67%SM supermarket workers). Partial support was also provided

for hypothesis three as primary control was positively related to job autonomy

(r = 0.46), however secondary control was not correlated (r = -0.18). These findings

are somewhat inconsistent, with the former suggesting that job autonomy influences

secondary control but not primary control, and the latter suggesting that job

autonomy influences primary but not secondary control. More emphasis is placed on

hypothesis three as it is based on the measured level of job autonomy rather than the

assumed level. Thus, these findings demonstrate that as job autonomy increases,

primary control increases.

These findings appear to be inconsistent with Abouserie’s (1996) study on

academics’ coping strategies. In this study, academics were given a list of strategies

and required to indicate which ones they use to handle stress. The following coping

strategies emerged as the most common; acceptance of the problem (58%), talking

with others (57.7%), and trying to come to terms with each problem (55.8%).

Although the most common strategy, “acceptance of the problem” appears to

be a secondary control strategy, it is different to secondary control. Secondary

control is often referred to as acceptance however it is not acceptance that the

problem exists; it is acceptance that the problem cannot be overcome. Acceptance of

the problem may be interpreted as recognising that the problem exists, which is not

secondary control. Thus, although Abouserie’s (1996) results suggest that academics

mostly use secondary control, this may not be the case.

150

One study partially supports the findings from the current study. Narayanan,

Marian and Spector (1999) studied the coping strategies reported by academics, sales

employees, and clerical workers. They used an open-ended questionnaire where

participants were asked how they handled a stressful event at work. The academics

tended to handle their problems at work by taking direct action (24% of sample), and

talking to the chair of department (26%). The clerical workers and the sales

employees, on the other hand, reported that they talked with their co-workers (22%,

29%, respectively), or friends (24%, 29%, respectively).

Although Narayanan et al., (1999) did not measure job autonomy, their

findings demonstrate that the employees expected to have higher job autonomy

(i.e., academics) tended to rely on primary control-type strategies. The employees

expected to have lower job autonomy (i.e., sales employees, clerical workers) tended

to rely on secondary control-type strategies. These findings were partially consistent

with the current study.

The finding that job autonomy is positively related to primary control

provides some support for the proposed model of job satisfaction presented in Figure

2. This model, based on the job demand-control model (Karasek & Theorell, 1990)

and the life span theory of control (Heckhausen & Schulz, 1995), proposes that

employees with high job autonomy are more likely to successfully change the

environment using primary control. Thus, as job autonomy increases, primary

control increases.

However, the findings in the present study must be examined cautiously as a

limitation has been identified. The primary and secondary control scale required

151

respondents to indicate their agreement with each type of strategy, from 1 (do not

agree at all) to 10 (agree completely). It is now recognised that the only information

this scale provides is whether the respondents have ever used the strategies, and not

how often they are using the strategies. The current findings only demonstrate that

as job autonomy increases, employees’ agreement with the primary control strategies

increases, not the frequency.

2.7.3 Does Job Autonomy Influence the Relationship Between the Control

Strategies and Job Satisfaction?

In addition to testing whether job autonomy influences the use of the control

strategies, the current study also tested whether job autonomy influenced the

relationship between the control strategies and job satisfaction. Consistently,

primary control (r = 0.44) was more positively related to job satisfaction than

secondary control (r = 0.04) for the academics. However, secondary control

(r = 0.14) was not more positively related to job satisfaction than primary control

(r = 0.38) for the supermarket workers. Further analyses demonstrated that job

autonomy did not moderate the relationship between the control strategies and job

satisfaction. As such, it appears that primary control is more adaptive than secondary

control for all employees, whether they have low or high job autonomy.

These findings do not support the discrimination model (Thompson et al.,

1998) which proposes that primary control is the most adaptive strategy in

controllable situations, and that secondary control is the most adaptive strategy in

uncontrollable situations. Rather, these findings support the primacy/back-up model

152

(Thompson et al., 1998), which proposes that primary control strategies are more

adaptive than secondary control strategies for people in low-control or high-control

situations.

Although these results appear to support the primacy/back-up model, closer

examination of the Primary and Secondary Control Scale (Heeps et al., 2000)

reveals several limitations. The most notable is that some of the primary control

items examined whether the employees believed that they could change their

situation, rather than examining how they could change their situation. For example,

the items “I think I can make it happen”, “I believe I can achieve it” and “I believe I

can overcome it” measure whether a person believes that they can change a situation.

These general and non-specific thoughts were assessed rather than specific perceived

behaviours (i.e., work harder) because it was assumed that there could be an

unlimited number of specific behaviours.

However, it is now questioned whether believing that one can change a

situation is a measure of primary control. A person may report that they can change

a situation for a variety of reasons, not just if they use primary control strategies

when they face difficulties. For example, a person may report that they can change

their environment because they have high optimism. Alternatively, they may be

using the secondary control strategy “illusory optimism” where they tell themselves

that “everything will work out okay in the end.” These examples serve to illustrate

that people who believe that they can change their environment may not necessarily

use primary control.

153

To overcome these limitations, the measure of primary control may need to

be more specific. Rather than assessing whether people generally believe they can

change their environment, the primary control scale needs to assess how people

change their environment using primary control strategies. Thus the scale needs to

examine perceived strategies (e.g., exerting more effort, working harder) rather than

beliefs. This would make the scale consistent with the secondary control scale,

which assesses specific strategies.

In summary, although the findings suggest that the relationship between the

control strategies and job satisfaction is not influenced by job autonomy, several

problems have been identified in the primary and secondary control scale. The

primary and secondary control scale needs to be revised so that the primary control

items refer to perceived strategies rather than beliefs, and the rating scale needs to

assess frequency.

2.7.4 Do the Control Strategies Mediate the Relationship Between Job

Autonomy and Job Satisfaction?

Hypothesis six tested an alternative explanation to Karasek and Theorell’s

(1990) proposal for the relationship between job autonomy and job satisfaction. This

explanation, developed in chapter 1, proposes that employees with high job

autonomy rely on more primary control strategies which are positively related to job

satisfaction, whereas employees with low job autonomy rely on more secondary

control strategies which are less positively related to job satisfaction. It must be

noted however that although secondary control strategies are less positively related to

154

job satisfaction, it is proposed that for workers with low job autonomy, secondary

control strategies are superior to primary control strategies. If these workers use

primary control, they are expected to experience primary control failure.

The results demonstrated that primary control, but not secondary control,

mediated the relationship between job autonomy and job satisfaction. This provides

empirical evidence supporting one mechanism by which job autonomy may

influence job satisfaction. The importance of these findings must not be

overemphasised however, as problems have been identified with the primary and

secondary control scale. As such, the mediating role of primary and secondary

control needs to be re-examined using a revised scale.

2.7.4.1 Summary

The major aim of this study was to test an explanation for the relationship

between job autonomy and job satisfaction. The explanation proposes that job

autonomy influences the use and adaptiveness of the control strategies. The results

from the current study have offered some support for job autonomy influencing the

use of primary control strategies, but less support for job autonomy influencing the

adaptiveness of the control strategies. However, as there are some methodological

problems with the scale, the proposition requires further examination.

2.7.5 Examining Occupational Differences in Job Satisfaction

The differences in job autonomy and primary and secondary control were

expected to influence job satisfaction, where the academics were expected to report

155

higher job satisfaction than the supermarket workers. This proposal was not

supported however, as the academics reported similar levels of job satisfaction

(M = 66%SM) as the supermarket workers (M = 59%SM). In order to understand

these levels of job satisfaction, past studies will be examined.

2.7.5.1 Past Studies on Job Satisfaction

As few studies have examined academics’ or supermarket workers’ levels of

job satisfaction, and as there does not appear to be any consensus as to what is the

normative level of job satisfaction, a review was conducted. A range of studies

(N=36), which examined the levels of job satisfaction reported by different

occupational groups, were selected from psychology databases. These studies,

displayed in Appendix G, examine several occupational groups including nurses,

teachers, managers, manufacturing employees, and social workers. Although these

studies relied on several different scales, including global and facet scales, they were

reasonably consistent. The average level of job satisfaction was 66.75%SM, and the

scores ranged from 44.75%SM (Laschinger, Finegan & Shamain, 2001) to 87%SM

(Fisher, 2000). This average is similar to the academics and supermarket workers

levels of job satisfaction.

A few studies have specifically examined academics’ and supermarket

workers’ levels of job satisfaction. For academics, researchers have reported the

following levels of job satisfaction; 57%SM (Leung et al., 2000), 65%SM (Hill,

1986), 66%SM (Lahey & Vihtelic, 2000), 74% (Carson, Lanier & Carson, 2001),

82% (Olsen, 1993) and 83%SM (Niemann & Dovidio, 1998).

156

Although these scores vary widely, it must be recognised that these studies

have relied on different scales of job satisfaction. Some relied on facets scales of job

satisfaction (Hill, 1986; Lahey & Vihtelic, 2000) whilst others relied on global scales

on job satisfaction (Carson, Lanier & Carson, 2001; Leung et al., 2000; Niemann &

Dovidio, 1998; Olsen, 1993). However, the facet versus global distinction does not

necessarily explain the differing levels of job satisfaction, as facet and global scales

of job satisfaction have been shown to be moderately correlated (Wanous et al.,

1997). Rather within the facet and global scales, there is extensive variability that

may account for the inconsistent levels of job satisfaction.

There are differences among the facet scales of job satisfaction. For example,

Hill’s (1986) facet scale of job satisfaction measures satisfaction with several

dimensions including economic, teaching administrative, collegial, recognition-

support, and convenience. In contrast, Lahey and Vihtelic (2000) focussed on the

work itself, pay, recognition, co-workers, and supervision. The difference between

Hill’s (1986) facets and Lahey and Vihtelic (2000) facets may be important. Hill’s

(1986) facets were designed to be specific to academia, however it appears that they

are focussing on the areas that academics traditionally cite as a source of stress, such

as recognition, finances (Leung et al., 2000), and administration (Abouserie, 1996).

As such, the academics in Hill’s (1986) study may have a reported a lower level of

job satisfaction than those in Lahey and Vihtelic (2000) study because the scale was

focussed on the more negative aspects of the job.

There are also differences among the global scales of job satisfaction. For

example, Niemann and Dovidio (1998) relied on a 3-item measure of job

157

satisfaction, which included the following items; “I am satisfied with my job”, “I find

fulfillment in my work” and “I feel free to do the work that is important to me.” The

level of job satisfaction reported by the academics in this study may have been

higher because of the inclusion of the third item, which may be confounded with job

autonomy.

A more valid global measure of job satisfaction was used in Olsen’s (1993)

study. Measuring job satisfaction through one-item (i.e., “All things considered, how

satisfied are you with your position”), they found academics reported a high level of

job satisfaction (M = 82%SM). It must be noted however that this level of job

satisfaction was reported by academics in their first year of appointment.

Interestingly, they re-tested these academics at the end of their third year, and found

that their level of job satisfaction had declined to 71.66%SM. This lower level is

more consistent with other studies.

In summary, it is extremely difficult to produce an average level of job

satisfaction for academics. Only a few studies have examined academics job

satisfaction and these have tended to rely on different scales. The level of job

satisfaction found in the current study fits within the range found by past studies. It

must be noted however that this range is reasonably large.

In regard to supermarket workers, the only studies that can be compared with

the current findings are those conducted on retail workers. These studies have

generally reported a higher level of job satisfaction than that found for the

supermarket workers. For example, Doran, Stone, Brief and George’s (1991) study

demonstrated that retail workers given the Minnesota Satisfaction Questionnaire

158

(Weiss et al., 1967) reported a level of job satisfaction that was 73%SM.

Furthermore, Leung’s (1997) study on retail workers reported similar findings using

Hackman and Oldham’s (1975) scale (70% SM).

Although these studies report higher levels of job satisfaction, it must be

noted that the workers in these studies were obtained from a department store (Doran

et al., 1991) and a casual apparel store (Leung, 1997), and as such, may have more

job autonomy that the supermarket workers. The supermarket register operators are

required to work on the cash register for the majority of their shift whereas retail

assistants can often choose among different task to complete. Thus, it is difficult to

compare these studies with the current findings.

2.7.5.2 Explaining the Levels of Job Satisfaction Reported by the Academics

and the Supermarket Workers

The finding that such two distinct occupational groups report similar levels of

job satisfaction is surprising. However, there may be differences between the groups

that can account for this. First of all, the nature of the work is very different for these

two groups. The supermarket workers engage in repetitive work, and as such, they

may face few novel difficulties. The academics, on the other hand, are expected to

be involved in several complicated activities and face many varied difficulties. As

such, although the current study was proposing that supermarket workers would have

lower job satisfaction because they have less autonomy, they may also have fewer

difficulties to overcome.

159

Another difference between the two occupational groups that may explain

their similar levels of job satisfaction concerns their different investments and

expectations. Whereas the supermarket workers have invested little time into

training, the academics have invested at least seven years studying at university. The

number of years invested in training or education may be particularly important, as it

has been suggested that education is positively correlated with expectations (Clark &

Oswald, 1996).

For example, Clark’s (1996) study of British employees demonstrated that the

percentage of employees who reported that they were very satisfied with their job

was greatest for the group with the lowest education (M = 78%SM). The next

highest reported a level which was 74%SM, and the highest educated group reported

a level of job satisfaction that was 73%SM. Although the differences between these

groups are small, it is surprising that the group with the lowest level of education

would report a level of job satisfaction that equalled those with a higher education,

let alone surpassed it. As such, the academics may have higher job expectations than

the supermarket workers.

In summary, inconsistent with the hypotheses, the academics and teachers

reported similar levels of job satisfaction. This finding may be partly attributed to

the supermarket workers experiencing fewer difficulties than the academics, or the

academics having higher job expectations than the supermarket workers.

160

2.7.6 Examining Occupational Differences in Life Satisfaction

Consistent with hypothesis seven, the academics reported higher overall life

satisfaction (M=78.22) than the supermarket workers (M=73.30). The academics’

levels of life satisfaction were expected to be higher because job satisfaction was

expected to be positively related to life satisfaction. Although the academics did not

report higher job satisfaction than the supermarket workers, they did report higher

life satisfaction.

In regard to the normative levels of life satisfaction, Cummins’ (1995, 2000b)

homeostatic theory of life satisfaction proposes that the mean life satisfaction across

population samples lies within the 70-80%SM range. This is because people have a

“set-point range” for their life satisfaction. This set-point range is determined by

personality variables, namely neuroticism and extroversion. Together, these two

variables provide an affective balance, where the mid-point for the set-point range is,

on average, 75%SM. This affective balance influences the second-order buffers (i.e.,

optimism, self-esteem and control) so that, on average, their set-point is also

75%SM. These second-order buffers can however be influenced by the external

world. Hence, the mid-point for the set-point range can range between 70-80%SM.

Consistent with this prediction, both the academics’ (M = 78.22%SM) and the

supermarket workers’ levels of life satisfaction (M = 73.30%SM) lay within this

range.

The academics’ level of life satisfaction was at the higher end of the

normative range. According to the homeostatic theory of life satisfaction, the ceiling

for population sample means is approximately 80%SM (Cummins, 2000b). This

161

value represents the theoretical maximum for sample means grouped as data where

the distribution of set-ranges is normal, and each person has achieved the upper value

of their set-range. As such, the academics’ level of life satisfaction, in relative terms,

is extremely high.

The supermarket groups’ level of life satisfaction was at the lower end of the

normative range. Cummins (2000b) proposes that when life satisfaction falls

towards the 70%SM mark, homeostatic devices operate to prevent it from falling

further. When the sample mean approaches 70%SM however, the homeostatic

machinery is defeated for a significant proportion of the sample. As this happens,

the distribution collapses and the standard deviations increase. Consistent with this

prediction, the standard deviation of life satisfaction for the supermarket workers

group (SD = 15.97) was greater than for the academic group (SD = 10.96). Hence, a

greater proportion of the supermarket workers may be experiencing homeostatic

defeat. In summary, the academics reported higher life satisfaction than the

supermarket workers, however both means lay within the normative range.

2.7.7 Predicting Job Satisfaction from Job Autonomy, Control Strategies,

Personality, and Life Satisfaction

Partial support was provided for hypothesis eight, as job autonomy and

primary control predicted job satisfaction for the academics, and primary control

predicted job satisfaction for the supermarket workers. Inconsistent with the

proposed model of job satisfaction however, secondary control did not predict job

satisfaction for either group, and job autonomy did not predict job satisfaction for the

162

supermarket workers. The finding that secondary control did not predict job

satisfaction clearly needs to be re-examined as there are several methodological

problems with the secondary control scale. The finding that job autonomy did not

predict job satisfaction for the supermarket workers requires further examination.

The finding that job autonomy did not predict job satisfaction for the

supermarket workers may reflect problems with the job autonomy scale. The job

autonomy scale was a multidimensional scale. The scale was thought to be superior

to other scales as it prompted employees to consider several aspects of their work

environment (Ganster & Fusilier, 1989). However, the scale may also be

problematic, as although it ensures that respondents think of the same facets, some

facets may not be appropriate for some employees.

An alternative is to use a global scale of job autonomy. For example, the Job

Descriptive Survey (Hackman & Oldham, 1975) measures job autonomy through

assessing whether the employee has the opportunity for independence and freedom in

their job. Using this scale, the respondents can just consider the areas that are

important to them. They can include facets that are not specified in the facet scale,

and exclude facets that are not relevant to their workplace. As such, the supermarket

workers, although reporting low job autonomy on the multidimensional scale, may

have higher levels of global job autonomy. As such, future studies will need to

assess job autonomy using a global measure.

163

2.7.8 Conclusion

This study has contributed to the development of the proposed model of job

satisfaction (refer to Figure 2). This model, adapted from Karasek’s (1979) job

demand-control model, proposes that job autonomy relates to job satisfaction through

influencing the way employees manage their work difficulties. The findings

demonstrated that workers with higher job autonomy do manage their work

difficulties differently from workers with lower job autonomy. Specifically, as job

autonomy increases, primary control increases.

In addition to examining how job autonomy influences the use of control

strategies, this study also proposed that job autonomy influences the adaptiveness of

the control strategies. Primary control strategies were, as predicted, the most

adaptive strategies for the academics, however secondary control strategies were not

the most adaptive strategies for the supermarket workers. These findings supported

the primacy/back-up model, suggesting that all employees, whether they have low or

high job autonomy, should rely on primary control strategies when they face a

difficulty at work. However, as problems have now been identified with the primary

and secondary control scale and job autonomy scale, further research needs to

re-examine these hypotheses.

164

3 Chapter 3 - Study Two

165

3.1 Abstract

This study aims to re-test the proposal that job autonomy influences the amount of

control strategies that employees use, the relationship between the control strategies

and job satisfaction. This study attempted to overcome the major limitations

identified in study one, concerning the primary and secondary control scale and the

job autonomy scale. Furthermore, this study examined the influence of two new

variables, namely need for job autonomy and social support at work. Two

occupational groups that were expected to differ in their levels of job autonomy

(i.e., secondary school teachers and academics) were compared. It was expected that

the academics would report higher job autonomy, higher primary control, and lower

secondary control than the teachers. Furthermore, it was expected that primary

control would be more adaptive for the academics, whereas secondary control would

be more adaptive for the teachers. These hypotheses were not supported however, as

both groups reported equally high levels of primary and secondary control, and

primary and secondary control were not related to job satisfaction. These

inconsistent results prompted a review of the underlying assumptions of the study.

Some methodological limitations were identified in the hypotheses examining job

autonomy and the control strategies. Despite this, support for the remaining

hypotheses highlighted the importance of social support at work in predicting job

satisfaction.

166

3.2 Proposal for Study Two

This study re-examines the proposal that job autonomy influences the use and

the adaptiveness of primary and secondary control strategies. It attempts to

overcome the limitations identified in study one. This study uses: a) a revised

version of the Primary and Secondary Control Scale; b) a new measure of job

autonomy; and c) different occupational groups for comparison. Furthermore, this

study incorporates recent research suggesting that the need for job autonomy

mediates the relationship between job autonomy and job satisfaction, and examines

how social support at work influences job satisfaction. These changes will now be

discussed.

3.2.1 a) The Primary and Secondary Control Scale

The Primary and Secondary Control Scale, developed by Heeps et al., (2000)

was implemented in study one because it was one of the best scales that concurred

with Rothbaum et al’s., (1982) and Heckhausen and Schulz’s (1995) definition of

control. However, the scale was exploratory, and study one highlighted some

problems with the scale. As such, a review was conducted on the scale in

collaboration with RoseAnne Misajon. This review, which was based on factor

analyses of the scale, highlighted several problems with the scale. These problems

involved: i) the stem of the item; ii) the content of the item; and iii) the rating scale.

From this review, a third and fourth edition of the Primary and Secondary Control

Scale was developed.

167

Factor analyses conducted on the first and second edition of the Primary

Control and Secondary Control Scale were reviewed (e.g., Cahill, 1998; Cousins,

2001; Maher & Cummins, 2001, Misajon, 2002; Misajon & Cummins, in press;

Spokes, 1998). These analyses were based on a variety of samples, including elderly

people, people with arthritis, people with multiple sclerosis, and academics. Each

researcher tended to make minor changes to the scale, where they may have excluded

some items, or changed the wording of others, to make the scale more suitable to

their sample. These researchers then conducted exploratory factor analyses on the

scale, and found that the items initially loaded on 3, 4 or 5 factors. As they were

often unable to explain these factors, they then requested two factors. The resulting

analyses are displayed in Table 16. In this table, items that were excluded from that

particular version of the scale are represented by NA. Items that did not load on any

factors, or alternatively loaded on both factors are represented by a dotted line. Items

that loaded on the primary control factor are bolded, whilst items that loaded on the

secondary control factor are not bolded. This table demonstrates that the primary

control items generally factored well, however the secondary control items often

loaded on both factors. The primary control scale will be discussed first.

168

Table 16- Factor Analysis of Primary and Secondary Control Scale

Study 1 2 3 4 5 6 7Primary Control ItemsNew ways to achieve goal 0.56 0.80 0.60 0.53 0.56 0.70 0.79Persistence 0.75 0.73 0.64 0.62 0.84 0.71 0.81Remove obstacles 0.71 0.70 0.66 0.56 0.74 0.51 NAInvest time 0.60 0.61 0.72 0.66 0.81 0.51 0.72Learn skills 0.59 0.74 0.76 0.65 0.49 NA 0.67Ask for help or advice 0.37 0.56 0.36 0.61 0.68 NA NAEffort to make it happen 0.31 0.53 ---- 0.54 0.82 0.39 0.65

Secondary Control ItemsPositive Re-interpretation 0.46 0.32 0.60 0.57 0.44 ---- 0.42Wisdom 0.47 0.58 0.49 0.51 0.41 ---- 0.47Illusory optimism 0.54 ---- 0.52 0.67 0.37 ---- ----Downward social comparison

0.65 0.50 0.68 0.43 0.77 ---- 0.46

Past success 0.71 0.52 0.71 0.31 0.81 ---- ----Vicarious 0.76 0.60 0.41 0.98 0.82 ---- ----Positive approach 0.63 NA ---- 0.65 0.54 ---- ----Goal disengagement ---- NA NA NA 0.82 0.72 0.76Present success 0.57 0.73 0.62 0.72 0.58 0.74 ----Predictive negative ---- 0.66 0.37 ---- 0.74 0.61 0.54Attribution 0.43 0.76 0.41 0.43 0.74 0.40 ----Behavioural avoidance 0.30 ---- ---- ---- 0.49 0.69 0.76Active avoidance ---- NA NA ---- 0.67 0.74 0.76Sour grapes 0.39 NA NA NA NA 0.61 0.54Support ---- 0.32 ---- 0.57 0.77 ---- ----Give up NA 0.59 -0.49 -0.58 NA ---- ----

Studies 1 = Maher (2001); 2 = Misajon (2000); 3 = Spokes (1998); 4 = Cahill (1998); 5 = Misajon (2001); 6 = Study one; 7 = Cousins (2001)

Bolded factor loadings refer to the primary control factor.

169

3.2.1.1 Primary Control Scale

3.2.1.2 i) Stem of Primary Control Items

The factor analyses in Table 16 demonstrate that the primary control items

generally factor well. However, one reason why the primary control items may have

loaded on a different factor to the secondary control items is that the primary and

secondary control items were presented separately in the scale. The primary control

items began with “when I find a goal that is difficult to reach”, “when I really want

something” and “when something gets in the way of a goal”, whereas the secondary

control items all began with “when something bad happens that I cannot change.”

The reason the two strategies have different stems is that it was originally

assumed that primary control strategies were only used when a person faced a

difficulty that they could change, and that secondary control strategies were only

used when the difficulty could not be changed. This assumption may be incorrect

however, as it is possible for people to use secondary control when they face a

situation that they can change. For example, an employee may be upset that a co-

worker is always late. They may know that if they use primary control and talk to

their supervisor about the problem, the co-worker will be reprimanded, and as such

begin to arrive on time. However, they may choose not to use primary control as

they may then lose their friendship with the co-worker. Rather, they may implement

secondary control, and tell themselves that the problem “doesn’t matter.”

Similarly, it is possible that people use primary control when they face

situations that they cannot change. For example, an employee may dislike their work

170

times, yet be aware that the work times cannot be changed. Even so, they may

attempt to change their working times through using primary control, and discussing

solutions with their supervisor. The supervisor would presumably reject their

proposal, and the primary control strategy would have failed. Despite knowing the

possibility of primary control failure however, the employee may have decided to

take a risk.

As it is possible for primary and secondary control to be used in controllable

and uncontrollable situations, the scale was changed so that the stems of the items are

the same. The revised scale includes the primary and secondary control items

together, with the following introductory sentence; “Here are ways people deal with

difficult situations in their lives. How often have you had these thoughts when

facing a difficulty over the past week?” Examples of these thoughts are “it will work

out okay in the end” and “I knew it would happen.” The other control items which

involved actions rather than thoughts had an alternative introductory paragraph;

“How often have you done these things when facing a difficulty over the past week?”

(i.e., “I kept trying”, “I told someone about it”, “I worked to overcome it”).

3.2.1.3 ii) Primary control Item Content

As demonstrated in Table 16, all studies found that the items assessing new

ways to achieve goals, persistence, remove obstacles, learn skills and invest time,

loaded on the primary control factor. The items measuring effort to make it happen,

and ask others for help or advice occasionally loaded on the secondary control factor.

These two items were deleted as they were criticised for being similar to secondary

171

control strategies. Specifically, the item referring to effort to make it happen,

generally worded as “I think I can make it happen” does not actually refer to the

person putting in effort, and rather is similar to the secondary control strategy of

illusory optimism (i.e., “I know it will work out okay in the end”). The other item

referring to asking for help or advice was also deleted as it difficult to separate it

from secondary control. Indeed asking the boss or someone who has some power

over the problem for help or advice may be a means of changing the environment.

However, discussions with people who have less power over the situation, such as

friends, may only serve to make the person accept the problem.

Although the remainder of the primary control items loaded on the primary

control factor, there were still conceptual problems with the items. For example, the

item referring to learning skills was deleted from the scale, as it is only relevant if the

person is attempting to achieve something, and cannot be applied to the new stem,

namely difficult situations. Furthermore, the item assessing investing time was

omitted, as it was not necessarily indicative of primary control. A person may spend

lots of time on a problem, yet not attempt to change the environment.

The remainder of the items were criticised as they examined whether the

employees believed that they could change their situation, rather than examining how

they change their situation. For example, the items “I think I can make it happen”, “I

believe I can achieve it” and “I believe I can overcome it” measure whether a person

believes that they can change a situation. As discussed in chapter 2, it is questioned

whether believing that one can change a situation is a measure of primary control. A

person may report that they can change a situation for a variety of reasons, not just if

172

they use primary control strategies when they face difficulties. For example, a

person may report that they can change their environment because they have high

optimism. Alternatively, they may be using the secondary control strategy illusory

optimism, where they tell themselves that “everything will work out okay in the

end.” As such, the revised primary control scale, displayed in Table 17, is changed

to examine perceived strategies (e.g., exerting more effort, working harder) rather

than beliefs.

Table 17- Original and Revised Primary Control Items

Primary Control Strategy 2nd Edition (Heeps et al., 2000)

4th Edition (Maher et al., 2001)

New ways to achieve goal I think about different ways to achieve it

I looked for different ways to overcome it

Effort to make it happen I think I can make it happen

I worked to overcome it

Invest time I think about it a lot NAPersistence I believe I can achieve it I kept tryingRemove obstacles I believe I can overcome

itI worked out how to remove obstacles

3.2.1.4 iii) Rating Scale

The primary control items were originally rated on a 10-point scale ranging

from 1 (do not agree at all) to 10 (agree completely). This rating scale indicates

whether an individual agrees that they have used a strategy, not how much they have

used a strategy. Two people may report that they agree completely that they have

used a strategy, however one may use it 10 times a day, whilst the other may use it

once a week. As the scale did not differentiate between these people, the primary

control rating scale was changed to assess frequency.

173

In order to reduce inaccuracies, the scale was changed from measuring the

control strategies that people generally use when they face a difficulty to examining

the strategies people have used over the past week. As such, the rating scale ranged

from 0 (never) to 10 (every time).

3.2.1.5 Secondary Control

3.2.1.6 i) Stem of Secondary Control Items

As previously discussed, it was thought that the primary control items may

have loaded on a different factor to the secondary control items because the stems of

the items were different. In order to overcome this, the secondary control items were

placed with the primary control items. The stem of the item was changed from

“when something bad happens that I cannot change” to “how often have you done

these things when facing a difficulty over the past week.”

3.2.1.7 ii) Item Content

As demonstrated in Table 16, a few secondary control items loaded on both

the primary control factor and the secondary control factor. There did not appear to

be a consistent pattern in these studies however, with some studies finding that an

item loaded on a secondary control factor, whilst others found that it loaded on a

primary control factor. It was originally expected that the secondary control items

would form one factor, however it is now proposed that each item measures a

different strategy and that these strategies are independent. One person may use one

174

secondary control strategy in all situations, and so not use any of the others. This

proposal has implications for the scoring of the secondary control scale, and also for

factor analyses of the scale.

In regard to scoring, the proposal that respondents’ scores on secondary

control items may not be consistent suggests that the secondary control items cannot

be aggregated. However, secondary control can still be measured by using the

highest scoring item. This scoring procedure will be explained in detail later.

In regard to factor analyses, the proposal that respondents’ scores on

secondary control items are not consistent may explain why the secondary control

items loaded on more than one factor. Respondents may report different scores on

all the secondary control items, and thus they would not be expected to cluster

together. As such, rather than eliminating any items which loaded on a primary

control factor, the items were examined in terms of their theoretical usefulness.

Many of the items were similar to others, such as past success and present

success, and positive approach and behavioral avoidance. Present success (“I think

about my success in other areas”) encompasses past success (“I remember I have

accomplished a lot in life”). Furthermore, positive approach (“I do something nice to

take my mind off things”) and behavioral avoidance (“I do some physical exercise or

try to relax”) could be combined to measure active avoidance (“I do something to

take my mind off things”).

Two items that had been deleted from the first edition of the scale, namely

denial and support, were reinstated. Denial, measured by the item “I ignored it” was

deleted from previous versions of the scale as it was thought to be similar to the item

175

for goal disengagement (i.e., “It doesn’t matter”). However, telling oneself that a

problem is not important is clearly different from denying that the problem exists.

Intuitively, goal disengagement may be more adaptive than denial.

Support, measured by the item “told someone about it” was also added to the

scale. It was originally deleted from the first edition of the scale as it was a

behavioural strategy. It was assumed that all secondary control strategies had to be

cognitive strategies. This is not the case however, and Heckhausen and Schulz

(1995) recommend that the distinction between primary and secondary control

should not be based on behavioural versus cognitive, rather whether it involves

changing the environment versus changing the self. Support allows the person to

change themselves and become more likely to accept a situation.

After this theoretical analysis, 12 secondary control strategies remained (refer

to Table 18). These strategies were grouped according to their purpose. All of the

strategies are designed to make the person feel better about their situation, however

they may do this by reducing negative feelings (i.e., self-protective) or by increasing

positive feelings (i.e., self-affirmative). As demonstrated in Table 19, people may

reduce negative feelings by telling themselves that a difficult situation is not their

fault, that they knew it would happen, or that it doesn’t matter. People may increase

positive feelings however by thinking that they are better off than many other people,

and thinking about areas of their life in which they have been successful.

176

Table 18- Original and Revised Secondary Control Items

Secondary control strategy

Second Edition 4th Edition

Positive re- interpretation

I can see that something good willcome if it

I looked for somethingelse that was positive in the situation

Wisdom I remember you can’t always getwhat you want

I can’t always get what I want

Illusory-optimism I know thing will work out OK inthe end

It will work out okay in the end

Downward socialOmparison

I remember I am better off thanmany other people

I am better off than many other people

Past success I remember I have alreadyaccomplished a lot in life

NA

Vicarious I remember the success of myfamily and friends

I thought of the success of my family or friends

Positive approach I think nice thoughts to take mymind off it

NA

Goal disengagement

I tell myself it doesn’t matter It doesn’t matter

Predictive- negative

I don’t feel disappointed because Iknew it might happen

I knew it would happen

Attribution I can see it is not my fault It was not my faultActive avoidance I ignore it by thinking about other

thingsI did something different, like going for a walk

Sour grapes I realise I didn’t need to control itanyway

NA

Present success I think about my success in otherareas

I thought about my success in other areas.

Denial NA I ignored itSupport NA Told someone about it

Table 19- Functions of the Secondary Control Strategies

Use Definition of use Secondary Control StrategySelf-protective

Reduces the negative impact of the situation

Illusory optimism, goal disengagement, predictive negative, attribution, denial, wisdom

Self-affirmation

Increases positive feelings about self

Downward social comparison, vicarious, present success, support, positive re-interpretation, active avoidance

177

3.2.1.8 iii) Rating Scale

As with the primary control items, the secondary control items were changed

to measure frequency. Each strategy was rated on an 11-point scale ranging from

0 (never) to 10 (every time).

3.2.1.9 Summary

Following a review of the factor analyses conducted on the primary and

secondary control scale, and an investigation of the item stem, the item content, and

the rating scale, a revised scale was developed. This scale, presented in Appendix H

will be implemented in the second study. One further point that requires discussion

however, is the scoring of the control scale.

3.2.1.10 Scoring the primary and secondary control scale

Previous versions of the Primary and Secondary Control Scale (Heeps et al.,

2000) averaged across the strategies to obtain an overall score for primary control

score and an overall score for secondary control. The problem with this method

however, is that a person may report that they use one secondary control strategy

every time (10) and report never (0) for the remaining strategies. Calculating the

average level of secondary control in this situation would result in a low score. As

they used a secondary control strategy every time they faced a difficulty in the

previous week, a low score is not representative of their secondary control use.

One solution to this problem is to take the highest score for primary control

and the highest score for secondary control. Using this method, a person who reports

178

10 for one secondary control strategy and 0 for the rest would receive a score of 10

for secondary control. Another person may report different scores for all secondary

control strategies, including a 5, 7, 8, 10, 2, 3, 4. This person would also receive a

score of 10, as it is the highest score. The fact that the second person has higher

scores on other secondary control items does not mean that the person uses more

secondary control strategies, only that they use a greater variety of secondary control

strategies.

3.2.2 b) Job Autonomy Scale

The next limitation that was identified in study one concerns the

measurement of job autonomy. The autonomy scale implemented in study one was

a multidimensional scale. The scale examined specific facets of the work, such as

variety of work, pace of work, scheduling of rest breaks, and interaction with others.

This multidimensional scale was advantageous as it prompted the employees to

consider all aspects of their work. This is important as employees may fail to

consider some facets of their work. They may have accepted for example that they

cannot change their pay, policies and amount of interruptions, and thus no longer

expect to be able to make choices in these areas. The multidimensional scale ensures

that all workers think about the same job facets.

However, it is now recognised that the multidimensional scale is also

problematic. Although the scale prompts employees to consider all aspects of their

work, some of the facets may not be appropriate for them, or important to them.

With a global scale, the respondent can include facets that are not specified in the

179

facet scale, and exclude facets that are not relevant to their workplace. As such, their

response is only based on facets that they think are important. They may exclude

some facets because they have lowered their expectations, however if they have

accepted them, then they are not expected to influence their levels of job satisfaction.

3.2.3 c) Occupational Groups

This study will compare two occupational groups that have different levels of

job autonomy, low and high. As in study one, university academic staff have been

selected for the high job autonomy group. Academics traditionally have flexibility in

their work and freedom to pursue their own research interests (Winefield, 2000).

They can often choose among a variety of tasks, including research, teaching, and

administration (Fisher, 1994). Whether this theoretical expectation existed in

practice was tested in study one. The results demonstrated that the academics

reported a level of job autonomy which was 53%SM. It could not be ascertained

whether this score was high however, as there was few comparative studies for

Ganster’s (1989, cited in Dwyer & Ganster, 1991) multidimensional scale of job

autonomy. Study two will overcome this problem by relying on a scale, which has

been used more extensively.

Secondary school teachers have been selected for the low job autonomy

group. Teachers have been selected rather than supermarket workers because this

study is attempting to minimise the differences between the groups. In study one, it

was demonstrated that although the supermarket operators reported higher job

autonomy, and higher secondary control than the academics, the two groups reported

180

similar levels of job satisfaction. However, there were differences between the two

occupational groups that may have accounted for the similar levels of job

satisfaction. The supermarket workers would have experienced fewer difficulties at

work than the academics, and may have had lower job expectations than the

academics.

Study two attempts to examine two groups which have similar experiences at

work, but which have differing levels of job autonomy, namely secondary school

teachers and academics. Although both occupational groups deliver education to

students and have similar roles, it is expected that teachers will report lower job

autonomy.

Although few studies have examined Australian teachers’ levels of job

autonomy, a recent report proposes that although the Government attempted to

empower schools and teachers through providing schools with more responsibility,

teachers are experiencing reduced autonomy (Senate Employment, Education and

Training References Committee, 1998). Teachers are reporting that they want to

have more involvement in decision-making. One study which interviewed 956

Australian teachers about the changes they felt were necessary to reduce stress

(Teacher Stress in Victoria, 1990) found that the most common change (80%) was to

increase staff collaboration and communications. They also mentioned increasing

consultations before major decisions are made.

The type of decisions that the teachers want to be consulted on concern

curriculum selection, development and implementation (Senate Employment,

Education and Training References Committee, 1998). It is particularly important

181

that the teachers are involved in curriculum selection so that they can have control

over the means of producing the results by which they will then be judged (Cole,

1989).

In summary, this study will test the major hypotheses by comparing

academics and teachers. Although little research has examined these two groups, it

is expected that the academics will report higher job autonomy than the teachers.

3.2.4 d) Need for Job Autonomy

In study one, it was assumed that high job autonomy was beneficial for all

employees. This assumption was based on Karasek and Theorell’s (1990, p. 12)

proposal that “if jobs were redesigned with high job decision latitude…demands

would be seen as challenges and would be associated with increased learning and

motivation, with more effective performance and less risk of illness.” However, it

must be noted that other researchers have suggested that people may differ in the

extent to which they like to exercise control over their environment (Burger &

Cooper, 1979; Parkes, 1989). This difference in need for autonomy may influence

the relationship between job autonomy and job satisfaction, where job autonomy

may have greater influence on job satisfaction when need for job autonomy is high.

Only a few studies have examined the moderating role of need for job

autonomy on the relationship between job autonomy and job related outcomes (e.g.,

de Jonge, Landeweerd & Breukelen, 1994, cited in de Rijk, Le Blanc, Schaufeli, &

de Jonge, 1998; Nicolle, 1994). These studies have tended to produce inconsistent

findings. For example, de Rijk et al., (1998) cite de Jonge et al’s., (1994) study as

182

providing evidence that the need for autonomy moderated the relationship between

job autonomy and emotional exhaustion and health complaints. When de Rijk et al.,

(1998) replicated the study however, they failed to find support for the moderating

role.

One other study, conducted by Nicolle (1994) provides some support for the

moderating role of need for autonomy. This study demonstrated that for nurses with

a low need for autonomy, job autonomy was positively related to absenteeism,

however for nurses with a high need for autonomy, job autonomy was not related to

absenteeism. These results must be interpreted with caution however, as only 3 of

the 36 analyses were significant.

One further study has been reported to provide evidence for the moderating

role of need for autonomy. De Rijk et al., (1998) cited Gaziel’s (1989) study on

school administrators as being supportive of the hypothesis. According to De Rijk et

al., (1998) this study demonstrated that for administrators who had a low need for

autonomy, job autonomy was not related to job satisfaction, whereas for

administrators who had a high need for autonomy, there was a positive relationship

between job autonomy and job satisfaction. Examination of the study demonstrates

that this is not the case however. Gaziel’s (1989) study did not examine the

relationship between autonomy and job satisfaction for workers with differing levels

of job autonomy. Rather, the study examined the major predictors of a perceived

deficiency in autonomy.

In summary, it has been suggested that employees may differ in their need for

autonomy and that the relationship between job autonomy and job satisfaction may

183

differ depending on this need. As only a few studies have examined this proposed

moderating effect, and as the studies tend to be inconsistent, clearly further research

is needed.

3.2.5 e) Addition of Social Support

As mentioned in chapter 1, the job demand-control model was extended to

include social support (Johnson & Hall, 1988; 1994; Johnson, Hall & Theorell, 1989;

Karasek & Theorell, 1990). Social support at work refers to “overall levels of

helpful social interaction available in the job from both co-workers and supervisors”

(Karasek & Theorell, 1990, p.69). Two major types of social support have been

identified, namely emotional support and instrumental support. According to

Karasek and Theorell (1990, p. 70), emotional support refers to the “degree of social

and emotional integration and trust between co-workers, supervisors and others”,

whereas instrumental support refers to “extra resources or assistance with work tasks

given by co-workers or supervisors.” The job demand-control-support model

proposes that social support at work predicts job satisfaction.

Study one did not examine social support at work as it focussed on

understanding how job autonomy influences the control strategies, and on personality

and life satisfaction. However, after examining research on the relationship between

social support and job satisfaction further, social support appears to be an extremely

important predictor, and as such, study two will examine social support at work in

more detail.

184

Social support at work has been shown to directly and indirectly increase job

satisfaction. In regard to the direct effects, several researchers have demonstrated

that social support at work is positively related to job satisfaction (r = 0.52; Dollard

et al., r = 0.66, Munro et al., 1998), and negatively related to job dissatisfaction

(r = -0.29, r = -0.28; LaRocco, House & French, 1980). These studies suggest that

workers who report higher social support tend to be more satisfied with their jobs.

One possible explanation for the positive relationship between social support

and job satisfaction is that social support reduces the negative effects of work

demands. This explanation, known as the buffering hypothesis, proposes that social

support at work buffers the potentially negative effects of high demands on job

satisfaction. Only a few studies have examined the buffering hypothesis for job

satisfaction.

A review of these studies, conducted by Van Der Doef and Maes (1999)

demonstrated that only two (i.e., Karasek, Triantis & Chaudry, 1982; Landsbergis,

Schnall, Dietz, Friedman & Pickering, 1992) of the six studies (Chay, 1993; de Jonge

& Landeweerd, 1993, cited in Van der Doef & Maes, 1999; Melamed, Kushnir &

Meir, 1991; Parkes & von Rabenau, 1993) that examined the buffering hypothesis

were supportive. Their review found no major differences among the studies to

account for the inconsistent findings except that both supportive studies used male

samples and the others used mixed or female samples.

One difference among the studies that may explain the findings is the

operationalisation of social support at work. For example, Karasek et al., (1982)

measured tolerance of supervisor, attentiveness of supervisor, instrumental support of

185

supervisor, demands of supervisor, number of co-workers, instrumental co-worker

support, and emotional co-worker support. Alternatively, Chay (1993) relied on the

Interpersonal Support Evaluation List (Cohen, Kamarack, Mermelstein & Hoberman,

1985) which measures appraisal support, belonging support, tangible support and

esteem support. A similar and briefer scale was employed by Landsbergis et al.,

(1992), who relied on Karasek and Theorell’s (1990) scale. This scale measures

emotional and instrumental support from co-workers and supervisors. There is

certainly no agreed upon way of measuring social support at work (Unden, 1996),

and as such, it is unclear if the operationalisation of social support influenced the

results. What is clear is that the buffering role of social support requires more

investigation.

In summary, although it is intuitively expected that social support at work

would reduce the negative effects of job demands or job stressors, the results are far

from consistent. As there are such few studies however, more research is required.

186

3.3 Model of Job Satisfaction

A revised model of job satisfaction, displayed in Figure 6, will be tested.

This model is similar to that presented in Figure 2, as the major proposal of the

model is that primary and secondary control mediate the relationship between job

autonomy and job satisfaction. However, this model includes new sections on social

support and need for job autonomy. In Figure 6, these changes are represented by

bolded variables and arrows. These new proposals will be discussed.

It is now proposed that the relationship between job autonomy and job

satisfaction is moderated by need for job autonomy. It should not be assumed that all

employees desire high autonomy. Indeed, some workers may have low job

autonomy yet still report high job satisfaction because they do not desire freedom

and independence in their job. Need for job autonomy and job autonomy predict the

interaction term (i.e., need for job autonomy x job autonomy), which in turn predicts

job satisfaction.

It is also proposed that social support at work influences job satisfaction

directly and indirectly. It is expected to be positively correlated with job satisfaction,

and to also moderate the effect of work difficulties on job satisfaction. In Figure 6,

this is represented by the interaction term. Difficulties at work and social support at

work together predict the interaction term (i.e., difficulties x social support), which in

turn predicts job satisfaction.

187

Figure 6 -Revised Model of Job Satisfaction for Study 2

Difficulties at work

Social Support at work

Job Autonomy x Need for job autonomy

Difficulties x Social Support

Need for job autonomy

Job Autonomy

Primary Control

Job Satisfaction

Personality Life Satisfaction

Secondary Control

Job Autonomy x Secondary Control

Job Autonomy x Primary Control

188

3.4 Aims and Hypotheses

This study will compare levels of job autonomy, control strategies, and job

satisfaction reported by university academic staff and secondary school teachers. It

will also test the extent to which job autonomy mediates the relationship between

primary and secondary control strategies using the revised measures of job autonomy

and control strategies. The hypotheses are as follows:

1) Job autonomy will be positively related to job satisfaction, and the academics will

report higher job autonomy than the teachers.

This hypothesis tests the basic assumptions of the study. It needs to be

demonstrated that job autonomy is related to job satisfaction, and that comparisons

made between the two occupational groups are valid.

2) The academics will report more primary control, and less secondary control than

the teachers.

As the academics have higher job autonomy, they are expected to be more

likely to successfully implement primary control strategies than the teachers. As

secondary control is used to compensate for, and avoid future primary control failure,

it is expected that the teachers will report more secondary control than the academics

3) Job autonomy will be positively related to primary control, and negatively related

to secondary control.

189

As in hypothesis two, this hypothesis is examining whether job autonomy

influences the use of primary and secondary control. However, unlike hypothesis

two, it is based on the measured level of autonomy rather than the expected

occupational level

4) Primary control will be more positively related to job satisfaction than secondary

control for the academics, and secondary control will be more positively related to

job satisfaction than primary control for the teachers.

This study proposes that job satisfaction results from a match between job

autonomy and control strategies. Based on the discrimination model, it is proposed

that primary control is most adaptive for employees who can control their work

environment (i.e., high job autonomy), and that secondary control is most adaptive

for employees who have little control over their environment (i.e., low job

autonomy). Although primary control is generally more adaptive than secondary

control, the teachers have a high probability of experiencing primary control failure

when implementing primary control strategies, and thus it is expected that, for them,

secondary control strategies will be more adaptive.

5) The relationship between the control strategies and job satisfaction is moderated

by perceived job autonomy.

This hypothesis, like hypothesis four, is testing whether job autonomy

influences the relationship between the control strategies and job satisfaction. Unlike

190

hypothesis four however, it is based on the measured level of job autonomy rather

than the assumed level of autonomy based on the occupation.

6) The relationship between job autonomy and job satisfaction is mediated by

primary and secondary control strategies.

This hypothesis is testing an explanation for the relationship between job

autonomy and job satisfaction. This explanation proposes that people who have high

job autonomy have high job satisfaction because of their use of the control strategies.

These workers use more primary control and less secondary control, and are thus

able to overcome their difficulties.

7) The academics will report higher job satisfaction and higher life satisfaction than

the teachers.

The academics are expected to report higher job satisfaction than the teachers

as they have higher job autonomy, and use more primary control and less secondary

control. This level of job satisfaction is expected to influence their level of life

satisfaction.

8) The influence of work difficulties on job satisfaction is moderated by levels of

social support at work.

This hypothesis is based on the job demand-control-support model (Karasek,

1979) which proposes that social support can reduce the effect of demands at work.

191

9) The relationship between perceived job autonomy and job satisfaction is

moderated by need for autonomy.

People may differ in their need for autonomy, and this will influence the

relationship between job autonomy and job satisfaction.

10) Job autonomy, control strategies, life satisfaction, personality, difficulties at

work, and social support at work, predict job satisfaction.

These are all of the variables included in Figure 6. These are the major

predictors of job satisfaction.

192

3.5 Method

3.5.1 Participants

The sample consisted of 108 university academic staff, and 97 secondary

school teachers. The academics were obtained from one university, whereas the

secondary school teachers were obtained from 20 Government schools. For the

academics, the response rate was 21%. The response rate of the teachers could not

be calculated as the questionnaires were collected from the staff room only if the

teachers were interested in completing the survey. The demographic characteristics

of the sample are displayed in Table 20. The bolded values demonstrate where the

largest proportion of the sample lies, which tends to be fairly consistent across the

groups.

193

Table 20- Demographics of the Academics and Teachers

Variable % Academic % TeachersGender Male 49 47

Female 51 53Age 18-25 0 2.1

26-35 8.3 15.536-45 32.4 37.146-55 44.4 39.256+ 14.8 6.2

Years in Occupation 0-5 13 11.36-10 27.8 11.311-15 22.2 13.416-20 7.4 18.620+ 29.6 45.4

Hours worked per week 21-30 6.5 4.131-40 7.4 12.441-50 47.2 45.451-60 32.4 27.861+ 6.5 10.3

3.5.2 Materials

Both the academics and the teachers received a plain language statement

(refer to Appendix I) and an anonymous questionnaire. The questionnaire consisted

of several scales, which measured job autonomy, need for job autonomy, primary

and secondary control, work difficulties, job satisfaction, life satisfaction, personality

and social support at work.


As discussed in the rationale for study two, a global measure of job autonomy

was administered. This scale developed by Hackman and Oldham (1975) is part of a

larger scale, the Job Diagnostic Survey. This scale is the most commonly used

instrument for measuring job autonomy (Spector, 1986). It consists of three items

194

that assess overall perceived job autonomy, such as “in my job, I can decide on my

own how to go about doing my work” (refer to Appendix J).

Although the psychometric properties of the scale have been questioned in the

past (Fried, 1991; Fried and Ferris, 1991), a major review which examined 15 years

of empirical research on the psychometric properties of the scale provided some

support. Taber and Taylor (1990) demonstrated that the average test-retest

correlations for the scale were moderate (r = 0.63), internal consistency was

moderate (0.69), and there was good discriminant validity.

Although these psychometric statistics are not exceptional, the use of the

scale has been supported in a recent review conducted by Boonzaier, Flicker and

Rust (2001). Furthermore, it must be noted that as mentioned by Breaugh (1989), a

better alternative is not available. Breaugh (1989, Breaugh, 1998) actually

developed a new measure of job autonomy, however this scale was deemed not to be

appropriate for this study as like Ganster’s (1989, cited in Dwyer & Ganster, 1991)

scale, it is multidimensional. As such, this study used the autonomy items of the Job

Diagnostic Survey (Hackman & Oldham, 1975). In this study, Cronbach’s Alpha

was 0.83.

3.5.2.2 Need for Job Autonomy

As there are only a few studies that have examined need for job autonomy,

the measures of need for job autonomy were reviewed. First, Fung-Kam (1998)

tested preference for job autonomy using Edwards (1959) Personal Preference

Schedule. This scale consists of 28 sets of paired statements representing different

195

personality traits and a score is given to the respondent who chooses the statement

representing the personality trait of need for autonomy. The major problem with this

scale is that it refers to general autonomy, and not specifically to autonomy at work.

Another need for job autonomy scale is Algera’s (1981, cited in Landeweerd

& Boumans, 1994) scale. This scale asks the respondents to rate the attractiveness of

various work situations. Although this scale may have been adequate, to date it has

only been published in Dutch, and as such was not viable.

An exploratory scale was developed by de Rijk et al., (1998). This scale

consists of four items which examine how important it is for the person to set the

pace of their tasks, have control over what they do at work and the way that they do

it, doing their own planning at work, and giving orders instead of receiving them.

This scale was selected for the current study even though psychometric statistics

have not been produced, as the items have face validity. These items were rated on a

10-point scale, ranging from 1 (not at all important) to 10 (could not be more

important; refer to Appendix K).

As this scale is exploratory, a factor analysis was conducted on the scale to

ensure that the items were measuring need for job autonomy. The assumptions were

met, where Bartlett’s test of sphericity was large and significant, and Kaiser-Meyer-

Olkin (KMO) measure of sampling adequacy exceeded 0.6. A principal components

analysis with direct oblimin rotation yielded one factor. Examination of the

eigenvalues however demonstrated that the second factor had an eigenvalue of 0.999,

and as such a two-factor solution was tested. This analysis, displayed in Table 21,

demonstrates that item four (i.e., “How important is it for you to give orders instead

196

of receiving them”) loaded on the second factor. Item four is different to the other

three items as it may also measure need for authority. As a result, item four was

deleted from the scale. When all four items were included in the scale, Cronbach’s

alpha was low (0.56), however when item four was deleted, Cronbach’s alpha was

adequate (0.77).

Table 21- Factor Analysis of the Need for Job Autonomy Scale

No. Item F1 F21 How important is it for you to set the pace of your

tasks at work.0.83

2 How important is it for you to have control overwhat you do at work and the way that you do it.

0.87

3 How important is it for you to do your own planning at work.

0.80

4 How important is it for you to give orders to work instead of receiving them.

0.99

Eigenvalues 2.098 0.999 % of variance 52.45 24.98 Cumulative variance 52.45 77.43 Cronbach's Alpha (for revised scale) 0.77

Loadings less than 0.40 are excluded; Bolded items are included in the scale

3.5.2.3 Primary control and Secondary Control

As discussed in the rationale, the Primary and Secondary Control Scale

developed by Heeps et al., (2000) was revised for this study (Maher et al., 2001).

The scale now includes four primary control items and 12 secondary control items

(refer to Appendix H). These items are rated on a 11-point scale ranging from 0

(never) to 10 (every time). Although the control strategies were aggregated in study

one, it now appears that this scoring method is flawed. The items cannot be

aggregated as people may use one strategy all the time, and never use the others.

197

Using the average, they would receive a score that is not representative of the

frequency of secondary control strategies (i.e., every time). As such, an alternative

solution used here is to record the highest frequency for primary control strategies

and the highest frequency for secondary control strategies.

3.5.2.4 Work Difficulties

Work difficulties were measured in the Primary and Secondary Control Scale

(Maher et al., 2001; refer to Appendix H). Prior to assessing how the employees deal

with their work difficulties, the scale assesses the frequency of work difficulties.

Specifically, the item is “how often do you have difficulty doing something at work.”

The rating scale ranges from 1 (never) to 10 (all the time).


Two scales of job satisfaction were administered; a facet scale and a global

scale. The facet scale was changed from the Job Descriptive Index (JDI) in study

one to the Minnesota Satisfaction Questionnaire (MSQ; Weiss et al., 1967; refer to

Appendix L). The MSQ was used because unlike the JDI, which only examines five

facets of the job, the MSQ examines 20 facets. It was thought that a greater

understanding of the groups could be obtained by using the MSQ. Furthermore, the

items in the MSQ can be aggregated to measure intrinsic and extrinsic job

satisfaction. Intrinsic job satisfaction refers to how people feel about the nature of

the tasks, whereas extrinsic job satisfaction refers to how people feel about aspects of

the work situation that are external to the work itself (Spector, 1997). This scale has

198

adequate reliability where Cronbach’s alpha ranges from 0.82 to 0.88 and

discriminant validity has been demonstrated (Hirschfeld, 2000).

The facet measure is only useful to gain insight into the teachers’ and

academics’ level of job satisfaction. It cannot be used as the dependent variable

however as a facet scale cannot be aggregated (Ironson et al., 1989). It may exclude

areas that are important to the respondent, or include areas that are unimportant to the

respondent. As such, a one-item measure of job satisfaction was used as the

dependent variable. Although internal consistency cannot be established with a

single-item measure, the single item measure of job satisfaction has been shown to

correlate with other measures of job satisfaction, where r = 0.63 (Wanous et al.,

1997).


As in study one, the subjective dimension of the Comprehensive Quality of

Life Scale (Com-QOL) developed by Cummins (1997) was used to assess

satisfaction with seven domains of life, including material well-being, health,

productivity, intimacy, safety, community and emotional well-being (refer to

Appendix E). An 11-point scale was utilised, ranging from 0 (completely

dissatisfied) to 10 (completely satisfied)

3.5.2.7 Personality

The extroversion and neuroticism subscales of the NEO Five Factor

Inventory, developed by Costa and McCrae (1992) were used to measure personality.

199

This scale, discussed in study one, contains 12 items to measure extroversion and 12

items to measure neuroticism (refer to Appendix F). Convergent and discriminant

validity of both of these personality factors has been established (Costa & McCrae,

1992, Leong & Dollinger, 1991; Tinsley, 1994).

3.5.2.8 Social Support at Work

Social support at work was measured by Karasek and Theorell’s (1990) scale

which has two components; supervisor support and co-worker support (refer to

Appendix M). Each component is measured by 4 items, and rated on a scale from

1 (not true at all) to 10 (could not be more true). Two items measure emotional

support, and two measure instrumental support. Emotional support measures the

degree of social cohesion in the work group, whilst instrumental support measures

the amount of assistance given with work tasks. Although the scale measures

emotional and instrumental support, the four items are summed to provide an overall

support score.

The items in the scale were changed slightly to ensure that they referred to

the employee. Some of them were quite ambiguous, such as “my supervisor shows

concern” and “my supervisor pays attention.” As these items could be interpreted in

regard to work tasks or other employees, they were changed to “my supervisor shows

concern for me” and “my supervisor pays attention to me.”

Past studies using the original scale have demonstrated that the scale has

adequate reliability with Cronbach’s alpha ranging from 0.69 to 0.89 (Karasek et al.,

1998), and 0.81 to 0.87 (Pelfrene, Vlerick, Mak, De Smets, Kornitzer & De Backer,

200

2001). Furthermore, factor analyses have demonstrated that the supervisor support

items load on a different factor to the co-worker support items (Pelfrene et al., 2001).

3.5.3 Procedure

Ethics approval was obtained from Deakin University, and the Department of

Education, Employment and Training (DEET). Consent was obtained from the

Heads of School to recruit the academics, and from the Principals for the teachers.

The recruiting procedure differed depending on the group. Five hundred academics

within one University were sent a questionnaire package. If they chose to participate

in the study, they completed the questionnaire and returned it using a reply paid

envelope. For the teachers, each Principal that agreed to assist with the study was

sent 10-15 questionnaires. They then discussed the questionnaires in their staff

meetings, and left them in the staff room for the teachers to collect. On occasion, the

Principals chose to distribute the questionnaires to a selection of staff members.

These questionnaires were sent back to Deakin University using a reply-paid

envelope. At the conclusion of the study, the participating Heads of School and the

Principals received a summary of the results.

201

3.6 Results


The data set for each occupational group was initially examined for missing

values, acquiescence, outliers, normality and linearity. Less than 5% of the values

for academics and teachers were missing for any one item. As there was no pattern

to these missing values, they were, as in study one, replaced with the group mean.

Univariate outliers were identified in the primary and secondary control scale (5), the

job autonomy scale (1), the facet job satisfaction scale (2), and the life satisfaction

scale (18). These values were recoded to lie within three standard deviations of the

mean.

Normality was assessed using the skew/standard error<3,

Kolmogorov-Smirnof values, frequency histograms, and normal probability plots.

For the academics, job autonomy (-5.76), and co-worker support (-3.24) were mildly

negatively skewed. For the teachers, job satisfaction (-3.79), supervisor support

(-3.44), and co-worker support (-5.16) were mildly negatively skewed. As in study

one, these variables were not transformed as transformations are not recommended

for data that are mildly and naturally skewed (Tabachnick & Fidell, 1996). Rather,

these were examined using the more conservative alpha level of 0.01. Linearity was

assessed through bivariate scatterplots, and these appeared to demonstrate reasonable

linear relationships.

202


Table 22 contains the means and standard deviations for the major variables

in the study for each occupational group. Whilst the teachers reported lower job

autonomy than the academics, they reported similar levels of job satisfaction, and

primary and secondary control. Table 23 displays the correlations among all of the

major variables for the academics and the teachers. This table demonstrates that

although job autonomy is correlated with job satisfaction, primary and secondary

control strategies are not.

Table 22- Means and Standard Deviations of Major Variables for Academics

and Teachers

Variable Academics TeachersM SD M SD

Job Satisfaction-One item 64.09 21.72 68.79 20.23Intrinsic Job Satisfaction 72.76 13.24 77.07 14.80Extrinsic Job Satisfaction 44.46 20.45 56.15 20.89Job Autonomy 74.93 16.81 66.32 17.79Primary Control 81.58 14.42 80.64 15.82Secondary Control 83.33 13.37 82.93 11.90Life Satisfaction 74.20 11.34 75.61 14.48Neuroticism 35.07 16.60 33.33 16.04Extroversion 61.62 13.06 63.92 15.69Supervisor Support 44.57 29.44 64.31 26.56Co-worker Support 71.11 18.90 77.49 15.57Difficulties at work 49.59 24.08 46.05 23.01

All scores have been converted to a percentage of scale maximum (%SM) which ranges from 0-100. The formula is (mean score for the original domain-1) x 100/ (number of scale points –1).

203

Table 23- Inter-Correlations for the Academics and Teachers

JS JA PC SC LS N E SS CSJS 0.51** 0.14 0.14 0.46** -0.29** 0.23* 0.64** 0.39**

JA 0.37** 0.06 0.06 0.38** -0.14 0.17 0.39** 0.34**

PC -0.07 0.12 0.56** 0.15 -0.19 0.01 0.01 -0.14

SC -0.06 0.00 0.40** 0.27** -0.08 0.19* -0.05 -0.06LS 0.38** 0.20* 0.09 0.03 -0.39** 0.52** 0.36** 0.26**

N -0.38** -0.38** -0.07 0.03 -0.48** -0.32** -0.15 0.03E 0.22* 0.16 0.20* 0.13 0.26** -0.44** 0.16 0.18SS 0.46** 0.26** -0.15 -0.09 0.19* -0.06 0.08 0.55**

CS 0.42** 0.15 0.07 0.05 0.23** -0.31** 0.38** 0.39**

Di -0.31** -0.20* -0.07 0.03 -0.33** 0.25** -0.09 -0.13 -0.26**

* p<0.05 , ** p>0.01; Correlations for teachers are bolded.

JS = Job satisfaction; JA = Job autonomy; PC = Primary control; SC = Secondary control; LS = Life satisfaction; N = Neuroticism; E = Extroversion; SS = Supervisor support; CS = Co-worker support; Di = Difficulties at work

3.6.3 Preliminary Examination of the Primary Control and Secondary

Control Scale

As the primary and secondary control scale is exploratory, it will be

examined here before the hypotheses are tested. The descriptive statistics displayed

in Table 22 and 23 indicate that primary and secondary control did not behave as

expected. Both the academics and the teachers reported high levels of primary

control (M = 82%SM, M = 81%SM), and secondary control (M = 83%SM,

M = 83%SM). Furthermore, the control strategies did not correlate with job

satisfaction. One interesting finding however is that primary control was positively

correlated with secondary control for both groups (r = 0.40, r = 0.56).

Overall however, these statistics are inconsistent with study one, where the

supermarket workers (M = 46%SM) reported significantly higher levels of secondary

204

control than the academics (M = 36%SM). Furthermore, primary control was

moderately correlated with job satisfaction (r = 0.31, r = 0.44).

A major difference between these two studies is the edition of the Primary

and Secondary Control Scale. The scale was changed for study two, where a new

scoring procedure was implemented. In the first study, the items were simply

aggregated for each control strategy, however in the current study, the highest

frequency for primary and secondary control was recorded. This method does not

appear to have been successful however in differentiating the respondents. The

frequency distribution, displayed in Table 24, demonstrates that 76% of the subjects

reported a level of primary control between 77%SM and 100%SM, and that 84% of

the subjects reported a level of secondary control between 77%SM and 100%SM.

This range is concerning, suggesting that there may have been a ceiling effect.

Table 24- Frequency of Primary and Secondary Control

Value Primary Control Frequency %

Secondary Control Frequency %

33.33 2.0 0.544.44 2.4 1.555.56 4.4 3.466.67 15.1 9.377.78 25.4 33.788.89 32.2 32.7100.00 18.5 19.0

In order to examine how much the scoring procedure in this study influenced

the resulting levels of primary and secondary control, the Primary and Secondary

Control Scale was also examined as in study one, where the average was calculated.

Before the items were aggregated, a factor analysis was conducted on the scale.

205

The assumptions were met, where Bartlett’s test of sphericity was large and

significant, and Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy

exceeded 0.60. A principal components analysis with direct oblimin rotation yielded

four factors. These four factors accounted for 57% of the variance. Only 9 of the 16

items loaded on only one factor. Two of the primary control items loaded on the first

factor, however a secondary control item also loaded on this factor. The remaining

secondary control items (i.e., 6) were equally distributed among the second, third and

fourth factor. There was no pattern to these items, and the four-factor solution could

not be interpreted. The analysis was repeated requesting three factors to determine if

a three-factor solution could be useful. This analysis was no clearer however, with

the primary control and secondary control items loading on all three factors. The

secondary control items that loaded on different factors did not appear to be

measuring different functions of secondary control.

In response to these analyses, a two-factor solution was requested. This

analysis provided a much clearer solution, with the four primary control items

loading on the first factor, and six of the 12 secondary control items loading on the

second factor. It must be noted however, that this two-factor solution only accounted

for 40% of the variance. Another factor analysis was conducted with only the bolded

items in Table 25. This analysis demonstrated that the two factors accounted for

49% of the variance.

206

Table 25- Factor Analysis of the Revised Primary and Secondary Control Scale

No. Item F1 F2pc1 I looked for different ways to overcome it. 0.67pc2 I kept trying. 0.77pc3 I worked to overcome it. 0.82pc4 I worked out how to remove obstacles. 0.59sc1 It will work out okay in the end. 0.41 0.53sc2 I knew it would happen. 0.71sc3 I can’t always get what I want. 0.68sc4 It doesn’t matter. 0.64sc5 I am better off than many other people. 0.40 0.40sc6 It was not my fault. 0.59sc7 I told someone about it. 0.61sc8 I thought of the success of my family or friends.sc9 I thought about my success in other areas. 0.58sc10 I did something different, like going for a walk. 0.41sc11 I ignored it. 0.54sc12 I looked for something else that was positive in the

situation.0.71

Eigenvalues 4.41 2.11 % of variance 26.98 13.19 Cumulative variance 27.59 40.78 Cronbach's Alpha (for revised scale) 0.82

Loadings less than 0.40 are excluded; Bolded items are included in the scale

When the control items were aggregated rather than separated into the highest

score, both groups still reported similar levels of the control strategies. For primary

control, the academics reported a mean of 79.09, whilst the teachers reported 80.18.

For secondary control, the academics reported 46.21, and the teachers reported

46.56. Furthermore, the control strategies were not strongly related to job

satisfaction. Primary control was not related at all to job satisfaction, whilst

secondary control was slightly negatively related to job satisfaction (r = -0.24).

These descriptive statistics suggest that even if the scales were aggregated, the results

would still not be significant.

207

As the scoring procedure does not appear to have drastically altered the

results, the hypotheses will be tested using the intended scoring procedure

(i.e., highest number). This scoring method, although problematic because of its

small range, is theoretically superior to the aggregated measure. However,

preliminary analyses using this scoring method clearly demonstrate that primary and

secondary control strategies are not behaving as expected, and as such, many of the

hypotheses will not be supported. In order to reduce the repetitiveness of these

findings, the hypotheses examining primary and secondary control will be examined

collectively. Specifically, this refers to hypotheses 2, 3, 4, 5, and 6.

208


In order to test the proposed model of job satisfaction, multivariate analyses

of variance were conducted to compare the levels of control strategies, job

autonomy, job satisfaction and life satisfaction reported by the academics and the

teachers. Multiple regression analyses were also conducted to examine the major

predictors of job satisfaction, and the moderating role of need for autonomy and

social support at work. As in study one, the alpha level was reduced to 0.01 in order

to reduce the risk of Type I errors.

3.7.1 Hypothesis One: Levels of Job Autonomy and Job Satisfaction

In order to test the first part of hypothesis one, proposing that job autonomy is

positively related to job satisfaction, the correlation coefficients for each

occupational group were examined. Consistently, job autonomy was positively

related to job satisfaction for both the academics (r = 0.37) and the teachers

(r = 0.51).

In order to test the second part of hypothesis one, proposing that the

academics would report higher job autonomy than the teachers, an analysis of

variance was employed. The assumption of univariate homogeneity of variance,

using Levene’s test, was met, F (1, 203)= 2.58, p > 0.05. The univariate test of

significance demonstrated that, as hypothesised, the academics (M = 75.13,

SD = 16.12) reported significantly higher job autonomy than the teachers,

(M = 66.34, SD = 17.74), F (1, 203) = 13.82, p = 0.00.

209

3.7.2 Hypotheses Two and Three: Examining how Job Autonomy Influences

the Amount of Primary and Secondary Control Strategies

Hypotheses two and three examine how job autonomy influences the amount

of control strategies that employees use. In order to test hypothesis two, that the

academics would report more primary control and less secondary control than the

teachers, a multivariate analysis of variance was employed. The variables were

normally distributed, and reasonably linear relationships were evident. There was no

evidence of multicollinearity as the determinant of the within-cell correlation

> 0.0001 (i.e., 0.774). Univariate homogeneity of variance, as assessed through

Levene’s test was met for primary control, F (1, 203) = 0.50, p > 0.05, and secondary

control, F (1, 203) = 1.68, p > 0.05. The assumption of multivariate homogeneity of

variance was also met using Box’s M test.

The multivariate test using Pillai’s criterion was not significant,

F (2, 202) = 0.09, p = 0.91. Inconsistent with hypothesis two, the academics

(M = 81.58, SD = 14.42) reported similar level of primary control as the teachers

(M = 80.64, SD = 15.81), F (1, 203) = 0.20, p = 0.66. Furthermore, the academics

(M = 83.33, SD = 13.37) reported similar levels of secondary control as the teachers

(M = 82.93, SD = 11.90), F (1, 203) = 0.05, p = 0.82.

In order to test hypothesis three that job autonomy is positively related to

primary control and negatively related to secondary control, job autonomy was

correlated with primary and secondary control. This analysis demonstrated that,

inconsistent with hypothesis three, primary control (r = 0.09) and secondary control

(r = 0.03) were not significantly related to job autonomy.

210

3.7.3 Hypotheses Four and Five: Examining how Job Autonomy Influences

the Relationship Between the Control Strategies and Job Satisfaction

Hypothesis four and five test the proposal that job autonomy influences the

relationship between job autonomy and job satisfaction. In order to test hypothesis

four, proposing that primary control will be more positively related to job satisfaction

than secondary control for the academics, and secondary control will be more

positively related to job satisfaction than primary control for the teachers, two

standard multiple regression analyses were conducted.

The assumptions of normality, linearity, and homoscedasticity of residuals, as

assessed through examination of the residual scatterplots, were met for both groups.

As demonstrated in Table 26, R was not significantly different from zero for the

academics, R = 0.08, F (2, 105) = 0.32, p = 0.73, or for the teachers, R = 0.16,

F (2, 94) = 1.16, p = 0.32. Inconsistent with hypothesis four, primary and secondary

control were not related to job satisfaction for either group.

211

Table 26- Multiple Regression of Primary and Secondary Control on Job

Satisfaction for Academics and Teachers

Group IV JS PC B sr2 (unique)Acad

PC -0.07 -0.08 -0.05 0.22SC -0.06 0.40 -0.07 -0.04 0.14

R =0.08 R2=0.006 AdjR2=-0.013Teach

PC 0.14 0.12 0.09 0.55SC 0.14 0.56 0.15 0.09 0.50

R =0.16 R2=0.02 AdjR2=-0.003

Acad – Academics; Teach – Teachers; PC - Primary control; SC – Secondary control; JS – Job satisfaction

For the academics, R is composed of 0.36% unique variance and 99.64% shared variance. For the teachers, R is composed of 1.05% unique variance and 98.95% shared variance.

Hypothesis five, similar to hypothesis four, examines how job autonomy

influences the adaptiveness of the control strategies, however it is based on the

measured level of job autonomy rather than the presumed level. Hypothesis five

proposes that the relationship between the control strategies and job satisfaction is

moderated by job autonomy. As discussed in study one, job autonomy is a

moderator rather than a mediator as it specifies when certain effects will hold. That

is, when job autonomy is high, primary control will be more strongly correlated with

job satisfaction, and when job autonomy is low, secondary control will be more

strongly correlated with job satisfaction.

In order to test the moderating effect of job autonomy on primary control and

secondary control, two hierarchical multiple regression analyses were conducted. In

212

these analyses the control strategy was entered first, then job autonomy, and then the

interaction term. The assumptions of normality, linearity, and homoscedasticity of

residuals, as assessed through examination of the residual scatterplots were met.

As demonstrated in Table 27, for primary control, R was not significantly different

from zero after the first step (i.e., primary control), R = 0.03, F (1, 203)= 0.15,

p = 0.70. The addition of job autonomy did result in an increase in R, where

R = 0.39, Finc (1, 202) = 36.29, p = 0.00. However, the addition of the interaction

term was not significant, R = 0.39, Finc (1, 201) = 0.12, p = 0.73.

For secondary control, R was not significantly different from zero after the

first step, R = 0.02, F (1, 203)= 0.09, p = 0.77. After job autonomy was entered, the

value of R increased, R = 0.39, F (1, 202) = 36.35, p = 0.00, however the addition of

the interaction term in step three was not significant, R = 0.39, F (1, 201) = 0.21,

p = 0.65. Thus, hypothesis five was not supported.

213

Table 27- Hierarchical Multiple Regression testing the Moderating role of the

Control Strategies on the Relationship Between Job Autonomy and Job

Satisfaction

Step IV DV B sr2 (unique)1 Primary control JS 0.04 0.032 Primary control -0.01 -0.009

Job Autonomy 0.48 0.39 15.21**3 Primary control 0.12 0.08

Job autonomy 0.62 0.51Primary control x job autonomy

-0.001 -0.16

R =0.39 R2=0.15 AdjR2=0.14

Step IV DV sr2 (unique)1 Secondary control JS 0.03 0.022 Secondary control 0.01 0.009

Job Autonomy 0.47 0.39 15.21**3 Secondary control 0.23 0.14

Job autonomy 0.72 0.59Secondary control x job autonomy

-0.002 -0.24

R =0.39 R2=0.15 AdjR2=0.14

*p<0.05, ** p<0.01; JS – Job satisfaction

3.7.4 Hypothesis Six: Examining the Proposed Explanation for the


In order to test hypothesis six, which proposes that the relationship between

job autonomy and job satisfaction is mediated by the control strategies, a hierarchical

multiple regression analysis was conducted.

214

The assumptions of normality, linearity and homoscedasticity of residuals

were met, and there was no evidence of multicollinearity. As demonstrated in Table

28, R was not significantly different from zero after primary and secondary control

were entered, R = 0.03, F (2, 202) = 0.08, p = 0.92. R did significantly increase after

job autonomy was added to the equation, R = 0.39, Finc (3, 201)= 36.16, p = 0.00.

Only job autonomy predicted job satisfaction accounting for 15% of the variance in

job satisfaction. As such, when primary and secondary control were controlled for,

job autonomy still predicted job satisfaction. Inconsistent with hypothesis six,

primary and secondary control did not mediate the relationship between job

autonomy and job satisfaction.

Table 28- Hierarchical Multiple Regression Testing the Mediating Role of the

Control Strategies

Step IV DV B sr2(unique)1 Primary control JS 0.03 0.02

Secondary control 0.02 0.01

R =0.03 R2=0.001 AdjR2=-0.009

2 Primary control JS -0.03 -0.02Secondary control 0.03 0.02Job autonomy 0.48 0.39 15.21**

R =0.39** R2=0.15 AdjR2=0.14

**p>0.01; JS – Job satisfaction

215

3.7.5 Hypothesis Seven: Occupational Differences in Job Satisfaction and

Life Satisfaction

In order to test hypothesis seven, that academics would report higher job

satisfaction than the teachers, an analysis of variance was conducted on the one-item

measure of job satisfaction. Years working in occupation and age were entered as

covariates in this analysis as the academics tended to be older, and had worked less

years than the teachers (i.e., refer to Table 20). Life satisfaction was normally

distributed, and reasonably linear relationships were evident. Univariate

homogeneity of variance, as assessed through Levene’s test was met,

F (1, 203)= 0.33, p > 0.05. The univariate test demonstrated that the academics

(M = 64.09, SD = 21.72) reported similar levels of job satisfaction as the teachers

(M = 68.79, SD = 20.23), F (1, 201) = 3.39, p = 0.07.

Although the one-item measure of job satisfaction is the dependent variable

in this study, the facet measure of job satisfaction was also explored to gain a greater

understanding of the two occupational groups. A multivariate analysis of variance

was conducted on the intrinsic and extrinsic facets of job satisfaction. The variables

were normally distributed, and reasonably linear relationships were evident. There

was no evidence of multicollinearity as the determinant of the within-cell correlation

> 0.0001 (i.e., 0.526). Univariate homogeneity of variance, as assessed through

Levene’s test was met for intrinsic job satisfaction, F (1, 203) = 1.63, p > 0.05, and

extrinsic job satisfaction, F (1, 203)= 0.03, p > 0.05. The assumption of multivariate

homogeneity of variance was also met using Box’s M test.

The multivariate test, using Pillai’s criterion was significant,

216

F (1, 199) = 5.96, p = 0.00. Examination of the univariate tests demonstrated that the

teachers (M = 56.15, SD = 20.89) reported higher extrinsic job satisfaction than the

academics (M = 44.46, SD = 20.45), F (1, 203)= 16.35, p = 0.00. The teachers

(M = 77.07, SD = 14.80) also reported higher intrinsic job satisfaction than the

academics (M = 72.76, SD = 13.24), F (1, 203)= 4.84, p = 0.03, however this finding

was not significant as the more stringent alpha level of 0.01. The means and

standard deviation for the intrinsic and extrinsic items for academics and teachers are

provided in Table 29.

217

Table 29- Means and Standard Deviations of the Intrinsic and Extrinsic Job

Satisfaction Items for Academics and Teachers

Item Academics TeachersM SD M SD

Being able to keep busy all the time 71.70 25.35 81.21 19.60The chance to work alone on the job 80.45 16.49 73.88 22.51The chance to do different things from time to time

77.77 21.69 78.24 21.75

The change to be somebody in the community

65.74 24.87 69.64 21.12

The way my boss handles his/her work 47.32 30.62 67.35 25.85The competence of my supervisor in making decisions

51.95 30.60 70.10 24.23

Being able to do things that don’t go against my conscience

68.72 25.01 78.92 21.90

The way my job provides for steady employment

77.47 25.21 87.51 21.35

The chance to do things for other people 77.77 18.10 87.05 14.85The chance to tell people what to do 59.16 22.52 67.12 20.34The chance to do something that makes use of my abilities

77.05 20.05 80.06 20.15

The way company policies are put into practice

30.34 24.28 48.68 24.91

My pay and the amount of work that I do 49.59 26.46 45.02 28.79The chance for advancement on the job 43.41 28.64 53.15 29.88The freedom to use my own judgement 74.18 20.42 72.62 22.22The chance to try my own methods of doing the job

73.25 20.50 76.51 21.51

The praise I get for doing a good job 44.14 30.96 52.57 28.31The feeling of accomplishment I get from the job

69.86 22.73 72.05 21.76

Bolded items measure extrinsic job satisfaction. Non-bolded items measure intrinsic job satisfaction.

In order to test the second part of hypothesis seven, that the academics would

report higher life satisfaction than the teachers, a univariate analysis of variance was

employed. Life satisfaction was normally distributed using the skewness/standard

error < 3 criterion. The assumption of univariate homogeneity of variance, as

218

assessed by Levene’s test, was not met, F (1, 203) = 0.96, p < 0.05, and as such, this

analysis proceeded with caution using an alpha level of 0.05.

The univariate test of significance demonstrated that, inconsistent with

hypothesis seven, there were no occupational differences in levels of life satisfaction,

F (1, 203) = 0.51, p = 0.11. As demonstrated in Table 30, the teachers average level

of life satisfaction was 74.21 (SD = 11.34) and the academics average level was

75.61 (SD = 14.48). The means and standard deviations for the seven domains of

life satisfaction are also presented to demonstrate that the two groups appear to be

more satisfied with safety, intimacy and material well-being, and less satisfied with

health and community.

Table 30- Means and Standard Deviations of the Domains of Life Satisfaction

for Academics and Teachers

Domain Academics TeachersM SD M SD

Material Satisfaction 76.87 17.91 76.08 16.92Health Satisfaction 68.73 21.37 67.35 23.22Productivity Satisfaction 75.03 14.58 73.03 18.84Intimacy Satisfaction 76.77 19.98 78.97 20.33Safety Satisfaction 81.80 16.87 84.56 16.67Community Satisfaction 70.24 18.72 72.98 20.57Emotional Satisfaction 72.91 18.79 76.59 19.62Overall life satisfaction 74.35 11.16 75.64 14.45

3.7.6 Hypothesis Eight: Examining how Social Support at Work Moderates

the Relationship between Difficulties at Work and Job Satisfaction

Hypothesis eight proposes that social support at work moderates the effect of

work difficulties on job satisfaction. Social support at work is proposed to be a

219

moderator, that is, a variable that affects the direction and/or strength of the

relationship between an independent variable (i.e., work difficulties) and a dependent

variable (i.e., job satisfaction). It is a moderator rather than a mediator because it

affects the relationship between work difficulties and job satisfaction, however it

does not explain why work difficulties and job satisfaction are related.

In order to test this moderation effect, a hierarchical multiple regression is

required. In the first step the independent variable is entered (i.e., work difficulties).

In the second step the moderator variable (i.e., social support) is entered. Finally, in

the third step the interaction term is entered (i.e., independent variable multiplied by

the moderator variable). Moderator effects are evident if the interaction term

predicts the dependent variable after the independent variable and the moderator

variables have been entered in steps one and two.

Two hierarchical multiple regression analyses were conducted for supervisor

support, and co-worker support. For both analyses, the assumptions of normality,

linearity, and homoscedasticity of residuals were met, and there was no evidence of

multicollinearity.

The moderating role of supervisor support on the relationship between work

difficulties and job satisfaction was tested first. R was significantly different from

zero at the end of the first step (i.e., work difficulties), R = 0.26, F (1, 203) = 14.94,

p = 0.00. The addition of supervisor support resulted in a significant increment in R2,

where R = 0.57, Finc (1, 203) = 78.86, p = 0.00. Supervisor support accounted for

26% of the variance in job satisfaction. As demonstrated in Table 31, the addition of

the interaction term (difficulties x supervisor support) did result in a significant

220

increment in R2, where R =0.59, Finc (1, 201) = 4.29, p = 0.04. It must be noted that

this finding was not significant using the more stringent alpha level of 0.01.

Table 31- Hierarchical Multiple Regression Analysis Examining if Supervisor

Support Moderates the Relationship between Work Difficulties and Job

Satisfaction


1 Freq. of Diff JS -0.23 -0.26 6.86**

R =0.26** R2=0.07 Adj R2=0.06

2 Freq. of Diff JS -0.18** -0.20 3.84Supervisor 0.37** 0.52 26.11

R =0.57** R2=0.33 Adj R2=0.32

3 Freq. of Diff JS -0.36 -0.40 4.04**Supervisor 0.16 0.22Diff x Sup 0.04 0.36 1.39*

R =0.59* R2=0.34 Adj R2=0.33

p<0.05 , ** p>0.01; JS – Job satisfactionAlthough not significant at 0.01, the interaction between difficulties and

supervisor support will be examined further for two reasons. First, only a few

studies have examined the moderating role of social support at work, and as such, the

pattern of the interaction requires investigation. Second, it is difficult to achieve

statistical significance in moderation analyses as the power is low (Bobko, 2001).

By having the independent variable, the moderator and the interaction term

(independent variable x moderator), there is an increased chance of multicollinearity

(Bobko, 2001). As the correlation between predictors increases, the standard

221

deviation of the regression weights increases, and it becomes less likely that the null

hypothesis will be rejected. In order to reach significance therefore, the analysis

needs to have large effects of large sample sizes (Bobko, 2001). Thus, as the

analysis was significant at 0.05, it will be examined further.

Work difficulties were regressed on job satisfaction separately for those with

low supervisor support, and those with high supervisor support. As proposed by

Cohen and Cohen (1983), the low and high distinction was defined as scores that fell

one standard deviation below or above the mean for supervisor support. As

demonstrated in Figure 7, the regression lines were consistent with the hypothesis,

where the slope of the regression line of work difficulties on job satisfaction was

steeper for high supervisor support than for low supervisor support.

222

Figure 7 - Relationship Between Work Difficulties and Job Satisfaction for

Employees with Low/High Supervisor Support

223

In order to test the moderating role of co-worker support, another hierarchical

multiple regression was conducted. R was significantly different from zero at the

end of the first step (i.e., work difficulties), R = 0.26, F (1, 202)= 14.91, p = 0.00.

The addition of co-worker support resulted in a significant increment in R2, where

R = 0.46, Finc (1, 202)= 36.78, p = 0.00. As demonstrated in Table 32, co-worker

support accounted for 7% of the variance in job satisfaction. The interaction term

(difficulties x co-worker support) did not result in a significant increment in R2,

R = 0.46, Finc (1, 202) = 0.06, p = 0.81. As such, co-worker support does not appear

to moderate the effect of work difficulties on job satisfaction.

Table 32- Hierarchical Regression Analyses examining whether Co-worker

Support Moderates the Relationship between Work Difficulties and Job

Satisfaction.


1 Freq. of Diff JS -0.23 -0.26 6.86**

R =0.26** R2=0.07 Adj R2=0.06

2 Freq. of Diff JS -0.18 -0.20 4.00**Co-worker 0.46 0.38 14.36**

R =0.46** R2=0.21 Adj R2=0.20

3 Freq. of Diff JS -0.13 -0.14 0.12Co-worker 0.50 0.42 2.25*Diff x Co-worker -0.008 -0.07 0.023

R =0.46 R2=0.21 Adj R2=0.20

p<0.05 , ** p>0.01; JS – Job satisfaction

224

3.7.7 Hypothesis Nine: The Moderating Role of Need for Autonomy on the


In order to test hypothesis nine that need for autonomy moderates the

relationship between job autonomy and job satisfaction, a hierarchical multiple

regression analysis was conducted. In this case, need for autonomy is a moderator

variable, as it specifies when the relationship between job autonomy and job

satisfaction will be strong or weak.

The assumptions of normality, linearity, and homoscedasticity of residuals

were met, and there was no evidence of multicollinearity. In the first step, job

autonomy was entered, followed by need for autonomy in the second step. Finally,

the interaction term (i.e., autonomy x need for autonomy) was entered. R was

significantly different from zero at the end of the first step, R = 0.39,

F (1, 203)= 36.61, p = 0.00. The addition of need for job autonomy did not result in

a significant increment in R2, where R = 0.39, Finc (1, 203) = 0.02, p = 0.89. The

addition of the interaction term did not result in a significant increment in R2, where

R = 0.39, Finc (1, 201) = 0.23, p = 0.63. As demonstrated in Table 33, need for job

autonomy does not moderate the relationship between job autonomy and job

satisfaction.

225

Table 33- Hierarchical Regression Analyses examining whether Need for Job

Autonomy Moderates the Relationship between Job Autonomy and Job

Satisfaction.

Step IV DV B sr2(unique)

1 Job autonomy JS 0.47 0.39 15.29**

R =0.39** R2=0.15 Adj R2=0.15

2 Job Autonomy JS 0.48 0.39 14.51**Need for job autonomy -0.19 -0.01

R =0.39 R2=0.15 Adj R2=0.15

3 Job Autonomy JS 0.78 0.64Need for autonomy 2.35 0.12Job Autonomy x Need for autonomy

-0.03 -0.31

R =0.39 R2=0.15 Adj R2=0.14

** p<0.01; JS – Job satisfaction

3.7.8 Hypothesis Ten: Major Predictors of Job Satisfaction

In order to test hypothesis ten, which examines several major predictors of

job satisfaction, two standard multiple regression analyses were conducted for the

academics and the teachers. The following predictors were included: primary and

secondary control; job autonomy; personality (neuroticism and extroversion); life

satisfaction; social support at work (supervisors and co-workers); and difficulties at

work. For both analyses, the assumptions of normality, linearity, and

homoscedasticity of residuals were met, and there was no evidence of

multicollinearity.

226

R was significantly different from zero for both the academics, R = 0.65,

F (9, 98) = 7.92, p = 0.00, and the teachers, R = 0.75, F (9, 87) = 12.49, p = 0.00.

The major predictors of job satisfaction were the same for both occupational groups,

namely job autonomy and supervisor support at work. As demonstrated in Table 34,

job autonomy accounted for approximately 2% and 5% of the variance in job

satisfaction for the academics and the teachers respectively. The finding for the

academics must be examined cautiously however, as it was not significant at the

more stringent alpha level of 0.01. Supervisor support at work accounted for 5% and

13% of the variance in job satisfaction for the academics and teachers respectively.

227

Table 34- Standard multiple Regression Predicting Job Satisfaction for

Employees with Low Autonomy and Employees with High Autonomy

Group Variable B sr2 (unique)Acad

Primary Control -0.13 -0.09Secondary Control -0.01 -0.01Job Autonomy 0.23 0.17 2.28*Life Satisfaction 0.29 0.15Neuroticism -0.19 -0.15Extroversion 0.02 0.01Co-worker Support 0.20 0.18Supervisor Support 0.21 0.28 5.61**Difficulties at Work -0.10 -0.16

R =0.65** R2=0.42 AdjR2=0.37Teach

Primary Control 0.03 0.02Secondary Control 0.19 0.11Job Autonomy 0.28 0.25 4.70**Life Satisfaction 0.15 0.11Neuroticism -0.14 -0.11Extroversion -0.01 -0.008Co-Worker Support 0.04 0.03Supervisor Support 0.36 0.47 13.54**Difficulties at Work -0.11 -0.12

R =0.75** R2=0.56 AdjR2=0.52

p<0.05 , ** p>0.01; Acad – Academics; Teach- Teachers

For academics, R is composed of 14.46% unique variance and 85.54% shared variance. For teachers, R is composed of 21.97% unique variance and 78.03% shared variance

As supervisor support appeared to account for the largest proportion of the

variance in job satisfaction for both occupational groups, a further regression

analysis was conducted to examine the value of R with only supervisor support. R

was significantly different from zero for the academics, R = 0.46, F (1, 106) = 27.81,

228

p = 0.00, and the teachers, R = 0.64, F (1, 95) = 67.29, p = 0.00. Supervisor support

at work accounted for 21% and 41% of the variance in job satisfaction for the

academics and the teachers respectively.

3.7.9 Conclusion

This study tested whether job autonomy influenced the use and adaptiveness

of primary and secondary control strategies. Inconsistent with the hypotheses, the

teachers and academics reported similar levels of primary and secondary control.

Furthermore, primary and secondary control strategies were not correlated with job

satisfaction. Although the control strategies did not mediate the relationship between

job autonomy and job satisfaction, the findings highlighted the importance of

supervisor support in predicting job satisfaction. These findings will now be

discussed.

229

3.8 Discussion

This study was designed to extend the job demand-control model (Karasek &

Theorell, 1990), testing an explanation for the positive relationship between job

autonomy and job satisfaction. This explanation proposed that job autonomy

influences the use and adaptiveness of the control strategies. Employees who

reported higher job autonomy were expected to successfully implement primary

control. The hypotheses were generally not supported however, and as such, a

thorough review of the assumptions and the hypotheses is required.

3.8.1 Assumption- The Academics Represent a High Job Autonomy Group

and the Teachers Represent a Low Job Autonomy Group

The major assumption underlying this study is that the academics were

expected to report higher job autonomy than the teachers as they have more freedom

and choice in many aspects of their work. Consistent with this expectation, the

academics (M = 75%SM) reported significantly higher job autonomy than the

teachers (M = 66%SM). Although this difference was significant, it must be

demonstrated that the difference is meaningful. One way to determine if the

difference is meaningful is to calculate the standard error of measurement (SEM;

Wyrwich, Nienaber, Tierney & Wolinsky, 1999). In the past, researchers have used

the SEM to determine clinically meaningful standards. It is estimated by multiplying

the standard deviation of the scale by the square root of one minus the reliability

coefficient, or

230

Although there is no consensus about how many SEMs an individuals score

must change for it to be considered significant, Wyrwich et al., (1999) suggest that a

2.77 SEM criterion is the safest. In their study however, they demonstrated that a

one SEM criterion reflected a minimal clinically important difference. In the current

study, the SEM is estimated to be 7.18. Thus, on average there appears to be one

SEM difference between the levels of job autonomy reported by academics and

teachers. This may not necessarily be meaningful as Wyrwich et al (1999) stress that

their results should not be generalised to other populations or tests.

Another way to examine if the difference is meaningful is to compare the

current levels of job autonomy with that reported by other occupational groups

(Hackman & Oldham, 1980). As demonstrated in Table 35, the means range from

58%SM to 73%SM. Compared to these occupational groups, the academics are in

the higher range and the teachers are in the lower range. However, it must be noted

that these data are relatively old. More recent studies have administered Hackman

and Oldham’s (1975) scale to different occupational groups, however they do not

separate the occupational groups (e.g., Renn & Vandenberg, 1995; Tiegs et al.,

1992).

231

Table 35- Normative Data for Hackman and Oldham’s (1980) Autonomy Scale

Occupation Normative Data (%SM)Professional 73.33Management 73.33Clerical 58.33Sales 63.33Service 66.66Processing 58.33Machine Trades 65.00Bench Works 60.00Structural Works 66.66

Although past studies do not shed much light on whether the differences in

levels of job autonomy reported by the academics and teachers are meaningful, it is

clear that the academics are reporting significantly higher job autonomy than the

teachers. These two groups may not represent the extremes of job autonomy,

however the difference should be great enough to demonstrate the expected

differences in the control strategies. It is assumed that the use of the control

strategies varies linearly with job autonomy over this range of values. As such, even

if the academics and teachers do not represent extremes of job autonomy, the

expected findings should be evident, albeit weaker.

In summary, it appears that the academics report higher job autonomy than

the teachers. It is difficult to ascertain whether this difference is meaningful,

however it is concluded that the difference should be great enough to demonstrate the

expected effects.

232

3.8.2 Hypothesis Testing

The major hypotheses tested in the study were that job autonomy influences

the use of primary and secondary control strategies, and also the relationship between

the control strategies and job satisfaction (i.e., adaptiveness). These hypotheses were

not supported, and as such possible explanations for the findings will be considered,

and the methodology will be re-examined.

3.8.3 Job Autonomy Influences the Amount of the Control Strategies

It was hypothesised that job autonomy would influence the amount of

primary and secondary control strategies. Academics, who have higher job

autonomy than the teachers, were expected to report using more primary control and

less secondary control. Furthermore, it was expected that job autonomy would be

positively correlated with primary control and negatively correlated with secondary

control. Inconsistent with these hypotheses however, the two occupational groups

reported similar levels of control strategies, and job autonomy was not related to the

control strategies.

These findings do not support the model of job satisfaction presented in

Figure 6. Specifically, the findings do not support the arrow from job autonomy to

primary and secondary control. This proposal was based on an extension of the life

span theory of control (Heckhausen & Schulz, 1995), which essentially proposes that

if a person can control a situation, they will attempt to change it. If they cannot

control the situation however, it is more likely that their attempts to change it would

fail, and thus they will seek to accept the situation. It is most surprising therefore

233

that the academics reported high secondary control, and that the teachers reported

high primary control. Explanations for these unexpected findings will be discussed.

3.8.3.1 Why did the Academics Report High Secondary Control?

The academics reported equally high levels of primary and secondary control

as the teachers. This means that when they face a difficulty at work, they use both

primary and secondary control strategies. It is interesting that they rely on secondary

control however, because theoretically, they should have less need for it than the

teachers. As they have reasonably high control over their environment, they are

expected to successfully implement primary control most of the time and rarely rely

on secondary control strategies. As this is clearly not the case, the use of secondary

control may need to be re-examined.

In addition to using secondary control to compensate for primary control

failure, secondary control may be used as a means of temporarily avoiding primary

control. If employees were faced with a large number of difficulties at work, they

may initially use secondary control. For example, workers may tell themselves that

it will work out okay in the end, or that it doesn’t matter. This may be necessary for

workers, such as academics, who face many difficulties, and must delay dealing with

some of them. Once they can deal with them however, it is expected that they do so

using primary control. Thus, secondary control may be used prior to primary

control.

The explanation that secondary control can be used prior to primary control

may explain why the academics reported high levels of secondary control. However,

234

this explanation does not account for the lack of occupational differences in

secondary control. Even if both groups use secondary control prior to dealing with

their difficulties, the teachers would be expected to rely on more secondary control

than the academics. After initially delaying dealing with a problem using secondary

control, it is expected that the academics would then use primary control, but that the

teachers would continue using secondary control.

3.8.3.2 Why did the teachers report high primary control?

Although the teachers were expected to rely mostly on secondary control,

they reported equally high levels of primary and secondary control. One explanation

for this finding is that the teachers may have avoided repeated primary control

failure. If they implemented primary control and failed, they were expected to rely

mostly on secondary control. Through relying on secondary control, and accepting

their situation rather than trying to change it however, they can then maintain their

perceptions of primary control. Thus, the teachers’ levels of primary control may be

explained, in part, by their reliance on secondary control strategies.

3.8.3.3 Summary

Both the academics and the teachers reported similar levels of primary and

secondary control. The academics reported higher secondary control than expected,

and the teachers reported higher primary control than expected. The academics’

level of secondary control may be explained by the proposal that secondary control

may also be used prior to primary control. The teachers’ levels of primary control

235

may be explained by the proposal that they can avoid primary control failure through

relying on secondary control strategies.

3.8.4 Job Autonomy Influences the Relationship Between the Control

Strategies and Job Satisfaction

It was hypothesised that job autonomy influences the adaptiveness of the

control strategies, such that primary control would be more positively related to job

satisfaction than secondary control for the academics, and that secondary control

would be more positively related to job satisfaction than primary control for the

teachers. Furthermore, it was hypothesised that the relationship between the control

strategies and job satisfaction would be moderated by job autonomy. Inconsistent

with both of these hypotheses however, primary and secondary control strategies

were not related to job satisfaction. This suggests that workers can use primary or

secondary control strategies to deal with their difficulties.

These findings are inconsistent with the primacy/back-up model and the

discrimination model (Thompson et al., 1998). The discrimination model proposes

that primary control is the most adaptive strategy in controllable situations, and that

secondary control is the most adaptive strategy in uncontrollable situations. The

primacy/back-up model proposes that primary control is the most adaptive strategy in

low-control and high-control situations. The current results demonstrated that

primary and secondary control strategies were not related to job satisfaction.

However, a major limitation has now been identified in the study that may render

these findings invalid.

236

3.8.5 Limitations in the Hypotheses Examining Job Autonomy and Control

Strategies

There was a methodological problem in this study that may have limited the

findings examining how job autonomy influences the use and the adaptiveness of the

control strategies. This problem concerns the specificity of the hypotheses. The

hypotheses tested whether job autonomy influenced the relationship between the

control strategies and job satisfaction at a general level, however the

primacy/back-up model and the discrimination model actually only refer to one

situation. These models propose that the controllability of a situation influences the

control strategies used to handle that situation. Thus, the hypotheses need to be

measured at a situational level, rather than at an occupational level.

This lack of consistency between the definition of the discrimination model

and the testing of the discrimination model is not limited to this study. All of the

studies that Thompson et al., (1998) claimed to test the discrimination model actually

fail to test it as specified in the definition.

To test the discrimination model, researchers need to correlate a measure of

perceived control over one situation with the control strategies used in that situation.

Past researchers have relied on specific measures of the controllability of the

situation, and the control strategies, however they aggregated them rather than

examining them separately.

For example, Thomson et al., (1996, 1994) developed a list of specific facets

relevant to living with HIV, such as progression of HIV infection, family

relationships, and quality of medical care. For each of these facets, respondents rated

237

how much control they had over them, and the extent to which they used primary and

secondary control to handle them. If they correlated each facet with the control

strategies used to handle that facet, they would be testing the discrimination model.

However, they aggregated the items to obtain an overall measure of perceived

control, an overall measure of primary control and an overall measure of secondary

control.

Another study measured specific controllability and control strategies, yet

failed to use this information to test the discrimination model. Thompson et al.,

(1998) examined control over physical appearance, measuring how much control

people had over the attractiveness of their hair, body strength and agility, weight and

body shape, skin and overall physical appearance. They also measured the primary

and secondary control strategies in relation to age-related changes over physical

appearance. Although they measured these specific variables however, they added

the perceived control scale to the primary control items to measure primary control.

Thus, they failed to examine whether the controllability of a situation influenced the

control strategies used in that situation.

It thus appears as though the current study and previous studies have failed to

adequately test the discrimination model. In order to do so, future studies need to

examine the controllability of the situation and the control strategies at a situational

level. It should then be tested whether the amount of autonomy an employee has

over a situation predicts the use and the adaptiveness of the control strategies in that

situation. As such, it is not necessary to examine the control strategies that

238

employees with high/low job autonomy are using, rather to examine which control

strategies all workers use in low-control and high-control situations.

3.8.5.1 Summary

Job autonomy did not influence the use or adaptiveness of the control

strategies. These findings were inconsistent with the proposed model of job

satisfaction, which attempted to explain the relationship between job autonomy and

job satisfaction. However, a major limitation was identified in the study, where it

appears as though the study has failed to test the discrimination model.

3.8.6 Other Predictors of Job Satisfaction

The remainder of the hypotheses will now be examined. These hypotheses

examine occupational differences in job and life satisfaction, the buffering role of

social support at work, and the moderating role of need for autonomy. Additionally,

the major predictors of job satisfaction that are included in the proposed model of job

satisfaction are examined.

3.8.7 Occupational Differences in Job Satisfaction and Life Satisfaction

Inconsistently, the academics did not report higher job satisfaction or life

satisfaction than the teachers. The finding that there was no difference in life

satisfaction is not surprising given that there were no occupational differences in job

satisfaction. Furthermore, the levels of life satisfaction reported by both groups were

within the normative range according to the homeostatic theory of life satisfaction

239

(Cummins, 1995, 2000b). It is surprising however that the academics (M = 64.09)

and the teachers (M = 68.79) reported similar levels of job satisfaction. The levels

of job satisfaction reported by both groups will firstly be compared with past studies.

3.8.7.1 Comparisons with Past Studies

Past research has reported varying levels of job satisfaction for academics.

Researchers have reported the following levels of job satisfaction; 57%SM (Leung et

al., 2000), 65%SM (Hill, 1986), 66%SM (Lahey & Vihtelic, 2000), 74%SM (Carson

et al., 2001), 82%SM (Olsen, 1993) and 83%SM (Niemann & Dovidio, 1998).

Furthermore, study one reported a level of 66%SM. The scales used in some of these

studies were criticised in study one, however a normative level of job satisfaction

was not established. All that can be concluded is that the academics in this study, as

with those in study one, report a level of job satisfaction that is within the range

found by other researchers.

In regards to teachers, researchers have reported several different levels of

job satisfaction. For example, Klecker and Loadman (1999) found a similar level of

job satisfaction (M = 68%SM) as the present study, however others have reported a

higher level of job satisfaction of 80%SM (Ma & Macmillan, 1999; Schonfeld,

2000). All of these studies are flawed however as they relied on poor measures of

job satisfaction.

For example, Ma and MacMillan (1999) included the following items to

measure job satisfaction; “I find my professional role satisfying”, “I look forward to

each day”, “I am committed to making our school one of the best in the province”

240

and “If I could start over, I would become a teacher again.” The item “I look

forward to each day” may actually be dependent on personality and quality of life, as

well as job satisfaction. Furthermore, the item “If I could start over, I would become

a teacher again”, is likely to be dependent on the teachers’ age and how much they

have invested into becoming a teacher. A teacher who has spent 20 years teaching

may agree with this item because they are satisfied with their job, or because they

want to justify why they are still in the profession. Until psychometric data are

obtained for this scale, the results are questionable.

Other researchers have relied on facet scales of job satisfaction. For example,

Klecker and Loadman (1999) measured satisfaction with salary, professional

advancement, professional challenge, autonomy, work conditions, interactions with

colleagues, and interactions with students. Facet scales are criticised however for

excluding facets that are important to the individual, or including facets that are not

important. Furthermore, it is expected that the teachers’ level of job satisfaction

would be lower in this scale, as it is dependent on their level of job autonomy. This

is problematic because although job autonomy is expected to be related to job

satisfaction, job autonomy may not be a domain of job satisfaction.

In general, it appears that the academics’ level of job satisfaction is

reasonably consistent with past studies. The teachers’ level of job satisfaction tends

to be slightly lower than previous studies, however these studies have relied on

inadequate measures of job satisfaction.

As few studies have examined academics’ and teachers’ levels of job

satisfaction, the findings can also be compared to the review conducted in study one,

241

which demonstrated an average level of job satisfaction of 66%SM. This average

level is consistent with both the academics and the teachers. Possible explanation for

the similar levels of job satisfaction will be presented.

3.8.7.2 Explaining the Similar Levels of Job Satisfaction

There are a number of explanations for the academics’ and teachers’ similar

levels of job satisfaction. First, the recruitment process was different for the teachers

and the academics. The academics were sent their questionnaire through internal

mail, whereas the teachers were, in some cases, given their questionnaire by the

Principal of the school. This may be problematic for the teachers, as the Principal, in

order to obtain positive results, may have given the questionnaires to happier

workers. Alternatively, even if social desirability was not important, the Principals

may have given the questionnaires to teachers that were more likely to agree, or were

more organised. Thus, it must be questioned whether the teachers included in the

sample are representative of the average teacher.

It is possible however that the teachers’ average level of job autonomy is

representative of the average teacher. Although teachers report a lower level of job

autonomy than the academics, there are certainly many other determinants of job

satisfaction. One major factor that was highlighted in this research was social

support at work. Examination of the descriptive statistics demonstrates that the

teachers report higher satisfaction with their supervisor support, and that supervisor

support was strongly correlated with job satisfaction.

242

3.8.8 The Influence of Social Support at Work on the Relationship Between

Work Difficulties and Job Satisfaction

Supervisor support, but not co-worker support moderated the relationship

between work difficulties and job satisfaction. The relationship between work

difficulties and job satisfaction was weaker when supervisor support was high. It

must be noted however that this hypothesis was only significant at an alpha level of

0.05, and not 0.01, suggesting that the effect may not be large. However, even if the

effect is not large, the findings suggest that supervisors should ensure that they

provide emotional and instrumental support to their employees. To do this, the

supervisor needs to show concern for their employees, and provide tangible

assistance (Karasek & Theorell, 1990).

The finding that supervisor support plays a greater role than co-worker

support has been reported by previous researchers (e.g., Beehr, 1985; Fenlason &

Beehr, 1994; Russell, Altmaier & Van Velzen, 1987). Co-workers have less

influence at work, and as such may have less influence over difficulties at work

(Fenlason & Beehr, 1994).

Consistent with the current findings, a few studies have demonstrated that

social support at work has positive moderating effects on job satisfaction (i.e.,

Karasek et al., 1982; Landsbergis et al., 1992). However, there are studies that have

failed to demonstrate the moderating role of social support (Chay, 1993; de Jonge &

Landeweerd, 1993, cited in Van Der Doef & Maes, 1999; Melamed at al., 1991;

Parkes & Von Rabenau, 1993).

243

One difference between these supportive and non-supportive studies is in the

measure of social support. Two of the supportive studies (i.e., current study and

Landsbergis et al., 1992) relied on Karasek and Theorell’s (1990) scale, whereas the

non-supportive studies relied on several different scales. However, there are too few

studies to draw conclusions about the influence of the scales. It is clear that further

research is needed to examine the moderating role of social support.

3.8.9 The Influence that Need for Job Autonomy has on the Relationship

Between Job Autonomy and Job Satisfaction

Need for job autonomy did not moderate the relationship between job

autonomy and job satisfaction. Although past empirical studies were equivocal, it

seemed intuitive that differences in need for autonomy would influence the

relationship between job autonomy and job satisfaction. The non-supportive finding

is consistent with a few past studies. For example, de Rijk et al., (1998) failed to

demonstrate that need for autonomy moderated the relationship between job

autonomy and emotional exhaustion and health complaints. Furthermore, Nicolle

(1994) found need for autonomy moderated the relationship between job autonomy

and absenteeism in only 3 of 36 analyses.

As the current study and previous studies generally fail to demonstrate that

need for job autonomy moderates the relationship between job autonomy and job

satisfaction/job stress, this hypothesis will no longer be investigated. Too few

studies have been conducted to simply conclude the effect does not exist, however it

seems that more testing is required to develop a valid measure of need for job

244

autonomy. Studies are relying on exploratory measures, and as such may not be

adequately measuring the need for job autonomy construct. As such, it is

recommended that researchers firstly attempt to develop a need for job autonomy

scale that is psychometrically sound.

3.8.10 Major predictors of Job Satisfaction

Inconsistent with the proposed model of job satisfaction, only job autonomy

and supervisor support uniquely predicted job satisfaction for both groups. The

relationship between these variables and job satisfaction was consistent with past

research. For job autonomy, r = 0.43 (Tiegs et al., 1992), and for social support at

work, r = 0.52 (Dollard et al., 2000), and r = 0.66 (Munro et al., 1998).

The control strategies, personality, life satisfaction, co-worker support, and

difficulties at work were moderately correlated with job satisfaction, however they

did not uniquely predict job satisfaction. As such, several changes need to be made

to the variables included in the model. The control strategies will be retained in the

model, however as discussed earlier, several changes will be made to the Primary

and Secondary Control Scale. Personality was a poor predictor of job satisfaction,

and as such, it will be excluded from study three. Life satisfaction was also a poor

predictor of job satisfaction, however it will be retained in the model as it is acting as

both an independent variable and a dependent variable. Supervisor support

explained much of the variance in job satisfaction, and as such both types of social

support (i.e., co-worker and supervisor) will be re-examined. Difficulties at work

245

will also be retained in the model, as it is necessary to demonstrate what the primary

and secondary control strategies are used for.

In summary, as a result of the model of job satisfaction only being partially

supported, several changes will be made in study three. These changes will be

explained further in chapter 4.

3.8.11 Conclusion

Although the findings demonstrated that job autonomy did not predict the use

or adaptiveness of the control strategies, one major limitation was identified in this

study. The hypotheses were criticised for being too general, and it was suggested

that study three should examine job autonomy and the control strategies at the

situational level rather than at the occupational level. Furthermore, this study

highlighted the importance of supervisor support at work, which will be examined

further in study three.

246

4 Chapter 4 - Study Three

247

4.1 Abstract

The major proposal of this study is that the controllability of a work difficulty

influences the use and adaptiveness of the control strategies used to handle that

difficulty. It was expected, based on the discrimination model, that in controllable

situations, employees would use more primary control than secondary control, and

that primary control would be the most adaptive. In uncontrollable situations

however, it was expected that employees would use more secondary than primary

control, and that secondary control would be the most adaptive strategy. These

proposals were not supported as employees reported using similar strategies for

controllable and uncontrollable difficulties. Furthermore, primary control strategies

were more adaptive than secondary control strategies for both types of difficulties.

These findings challenge the belief that control strategies are influenced by

situational variables and also question the assumption that primary control failure

negatively affects job satisfaction. The implications of these findings for the

proposed model of job satisfaction are discussed.

248

4.2 Proposal for Study Three

This study continues to test the proposal that job autonomy influences the use

and the adaptiveness of primary and secondary control strategies, however unlike

previous studies, it will be examined at a situational level, rather than an

occupational level. As such, changes are made to the specificity of the hypotheses

and the primary and secondary control scale. Further changes are made to the model

of job satisfaction, where it is proposed that the control strategies and social support

at work moderate the relationship between controllable and uncontrollable work

difficulties and job satisfaction.

4.2.1 Specificity of Hypotheses Testing the Proposal that Job Autonomy

Influences the Control Strategies

In this study, the proposal that job autonomy influences the use and

adaptiveness of the control strategies is examined at a more specific level. As

discussed in chapter 3, studies one and two were criticised as they were not

consistent with the definition of the discrimination model. The discrimination model

proposes that when the situation is controllable, primary control is more adaptive,

and when the situation is uncontrollable, secondary control is more adaptive.

However, empirical tests of the model, including studies one and two, have examined

perceived control and control strategies at a general level (i.e., Thompson et al.,

1996; Thompson et al., 1994; Thompson et al., 1998).

The difference between the definition and empirical tests of the

discrimination model may be important. If the discrimination model is tested at a

249

more specific level, the relationship between the two variables may be stronger. All

employees, whether they be low autonomy or high autonomy, are expected to have

high control over some aspects of their job and less control over other aspects.

Furthermore, all employees, whether they be low or high job autonomy, are expected

to use primary control in some situations and secondary control in others. By

correlating their general level of job autonomy with their general level of control

strategies, the results become less extreme, the low and high autonomy groups

become more similar, and the correlations become weaker.

In order to accurately test the discrimination model, study three will examine

whether the controllability of a situation influences the use and adaptiveness of the

control strategies used in that situation. Past research examining these proposals will

be examined.

4.2.2 Examining how the Controllability of a Difficulty Influences the Use of

the Control Strategies

As proposed by the life span theory of control, it is expected that all

individuals will implement primary and secondary control strategies (Heckhausen &

Schulz, 1995). However, the ratio of these strategies is expected to be influenced by

the controllability of the situation. If the situation is appraised as being controllable,

it is expected that people will try to change it using primary control. The situation is

amenable to change, and as such, it is expected that attempts to change the

environment using primary control would be successful.

250

If however the situation is appraised as being uncontrollable, it is expected

that people will attempt to change themselves using secondary control. If the person

tried to change the environment using primary control, they would be likely to

experience primary control failure. In order to avoid primary control failure

therefore, it is expected that people would rely on more secondary control. Hence, it

is proposed that the controllability of the situation is inversely related to the

probability of primary control failure, which in turn, influences the use of secondary

control strategies.

4.2.3 Empirical Studies Examining if the Controllability of a Situation

Influences the Use of Control Strategies

Although chapter 1 identified some studies that examined the amount of

general primary and secondary control reported by employees to handle general work

difficulties, no studies have been located which report the amount of primary and

secondary control people use in controllable and uncontrollable situations. One

study has examined the control strategies reported by people only in low-control

situations (i.e., HIV-positive men in prison). According to the discrimination model,

it would be expected that these men would rely on more secondary control than

primary control. This was not the case however, as Thompson et al., (1996)

demonstrated that the men reported slightly more primary control (M = 48.5%SM)

than secondary control (M = 45%SM).

Although there are no studies examining the use of the control strategies in

controllable and uncontrollable situations, there are some studies that have been

251

conducted on coping strategies. These studies generally examine the amount of

problem-focussed and emotion-focussed coping strategies reported by people in

controllable and uncontrollable situations (e.g., Bowman & Stern, 1995; Folkman et

al., 1986; Forsythe & Compas, 1987; Terry & Hynes,1998; Valentiner, Holahan &

Moos, 1994; Vitaliano, DeWolfe, Maurio, Russo & Katon, 1990).

One study has specifically examined coping strategies at work. Bowman and

Stern (1995) asked nurses to describe two stressful events, one that they “could

control, could change or could do something about” and one that was “difficult to

control, that you had to accept or had to get used to.” Participants then rated the

controllability of the situation, and completed Lazarus and Folkman’s (1984) coping

scale. Unfortunately however, Bowman and Stern (1995) did not examine the mean

coping strategies separately for each stressful situation. Instead they aggregated

them, providing mean scores for avoidance coping, problem-reappraisal coping, and

problem solving coping. It must be noted however that even if the means were

provided, the validity of the research design is questioned. There are problem with

using the terms “change” and “do something about” for controllable situations and

“accept” and “get used to” for uncontrollable situations. These terms may bias the

employees to respond in ways that are consistent with the discrimination model. By

their nature, situations that have been changed are those where primary control

strategies have been used, and situations that have been accepted are those where

secondary control strategies have been used.

Other studies have examined coping strategies in controllable and

uncontrollable non-work situations. These studies have provided somewhat mixed

252

support. For example, Forsythe and Compas (1987) demonstrated that people used

more problem-focussed coping for events appraised as controllable (M = 13.82) than

uncontrollable (M = 10.81), however there were no differences in emotion-focussed

coping. Furthermore, Folkman et al., (1986) found that married couples tend to use

more problem-focussed coping in situations perceived as changeable, and more

emotion-focussed coping in situations perceived as having to be accepted.

However, Valentiner et al., (1994) demonstrated that college students did not report

more problem-focussed type coping (M = 55.39) than emotion-focussed type coping

(M = 57.78) in a controllable event. Furthermore, they did not report more emotion-

focussed type coping (M = 53.50) than problem-focussed type coping in an

uncontrollable event (M = 54.34).

Instead of reporting the average level of coping strategies in controllable and

uncontrollable situations, other studies have examined the correlations between

perceived control and coping strategies. In this case, it would be expected that

perceived control would be positively correlated with problem-focussed coping and

negatively correlated with emotion-focussed coping. Overall however, these studies

have tended to be inconsistent.

For example, Osowiecki and Compas (1999) demonstrated that problem-

focussed and emotion-focussed coping were not significantly related to perceived

control. A similar result was found by Conway and Terry (1992) where problem-

focussed coping, self-denigration and escapism did not correlate with the

controllability of an event. However, Park, Folkman and Bostrom (2001)

demonstrated that controllability appraisal was positively correlated with problem-

253

focussed strategies (i.e., planful problem solving, r = 0.23), and negatively correlated

with emotion-focussed strategies (distancing, r = -0.29).

When examining these studies on coping strategies, the flaws in the

conceptualisation of problem-focussed and emotion-focussed coping must be

considered. As discussed in chapter 1, the theory underlying problem-focussed and

emotion-focussed coping and the questionnaire designed to assess these strategies

(i.e., Ways of Coping Questionnaire; WCQ) has methodological limitations

(Edwards & O’Neill, 1998). The most concerning problem is that there is overlap

among the coping dimensions, where some problem-focussed coping strategies

resemble emotion-focussed coping strategies (Edwards & O’Neill, 1998).

The conceptualisation of primary and secondary control is superior to

problem-focussed and emotion-focussed coping because it maintains the distinction

between changing the environment (i.e., primary control), and changing the self (i.e.,

secondary control). Thus, the proposal that the controllability of the situation

influences the amount of control strategies will be tested in this study.

4.2.3.1 Summary

Based on the proposals of the life span theory of control, it is expected that

when employees have a controllable difficulty, they use more primary control than

secondary control. When they have an uncontrollable difficulty, it is expected that

they will attempt to avoid primary control failure, and thus report more secondary

control than primary control. Although no studies have examined the control

254

strategies reported by people in controllable and uncontrollable situations, studies

using coping strategies have offered, at best, mixed support.

4.2.4 Examining how Controllability Influences the Adaptiveness of the

Control Strategies

Based on the discrimination model (Thompson et al., 1998), it is expected

that primary control is the most adaptive strategy in controllable situations, and that

secondary control is the most adaptive strategy in uncontrollable situations. This is

consistent with the life span theory of control which proposes that although primary

control is the more adaptive strategy, primary control failure can have negative

consequences (Heckhausen et al., 1997). It is postulated that the controllability of

the situation is inversely related to the probability of primary control failure.

As discussed in chapter 1, only a few studies have examined the most

adaptive control strategies in low-control situations (Thompson et al., 1996; 1994;

1993; 1998). These studies suggest that primary control strategies are more adaptive

in controllable and uncontrollable situations, however these studies were criticised

for their measurement of perceived control, and primary and secondary control

strategies.

It must be noted however that a similar hypothesis was being developed in

the coping literature. Several researchers have tested this proposition, referred to as

the “goodness of fit” hypothesis (Carver, Scheier & Weintraub, 1989; Conway &

Terry, 1992; Folkman et al., 1986; Roberts, 1995; Vitaliano et al., 1990). They

recognise that “it is not the coping response per se that is the key to reduce emotional

255

distress, but rather how well the coping strategy fits the perceived situation”

(Osowiecki & Compas, 1998, p. 485).

Although there is significant overlap among the coping studies and control

strategy studies, researchers are yet to integrate the results. As there are few such

studies, this integration is essential to gain a greater understanding about whether the

controllability of the situation influences the use and adaptiveness of the

control/coping strategies in that situation. It must be noted however that the majority

of studies have relied on Lazarus and Folkman’s (1984) problem-focussed coping

and emotion-focussed coping, which was criticised in chapter 1.

4.2.4.1 Integrating Empirical Studies on the Discrimination Model and the

Goodness of Fit Model

Empirical studies examining whether the controllability of the situation

influences the coping/control scales used in that situation provide mixed support.

Generally, these studies demonstrate the primary control-type strategies are more

adaptive than secondary control-type strategies in controllable situations. However,

some of the studies fail to demonstrate that secondary control-type strategies are

more adaptive than primary control-type strategies in uncontrollable situations (e.g.,

Bowman & Stern, 1995; Conway & Terry, 1992; Osowieki & Compas, 1998, 1999;

Park, Folkman & Bostrom, 2001; Vitaliano et al., 1990).

For example, Thompson et al’s., (1996) study on HIV-positive men

demonstrated that primary control was negatively related to distress and secondary

control was positively related to distress. However, other studies have demonstrated

256

that secondary control is more adaptive than primary control. For example, Terry

and Hynes (1998) demonstrated that for women coping with in vitro fertilization,

problem management coping (i.e., trying to solve the problem) was related to more

distress. Secondary control-type strategies (i.e., problem appraisal, and emotional

approach) were related to less distress.

Integration of the studies testing the goodness of fit model and the

discrimination model highlights the inconsistencies in the area. Generally, it appears

as though primary control is adaptive in controllable situations, but that secondary

control may not be the most adaptive strategy in uncontrollable situations. These

results may be limited, as problems have been identified in the design of the studies.

4.2.4.2 Research Design

Researchers have typically relied on two major types of designs to measure

the goodness of fit hypothesis and the discrimination model. The first design

assesses how people handle one stressful situation (e.g., Carver et al., 1989; Conway

& Terry, 1992; Folkman et al., 1986; Roberts, 1995; Vitaliano et al., 1990).

Typically, the person reports on the most stressful encounter they had during the

previous week, indicating how much they could control the situation, and what they

did to handle the situation. The researcher then correlates the controllability of the

situation with the coping strategies. This measure is problematic however as the

respondent chooses whether they report a controllable or uncontrollable situation and

the researcher cannot influence this variable.

257

The second type of design, much like studies one and two, examines the

control strategies used by people in low-control situations. Researchers have studied

various groups such as cancer patients (Osowiecki & Compas, 1998, 1999),

HIV- positive men (Thompson et al., 1996; Thompson et al., 1994), people

experiencing age-related physical changes (Thompson et al., 1998), and children

experiencing homesickness (Thurber & Weisz, 1997). These studies generally assess

how much perceived control the person has over the situation (e.g., cancer) and then

assesses which control/coping strategies they used to handle the situation

(e.g., Osowiecki & Compas, 1998). This design is criticised however as it is

inconsistent with the proposed models. Both the discrimination model and the

goodness of fit model refer to one situation. In order to test whether the

controllability of a situation influences the control strategies used in that situation,

the scale needs to be more specific.

4.2.4.3 Summary

Research examining the discrimination model and the goodness of fit

hypothesis is equivocal. It appears however that the majority of studies find that

primary control is adaptive in controllable situations, but less demonstrate that

secondary control is adaptive in uncontrollable situations. The validity of these

findings are questioned however, as the research designs are criticised. In order to

accurately test the discrimination model, a more specific scale is required which

assesses how people handle controllable and uncontrollable difficulties.

258

4.2.5 Developing a Situation Specific Primary and Secondary Control Scale

In order to test the proposal that the controllability of a situation influences

the use and adaptiveness of the control strategies in that situation, a situation specific

primary and secondary control scale is developed. This scale overcomes many of the

limitations identified in previous scales, as it: a) assesses how people react in

controllable and uncontrollable situations; b) can be used by workers in any

occupation; and c) contains few items.

The Situation Specific Primary and Secondary Control Scale (Maher &

Cummins (2002) is an extension of the 4th edition of the Primary and Secondary

Control Scale (Maher et al., 2001). The scale includes four primary control

strategies and 12 secondary control strategies. Respondents are asked to indicate

how often, during the past week, they have used various strategies when facing a

difficulty at work. The major change made to this scale is that rather than thinking

about any difficulty at work, the respondents now think about one difficulty that they

can control and one difficulty that they cannot control.

The major issue in designing a scale to measure low-control and high-control

situations was deciding on the wording of the situation. Only one study was located

that tested respondents in controllable situation and uncontrollable situations at work

(Bowman & Stern, 1995). For the controllable situation, the employee was told to

consider a situation that they “could control, could change, or do something about.”

For the uncontrollable situation, the employee was told to consider a situation that

was “difficult to control, that you had to accept or get used to.”

259

Other researchers examining one situation have asked “how much control do

you have over” (Thompson et al., 1994; Thurber & Weisz, 1997), “how much

influence do you have over” (Conway & Terry, 1992) and “how much can you

change” (Carver et al., 1989, Folkman et al., 1986).

As mentioned previously, there are problem with using the terms “change”,

“do something about”, and “influence” for controllable situations and “accept” and

“get used to” for uncontrollable situations. These terms may bias the employees to

respond in ways that are consistent with the discrimination model.

A viable alternative that does not imply that the situation has been changed is

“control.” The employees could be asked to think of a difficulty that they can

control and a difficulty that they cannot control. Control is superior to the other

constructs, as it does not bias the respondents to nominate primary control strategies

in a high-control situation.

Changes were also made to the primary control items. The primary control

items were designed to be general, assessing whether the person looks for different

ways to overcome difficulties, persists, puts in effort, and works out how to remove

obstacles. Closer examination of these items however revealed several problems.

For example, the item, “looked for different ways to overcome it” may not actually

represent primary control. A person who looks for different ways to overcome their

difficulties does not necessarily attempt to change the environment to suit their

needs. They may think about the different ways, decide that they are all too risky,

and then resort to secondary control strategies. To demonstrate primary control, a

person needs to do more than just think of different ways to overcome the difficulty,

260

rather they need to act on their environment. As such, this item was changed to a

strategy identified by Folkman and Lazarus (1980) termed “choose and act on a

potential solution.”

The item, “work hard to overcome it” is also criticised as it implies that the

strategy must be successful for it to represent primary control. According to this

item, the person must not only work harder, but must also overcome the difficulty.

Primary control does not necessarily involve overcoming the problem, only that the

person attempts to change the environment. As such, this item was changed to

“work harder.”

Another problematic item is “work out how to remove obstacles.” This item,

adapted from Heckhausen et al’s., (1997) scale, refers to goals rather than

difficulties. This item is appropriate for goals as, if a goal is not obtained, there must

be obstacles in the way of it. However, there may not necessarily be obstacles in the

way of a difficulty. This item does not appear to fit with the control scale, which

focuses on difficulties at work, and as such was deleted from the scale.

Based on these analyses, the following three items were included in the

primary control scale; “choose and act on a potential solution”; “keep trying”; and

“work harder.” One extra item was developed to account for the fact that other

people, such as management staff, often control many problems in the workplace. A

major way that a person may change difficulties in the workplace is through

discussions or confrontations. An item developed in Latack’s (1986) coping scale to

measure this is “discussing the problem with the people involved.” As discussing the

261

problem may not necessarily mean that the environment is changed, the item was

revised to “discuss solutions with the people involved.”

The addition of this item led to a review of the “support” item in the

secondary control scale to ensure that the two were different. Indeed, it is difficult to

separate support as a primary control strategy and support as a secondary control

strategy. The main distinction between the two, however, is that support as a

primary control strategy involves the person changing the environment, whereas

support as a secondary control strategy involves the person changing themselves to

accept the problem.

The secondary control strategy of support was measured by the item “told

someone about it” in previous editions of the scale. In order to ensure that this item

is distinct from the primary control strategy, it was changed to clearly demonstrate

that it involves changing the self (i.e., “I told someone about the difficulty to make

me feel better”).

One final change to the primary and secondary control scale concerns the

scoring. In study two, the highest score for primary and secondary control was

recorded. The items were not aggregated because the resulting score was deemed to

be unrepresentative of the control strategies. For example, using an aggregated

score, a person who reported that they use one secondary control strategy every time

(10), and reported never (0) for the remaining strategies would receive a low score.

In order to demonstrate that this person is using one secondary control strategy all

the time, the highest score for secondary control was recorded (10), and the person

received a high score. However, as demonstrated in study two, using the highest

262

score does not appear to be the answer. This method did not differentiate the

respondents, with 76% of the subjects reporting a level of primary control between

77%SM and 100%SM, and 84% of the subjects reporting a level of secondary

control between 77%SM and 100%SM. This range is concerning, suggesting that

there may have been a ceiling effect. As a result, the aggregated scoring procedure

implemented in study one will be used in the current study. However, it must be

noted that the aggregated scoring method, although used by the majority of

researchers in the field, is biased towards people who use a greater variety of

strategies.

4.2.5.1 Summary

In order to test the proposal that the controllability of a situation influences

the use and adaptiveness of the control strategies in that situation, a situation specific

primary and secondary control scale was developed. The major change made to the

scale is that rather than thinking about any difficulty at work, the employees are now

required to think about one difficulty that they can control and one difficulty that

they cannot control. Changes were also made to the wording of the primary control

items and the scoring procedure.

4.2.6 Examining the Moderating Role of Primary and Secondary Control

Strategies

In addition to examining the amount and adaptiveness of the control

strategies in controllable and uncontrollable situations, this study also tests whether

263

the control strategies moderate the relationship between work difficulties and job

satisfaction. Although no other studies have examined the moderating role of the

control strategies, several researchers in the coping literature have suggested that it is

not the stressor that predicts job satisfaction, but rather how the person deals with the

stressor (Aldwin & Revenson, 1987, Ashford, 1988; Parkes, 1990, 1994; Perrewe &

Zellars, 1999; Osipow, Doty & Spokane, 1985).

For example, Ashford (1988) demonstrated that coping moderated the effect

of organisation transitions on job stress, where employees who shared emotions

experienced less stress after organisational change. Parkes (1990) also demonstrated

that coping moderated the effect of work demands on general health, however they

found that employees who reported more direct coping (i.e., problem-focussed

coping) had better health. As only a few studies have examined this proposal in the

workplace, and as they have relied on varying measures of coping, this proposal will

be examined further.

It is proposed that primary control strategies are only useful in reducing stress

when the situation is controllable. In these situations, primary control can be

implemented successfully, and the negative effects of the difficulty can be reduced.

When the situation is uncontrollable however, secondary control strategies may be

useful in helping the person adjust to the situation. If they use primary control

strategies, they are likely to experience primary control failure, which may increase

their stress. However, if they use secondary control, they can reduce their stress by

accepting the situation. These exploratory proposals will be tested.

264

4.2.7 Examining the Moderating Role of Social Support at Work

In addition to primary and secondary control, social support at work may

moderate the relationship between controllable and uncontrollable work difficulties

and job satisfaction. In study two, the moderating roles of co-worker and supervisor

support were examined. This study will also examine the major types of social

support, namely instrumental and emotional support. These two types of social

support are expected to play different roles (Ducharme & Martin, 2000; Wong,

Cheuk & Rosen, 2000). Instrumental support is expected to buffer work difficulties

because it helps workers to cope effectively with problems, whereas emotional

support is not expected to buffer work difficulties as it does not directly alter the

stressor (Wong et al., 2000). Some support has been provided for these proposals,

where instrumental supervisor support, but not emotional supervisor support, has

been shown to buffer the effects of job stress on job satisfaction (Wong et al., 2000).

A more specific explanation is developed for this study, where it is proposed

that both instrumental and emotional support buffer the effects of work difficulties.

Specifically, it is proposed that instrumental support buffers the effects of

controllable difficulties and emotional support buffers the effects of uncontrollable

difficulties. Instrumental support is useful if the difficulty is controllable as other

people may help the person to overcome the problem. However, when the difficulty

is uncontrollable, instrumental social support may not be useful as there is nothing

that can be done to overcome the difficulty. Rather, in these situations, emotional

social support may help the person to accept these difficulties. As this study is

examining both controllable and uncontrollable difficulties, it must be considered

265

whether emotional and instrumental social support moderate both types of

difficulties.

4.2.7.1 Summary

This study proposes that the control strategies and social support at work

moderate the relationship between work difficulties and job satisfaction.

Specifically, it is expected that primary control and instrumental support moderate

the relationship between controllable work difficulties and job satisfaction.

Furthermore, it is expected that secondary control and emotional support moderate

the relationship between uncontrollable work difficulties and job satisfaction.

266

4.3 Revised Model of Job Satisfaction

This study aims to test the model of job satisfaction presented in Figure 8,

which is fundamentally different to the previous two models. The model proposes

that employees experience controllable and uncontrollable difficulties, which are

negatively related to job satisfaction.

In response to these difficulties, employees can implement primary and

secondary control strategies. It is expected that the ratio of control strategies will

vary depending on whether the difficulty is controllable or uncontrollable. For

controllable difficulties, it is expected that workers will rely on primary more than

secondary control. For uncontrollable difficulties, it is expected that workers will

rely on secondary more than primary control.

The adaptiveness of the control strategies is also expected to vary depending

on whether the difficulty is controllable or uncontrollable. For controllable

difficulties, it is expected that primary control will be more adaptive and therefore

more positively related to job satisfaction than secondary control. As the situation is

controllable, it is likely that a person can change it.

For uncontrollable difficulties, secondary control will be more positively

related to job satisfaction than primary control. As the situation is uncontrollable, it

is unlikely that a person can change the situation using primary control, and thus

secondary control would be more adaptive than primary control failure.

Both types of difficulties (controllable and uncontrollable) are expected to be

directly and indirectly related to job satisfaction. Employees who report more

difficulties at work are expected to report lower job satisfaction. However two

267

variables that may moderate the relationship between work difficulties and job

satisfaction are primary and secondary control and social support at work. These

moderation effects are represented by the interaction terms in Figure 8.

In regard to primary and secondary control, it is proposed that primary

control strategies are useful in reducing the effects of work difficulties when the

situation is controllable. When the situation is uncontrollable, secondary control

strategies may reduce the effects of work difficulties on job satisfaction.

In regard to social support at work, it is expected that instrumental support

will buffer the effects of controllable difficulties on job satisfaction. Emotional

support is expected to buffer the effects of uncontrollable difficulties on job

satisfaction.

In addition to work difficulties, job autonomy and life satisfaction are

expected to directly predict job satisfaction. Both of these relationships have been

demonstrated in studies one and two.

In summary, controllable and uncontrollable work difficulties, the primary

and secondary control strategies used to handle such difficulties, and social support

at work, are expected to determine job satisfaction. Job satisfaction is, in turn,

expected to influence, and be influenced by, life satisfaction.

268

Figure 8- Revised Model of Job Satisfaction

Controllable Difficulties

Uncontrollable Difficulties

Instrumental Support

Emotional Support

Job Autonomy

Job Satisfaction

Secondary Control

Primary Control

Control Diff x Instrumental Support

Uncontrol Diff x EmotionalSupport

ControllableDiff x Primary Control

Uncontrollable Diff x Secondary Control

Life Satisfaction

Secondary Control

269

Primary Control

270

4.4 Hypotheses

1) When workers face controllable difficulties, they are expected to use primary

control more than secondary control. Conversely, when workers face uncontrollable

difficulties, they are expected to use secondary control more than primary control.

Workers should match their control strategies to the controllability of the

situation. When the situation is controllable, the workers will be likely to change it

using primary control. When the situation is uncontrollable, the workers will be

likely to accept the situation using secondary control.

2) When workers face controllable difficulties, primary control is expected to be

more positively related to job satisfaction than secondary control. Conversely, when

workers face uncontrollable difficulties, secondary control is expected to be more

positively related to job satisfaction than primary control.

This hypothesis is based on the discrimination model, which proposes that

primary control is more adaptive in controllable situations, and that secondary

control is more adaptive in uncontrollable situations.

3) Primary control will moderate the effect of controllable difficulties on job

satisfaction and secondary control will moderate the effect of uncontrollable

difficulties on job satisfaction.

This hypothesis proposes that primary control strategies are useful in

reducing the influence of work difficulties on job satisfaction when the situation is

271

controllable. When the situation is uncontrollable however, secondary control

strategies are expected to reduce the effects of work difficulties on job satisfaction.

4) Instrumental social support will moderate the effects of controllable work


When the difficulty is controllable, it is proposed that other people may help the

person to overcome the problem.

5) Emotional social support will moderate the effects of uncontrollable work


Emotional social support is expected to help reduce the influence of uncontrollable

difficulties on job satisfaction. In these cases, instrumental social support may not be

useful as the situation cannot be overcome, however emotional social support may

help reduce the severity of these difficulties.

6) Work difficulties, the control strategies used to handle work difficulties, social

support at work, job autonomy and life satisfaction, will predict job satisfaction.

These variables are expected to be major predictors of job satisfaction, as

demonstrated in Figure 8.

272

4.5 Method

4.5.1 Participants

The sample consisted of 214 general employees, obtained from a database

and through convenience sampling. The age of the participants ranged from 21-73

years, with the average being 44.78 years (SD=15.18). The demographic

characteristics of the sample are displayed in Table 36.

Table 36- Demographic Characteristics of the Sample

Variable % sampleGender

Male 47.7Female 47.7

OccupationProfessional 46.7Business 13.1Trade 10.7Clerical 13.1Retail 6.1Labourer 1.9Other 3.3

Hours worked per week1-20 13.621-30 14.031-40 27.641-50 27.151-60 12.161+ 4.7

Bolded values indicate the largest proportion.

273

4.5.2 Materials

All of the respondents received a plain language statement (refer to Appendix

N) and a questionnaire. The questionnaire consisted of scales measuring controllable

and uncontrollable difficulties, primary and secondary control for such difficulties,

job autonomy, job satisfaction, life satisfaction and social support at work.

4.5.2.1 Controllable and Uncontrollable Difficulties

Controllable and uncontrollable difficulties were measured in the Situation

Specific Primary and Secondary Control Scale (Maher & Cummins, 2002; refer to

Appendix O). Although the most direct way to measure this variable would be to

ask, “how often do you experience difficulties that you can control/cannot control”,

this item was deemed to be too cognitively taxing and prone to errors. Rather the

employees were given a list of potential difficulties that they may face at work such

as supervisors, co-workers, kind of work, pay, work-place rules, promotion, time

management and others. They indicated which difficulties they experienced that

they could control.

In order to determine how frequently they experienced controllable

difficulties, they were asked to consider the difficulty that they experienced most

often and could control, and indicate how often they experienced it. This process

was repeated for uncontrollable difficulties.

Although this question only refers to one difficulty, it was deemed to be the

best method. One alternative is to ask them on average how often they experience

the difficulties they ticked. However, it is extremely difficult for employees to

274

mentally calculate how often they experience each difficulty they ticked and then

calculate the average. Another alternative would be to get them to indicate how

often they experienced each difficulty. This was problematic however, as there are

an unlimited number of work difficulties, particularly as this study is relying on a

general sample of employees.

4.5.2.2 Primary Control and Secondary Control

The Situation Specific Primary and Secondary Control Scale (Maher &

Cummins, 2002; refer to Appendix O) was used in this study. The main difference

between this scale and earlier scales is that the respondents now indicate which

control strategies they use for controllable difficulties and uncontrollable difficulties.

There are four primary control, and 12 secondary control strategies from which to

choose. There is also the option to list other strategies that they use. Each strategy is

rated on a five-point scale ranging from 0 (never) to 4 (always).


Job autonomy was measured by the job autonomy items in the Job Diagnostic

Survey (Hackman & Oldham, 1975). This scale consists of three items that assess

overall perceived job autonomy, such as “In my job, I can decide on my own how to

go about doing my work” (refer to Appendix B). The items were rated on a 10-point

scale ranging from 1 (do not agree at all) to 10 (agree completely). As discussed in

study two, the psychometric statistics are adequate.

275


Job satisfaction was measured by a one-item measure. The item “taking into

consideration all the things about your job, how satisfied are you with it” was rated

on an 11-point scale. Although internal consistency cannot be established with a

single-item measure, this measure of job satisfaction has been shown to correlate

with other measures of job satisfaction, where r = 0.63 (Wanous et al., 1997).

In studies one and two, a facet measure of job satisfaction was also used. As

this facet measure was only used to gain insight into the particular occupational

groups, it is not necessary in this study which is relying on a general sample of

employees.


Life satisfaction was measured by the Personal Well-being Index developed

recently by Cummins et al., (2001). This scale attempts to overcome some

methodological problems that were identified in the Comprehensive Quality of Life

Scale, which was used in studies one and two. Most of these problems involve the

objective scale, or the importance scale, rather than the satisfaction scale. However,

some problems were identified with the seven domains of life satisfaction, such as

the domain of emotional well-being or happiness refers to an affective state rather

than a domain of life (Cummins, 2002). Furthermore, the wording of some of the

items were not optimal (Cummins, 2002). As such, rather than happiness, future

security is included as a domain of life satisfaction. Thus, the Personal Well-being

Index consists of seven domains of satisfaction which are rated on an 11-point scale

276

(refer to Appendix P).


In study two, social support at work was measured by Karasek and Theorell’s

(1990) scale. This scale consisted of both instrumental and emotional support,

however the items were designed to be summed to provide an overall support score.

These items were examined more closely in this study to ensure that they were

appropriate for a range of occupations.

Some problems were identified with the instrumental support scale. For

example, the item, “my supervisor creates a good teamwork environment for me”

may not measure instrumental support. It is not necessary that an employees works

in a teamwork environment for them to receive instrumental support. To provide

instrumental support, the employer only needs to offer some material assistance. A

further problem with the scale concerns the item “my co-workers are competent.”

This item, although intended to measure instrumental support, only assesses the co-

workers competence. Co-workers may indeed be competent, however this does not

mean that they offer assistance when required. These problems, although they were

not recognised in study two, will be rectified in this study.

A review of the other major measures of social support at work was

undertaken. This review demonstrated that although there are many scales that claim

to measure social support at work, few adequately measure emotional and

instrumental support. Some researchers rely on a one item measure, such as “how

true is it that your supervisors are warm/friendly when you have problems” and “how

277

true is it that your supervisor helps you complete a given task” (Himle & Jayaratne,

1991; Wong et al., 2000). These scales are criticised however for failing to capture

the different ways that emotional or instrumental support can be offered.

Some studies have relied on scales which only focus on emotional support

and fail to measure instrumental support (Dollard et al., 2000, Rodriguez, Bravo,

Peiro & Schaufeli, 2001). Others do not claim to measure either component (Caplan

et al., 1975; Van der Doef & Maes, 1999), whilst others still claim to measure five

types of social support (Unden, 1996).

One exploratory scale recently developed by Ducharme and Martin (2000)

does not appear to suffer from any of these problems. Their scale, developed only to

assess co-worker support, includes five items assessing instrumental support and five

items assessing emotional support. Consistently, a factor analysis on the scale

demonstrated that two factors emerged. To make this scale appropriate for the

current study, it was posited that the co-worker items could also be applied to

supervisors. If this were done, the scale would consist of 20 items. As this may be

unnecessarily long, some items were deleted. Specifically, only the three highest

loading items were selected each for instrumental and emotional support. For

emotional support, these were “your co-workers really care about you”, “you feel

close to your co-workers” and “your co-workers take a personal interest in you.”

These three items were also changed to be applicable to supervisors.

For instrumental support, the items were “your co-workers would fill in while

you’re absent”, “your co-workers are helpful in getting the job done” and “your

278

co-workers give useful advice on job problems.” These three items also needed to be

applicable to supervisors. As one of them was not (i.e., “your supervisors would fill

in while you were absent”), the next highest loading item in the factor analysis was

selected. This was “your co-workers assist with unusual work problems.” The

resulting scale is a six-item scale for co-worker support and a six-item scale for

supervisor support, that both assess instrumental and emotional support (refer to

Appendix Q).

4.5.3 Procedure

Ethics approval was obtained from Deakin University. The majority of the

sample was obtained from a database developed by Australian Unity and Deakin

University. This database contains information for 900 people that have been

randomly selected from the population, and have agreed to participate in a survey.

The employment status of these people was unknown, and as such two

questionnaires were sent to them, one if they were employed, and one if they were

unemployed. A total of 250 (27%) questionnaires were returned however only 130

(14.44%) of these were completed by people that were employed. The remainder of

the sample was obtained through convenience, and snowballing.

279

4.6 Results


The data set was examined for missing values, outliers, normality and

linearity. There were very few missing values for measures of life satisfaction, job

satisfaction, job autonomy, job demands, and co-worker support (i.e., < 5%). There

was a higher rate of missing values for supervisor support (19%) and primary and

secondary control (6%, 11%, respectively). The treatment of these values depended

on their context. If the participant had completed the majority of the scale, the

missing values were replaced with the group mean. If however, the person had failed

to complete any of the scale, they were excluded from analyses using that scale.

Overall, this treatment resulted in less than 5% of the missing values being replaced

with the group mean.

Univariate outliers were identified in the measures of life satisfaction

(2 cases), job satisfaction (3 cases), job autonomy (1 case), co-worker support

(4 cases), and primary and secondary control (3 cases). These values were re-coded

to lie within three standard deviations of the mean.

Many of the scales were negatively skewed, where the skew/standard

error > 3. These include life satisfaction (-4.44), job satisfaction (-5.09), job

autonomy (-7.07), and co-worker and supervisor support (-4.56, -5.15, respectively).

As transformations are not recommended for variables that are mildly and naturally

skewed (Tabachnick & Fidell, 1997), these variables were not transformed.

Reasonably linear relationships were evident among the variables.

280


Table 37 contains the means and standard deviations of the major variables.

This table demonstrates that employees are using primary more than secondary

control in controllable and uncontrollable situations. Other interesting findings are

that employees report that their co-workers offer more instrumental and emotional

support than their employers. Additionally, the level of life satisfaction is in the

expected range.

Table 38 displays the correlations among the major variables. Several

variables correlate well with job satisfaction including job autonomy, life

satisfaction, social support at work, and difficulties at work. Strong correlations are

also observed among primary control in controllable situations and primary control

in uncontrollable situations. Similarly, secondary control in controllable situations is

strongly correlated to secondary control in uncontrollable situations.

281

Table 37- Means and Standard Deviations of the Major Variables

Variable M SDJob Satisfaction 71.67 18.41Job Autonomy 78.18 20.34Life Satisfaction-domain 73.68 13.76Co-worker emotional support 74.63 20.81Co-worker instrumental support 78.03 21.30Supervisor emotional support 65.86 26.66Supervisor instrumental support 70.55 26.32Frequency of controllable difficulties 58.70 20.33Frequency of uncontrollable difficulties 57.07 22.25Primary control for controllable difficulty 71.98 13.05Secondary control for controllable difficulty 53.74 12.12Primary control for uncontrollable difficulty 65.56 16.73Secondary control for uncontrollable difficulty 53.10 12.68

All scores have been converted to a percentage of scale maximum (%SM) which ranges from 0-100. The formula is (mean score for the original domain-1) x 100/ (number of scale points –1)

Table 38- Inter-Correlations among Major Variables

JS JA Cdif Udif PcC ScC PcU ScU LS SupJSJA 0.57Cdif -0.37 -0.22Udif -0.35 -0.32 0.42PcC 0.27 0.34 -0.13 -0.14ScC -0.12 -0.25 -0.03 0.10 -0.03PcU 0.23 0.18 -0.12 -0.11 0.69 0.06ScU -0.04 -0.16 -0.05 0.10 0.07 0.64 0.20LS 0.46 0.34 -0.14 -0.13 0.13 -0.13 0.19Sup 0.37 0.40 -0.21 -0.24 0.10 0.04 0.08 0.08 0.26Cow 0.48 0.41 -0.14 -0.14 0.07 -0.06 0.09 0.04 0.31 0.57

Bolded items p<0.01 JS - job satisfaction; JA - job autonomy; Cdif - controllable difficulties; Udif -uncontrollable difficulties; PcC - primary control for controllable difficulties; ScC - secondary control for controllable difficulties; PcU - primary control for uncontrollable difficulties; ScU - secondary control for uncontrollable difficulties; LS - life satisfaction; Sup - supervisor support; Cow - co-worker support.

282

4.6.3 Factor Analyses

Factor analyses were conducted on the two exploratory scales in the study

measuring primary and secondary control strategies and social support at work.

4.6.4 Primary and Secondary Control Scale

Two factor analyses were required to examine the primary and secondary

control items for a controllable difficulty and an uncontrollable difficulty. For both

of these analyses, the assumptions were met where Bartlett’s test of sphericity was

significant, and Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy

exceeded 0.60.

For controllable difficulties, a principle component analysis with direct

oblimin rotation yielded five factors. These five factors accounted for 53.8% of the

variance, however only 6 of the 16 items loaded on only one factor, and as such, this

five-factor solution could not be interpreted. The analysis was repeated using a four-

factor solution. Ten items loaded on only one factor, and the primary control items

loaded on a separate factor to the secondary control items. However, there was no

pattern to the secondary control items that loaded on the remaining three factors. In

response to this, a three-factor analysis was conducted. This analysis was much

clearer, demonstrating that all of the primary control items loaded on Factor 1, and

the secondary control items were mainly divided among Factor 2 and Factor 3.

As demonstrated in Table 39, Factor 1 consists of all four primary control

items. Two of the secondary control items negatively loaded on this factor as well,

and as such, they will be excluded from the analyses. Factors 2 and 3 contain the

283

remaining secondary control items. The items that loaded on Factor 2 were support,

vicarious, past success, behavioural avoidance and positive re-interpretation, whilst

the items that loaded on Factor 3 were predictive-negative, wisdom, and attribution.

There is a theoretical distinction among these items that was introduced in chapter 3.

As demonstrated in Table 19, two functions of secondary control are posited, namely

self-protective and self-affirmative.

Self-protective secondary control strategies reduce the negative impact of the

situation, whilst self-affirmative secondary control strategies increase positive

feelings about the self. All of the items that loaded on Factor 2 involve

self-affirmation, whilst the items that loaded on Factor 3 involve self-protection. It

must be noted however that some strategies that were expected to load on the two

factors did not. The self-affirmative strategy of downward social comparison did not

load on Factor 2. Furthermore, the self-protective strategies of goal disengagement,

illusory optimism, and denial did not load on Factor 3. Despite this however, overall

the factor analysis supports the two types of secondary control.

284

Table 39- Factor Analysis of Primary and Secondary Control Item in

Controllable Situations

Item F1 F2 F3Pc1 Discuss solutions with the people involved. 0.56Pc4 Choose a solution and act on it. 0.70Pc7 Work harder. 0.42Pc10 Keep trying. 0.60Sc2 Think that the difficulty doesn’t matter. -0.61Sc3 Think that this difficulty will work out okay

in the end.Sc5 Think that I knew this difficulty would

happen.0.68

Sc6 Think that I can’t always get what I want. 0.73Sc8 Think that I am better off than many other

people.Sc9 Think that this difficulty is not my fault. 0.42Sc11 Tell someone about this difficulty to make me

feel better.0.54

Sc12 Think of the success of my family/friends. 0.61Sc13 Think about my success in other areas. 0.74Sc14 Do something different, like going for a walk. 0.66Sc15 Ignore this difficulty. -0.63Sc16 Look for something else that is positive in the

situation.0.65

Eigenvalues 2.40 2.25 1.46% Variance 15.01 14.04 9.12

Cumulative variance 15.01 29.05 36.17Items with loadings less than 0.30 are not shown.Self-protective secondary control items are bolded. Self-affirmative secondary control items are italicised.Factor 1- Primary Control, Factor 2- Self-affirmative, Factor 3- Self-protective

In addition to the controllable difficulties, a principal component analysis

with direct oblimin rotation was conducted on the strategies used for uncontrollable

difficulties. This analysis yielded six factors, however as only one item loaded on

one factor, a five-factor solution, and a four-factor solution were requested. Both

analyses could not be interpreted, as there was no pattern to the secondary control

285

items. As such, a three-factor solution was conducted. This solution was remarkably

similar to the analysis of controllable situations. As demonstrated in Table 40, all

four primary control items loaded on Factor 1, and the secondary control items

loaded on their respective types of secondary control (i.e., self-protective and

self-affirmative). As with the analysis for controllable situations however, there are

a few exceptions, where goal disengagement and denial loaded on Factor 1. Unlike

the controllable analysis, Factor 2 included the self-affirmative strategy of downward

social comparison, however it also included illusory optimism, which is a self-

protective strategy. Finally, Factor 3 was the same in both analyses where it

excluded goal disengagement, illusory optimism, and denial.

286

Table 40- Factor Analysis of Primary and Secondary Control Items in

Uncontrollable Situation

Item F1 F2 F3Pc1 Discuss solutions with the people involved. 0.66Pc4 Choose a solution and act on it. 0.70Pc7 Work harder. 0.59Pc10 Keep trying. 0.61Sc2 Think that the difficulty doesn’t matter. -0.68Sc3 Think that this difficulty will work out okay

in the end.0.41

Sc5 Think that I knew this difficulty would happen.

0.63

Sc6 Think that I can’t always get what I want. 0.73Sc8 Think that I am better off than many other

people.0.58

Sc9 Think that this difficulty is not my fault. 0.54Sc11 Tell someone about this difficulty to make me

feel better.0.39

Sc12 Think of the success of my family/friends. 0.57Sc13 Think about my success in other areas. 0.70Sc14 Do something different, like going for a walk. 0.63Sc15 Ignore this difficulty. -0.65Sc16 Look for something else that is positive in the

situation.0.66

Eigenvalues 2.90 2.29 1.60% Variance 18.16 14.33 10.02

Cumulative variance 18.16 32.49 42.51Items with loadings less than 0.30 are not shown.Self-protective secondary control items are bolded. Self-affirmative secondary control items are italicised.Factor 1- Primary Control, Factor 2- Self-affirmative, Factor 3- Self-protective

Commonalties among the two factor analyses were examined. This study

aims to compare the primary and secondary control strategies for controllable with

uncontrollable difficulties, and as such, the comparisons should be based on the same

items. For primary control, all four items loaded on Factor 1 in both analyses, and as

such they will all be included. For self-affirmative secondary control, the common

287

items were sc11, sc12, sc13, sc14, and sc16. As such, self-protective secondary

control will be measured by these items (i.e., support, vicarious, present success,

active avoidance, and positive re-interpretation). For self-protective secondary

control, the common items were sc5, sc6, and sc9, which are predictive-negative,

wisdom, and attribution. Table 41 demonstrates which self-protective and self-

affirmative strategies were included in the analyses.

Table 41-Secondary Control Items included in Analyses

Type of Strategy Strategy Item Current StudySelf-protective Attribution Self-protectiveSelf-protective Predictive-Negative Self-protectiveSelf-protective Wisdom Self-protectiveSelf-protective Goal Disengagement --Self-protective Illusory Optimism --Self-protective Denial --

Self-affirmative Support Self-affirmativeSelf-affirmative Vicarious Self-affirmativeSelf-affirmative Present Success Self-affirmativeSelf-affirmative Active Avoidance Self-affirmativeSelf-affirmative Positive Re-interpretation Self-affirmativeSelf-affirmative Downward Social Comparison --

4.6.5 Social Support at Work

A principal components factor analysis with direct oblimin rotation was

conducted on the social support scale to ensure that the items were measuring four

types of support, namely co-worker instrumental support, co-worker emotional

support, supervisor instrumental support and supervisor emotional support. The

assumptions were met where Bartlett’s test of sphericity was significant, and

Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy exceeded 0.60.

288

As demonstrated in Table 42, all of the supervisor items loaded on Factor 1,

and all of the co-worker items loaded on Factor 2. The instrumental support and the

emotional support items for supervisors and co-workers did not load on different

factors. As these factor loadings are very high, the scale should only be divided into

co-worker support and emotional support. However, as the hypotheses refer to the

specific types of support, the instrumental and emotional support items will remain

separate.

Table 42- Factor Analysis of the Social Support at Work Scale

Item F1 F2My co-workers really care about me. 0.82I feel close to my co-workers. 0.88My co-workers take a personal interest in me. 0.93My co-workers assist with unusual work problems. 0.88My co-workers are helpful in getting the job done. 0.91My co-workers give useful advice on job problems. 0.77My supervisor really cares about me. 0.86I feel close to my supervisor. 0.91My supervisor takes a personal interest in me. 0.88My supervisor assists with unusual work problems. 0.91My supervisor is helpful in getting the job done. 0.93My supervisor gives useful advice on job problems. 0.93

Eigenvalues 7.34 2.05% Variance 61.97 17.05

Cumulative % variance 61.97 79.02


289


In order to test the hypotheses, repeated measures analyses of variance were

conducted to compare the use of control strategies in different situations, and

multiple regression analyses were used to examine the adaptiveness of the control

strategies. Hierarchical regression analyses were conducted to examine the

moderating role of the control strategies and social support on the relationship

between work difficulties and job satisfaction. Finally, a standard multiple

regression analysis was conducted to examine the major predictors of job

satisfaction. As in study two, the alpha level was reduced to 0.01 in order to reduce

the risk of Type I errors.

4.7.1 Hypothesis One- Use of Control Strategies for Controllable and


Hypothesis one proposes that when workers face controllable difficulties,

they use more primary control than secondary control, and when workers face

uncontrollable difficulties, they use more secondary control than primary control.

Before examining this hypothesis, the types of difficulties that employees are

reporting as being controllable or uncontrollable are examined. As demonstrated in

Table 43, the most common controllable difficulties were with time management,

motivation and co-workers. The most uncontrollable difficulties were with pay,

amount of work, work-place rules, and promotion.

290

Table 43 - Controllable and Uncontrollable Difficulties Reported by Employees

Difficulty % yes and controllable

% yes and uncontrollable

Difficulties with supervisors 38.8 35.5Difficulties with co-workers 60.7 17.8Difficulties with kind of work 52.8 29.9Difficulties with pay 23.4 46.3Difficulties with work-place rules 30.4 44.4Difficulties with promotion 12.1 40.2Difficulties with time management 62.1 19.6Difficulties with motivation 62.6 11.2Difficulties with work times 49.5 23.4Difficulties with amount of work 42.1 44.9Other 4.7 4.2

To test whether workers use more primary control when they face

controllable difficulties and more secondary control when they face uncontrollable

difficulties, two repeated-measures analyses of variance were conducted. The

variables were normally distributed, and the homogeneity of variance was met.

Keppel (1991) proposes that the adequacy of the sphericity tests has been questioned

and as such, researchers should assume that sphericity does not hold, and make the

adjustments.

For the controllable difficulties, Mauchly’s test of sphericity was not violated.

However, Greenhouse-Geisser epsilon was greater than 0.75, and as such, the

degrees of freedom were calculated using the Huynh-Feldt epsilon. The difference

was significant, F (1.984, 400.712) = 99.64, p = 0.00. As demonstrated in Table 44,

employees reported significantly more primary control than affirmative secondary

control, F (1, 202) = 187.94, p = 0.00, and self-protective secondary control,

291

F (1, 202) = 32.27, p = 0.00. Employees did not report more protective secondary

control than affirmative secondary, F (1, 202) = 1.20, p = 0.27. Thus, consistent with

the first part of hypothesis one, employees reported more primary control than

secondary control for a controllable difficulty.

For the uncontrollable difficulties, Greenhouse-Geisser epsilon was greater

than 0.75, and as such, Huynh-Feldt epsilon was used. This demonstrated that the

difference was significant, F (2, 386) = 38.57, p = 0.00. As demonstrated in Table

44, employees reported more primary control than self-affirmative secondary control,

F (1, 193) = 80.27, p = 0.00, and self-protective secondary control,

F (1, 193) = 29.94, p = 0.00. Employees also reported more self-protective

secondary control than self-affirmative secondary control, F (1, 193) = 8.78,

p = 0.003. Overall however, inconsistent with the second part of hypothesis one,

employees reported more primary than secondary control for uncontrollable

difficulties.

Table 44- Employees Use of Primary and Secondary Control in Controllable

and Uncontrollable Situations

Situation Strategy M SDControllable Primary Control 71.98 13.05

Secondary Control- Self affirmation 52.88 16.70Secondary Control- Self protective 54.60 16.21

Uncontrollable Primary Control 64.56 16.73Secondary Control-Self-affirmation 50.74 16.76Secondary Control- Self-protective 55.45 16.90

292

To determine the most commonly used control strategies to handle

controllable and uncontrollable difficulties, the means for each individual strategy

were examined. As demonstrated in Table 45, the most commonly used primary

control strategy for controllable and uncontrollable difficulties is “keep trying” and

the least common strategy is “discuss solutions with the people involved.” The most

common secondary control strategy for controllable and uncontrollable difficulties is

“think that I am better off than many other people.” The least common secondary

control strategy for controllable difficulties is “ignore this difficulty” and for

uncontrollable difficulties is “think that the difficulty doesn’t matter.”

293

Table 45- Means and Standard Deviations of Individual Control Strategies

Strategy Controllable UncontrollableM SD M SD

Discuss solutions with the people involved. (pc)

66.00 23.64 55.03 27.00

Choose a solution and act on it. (pc) 73.28 19.35 61.73 25.21Work harder. (pc) 67.49 20.56 63.92 24.40Keep trying. (pc) 77.58 18.52 77.58 18.52Think that the difficulty doesn’t matter. (sc) 33.99 26.72 30.15 26.19Think that this difficulty will work out okay inthe end. (sc)

53.33 27.78 49.36 25.63

Think that I knew this difficulty would happen. (sc)

55.67 25.19 55.67 25.19

Think that I can’t always get what I want. (sc) 51.42 24.83 51.42 24.83Think that I am better off than many other people. (sc)

67.49 22.98 67.49 22.98

Think that this difficulty is not my fault. (sc) 56.16 23.44 56.16 23.44Tell someone about this difficulty to make me feel better. (sc)

56.40 27.51 56.44 26.39

Think of the success of my family/friends. (sc) 46.43 27.30 41.23 28.00Think about my success in other areas. (sc) 54.56 23.48 53.22 24.53Do something different, like going for a walk.(sc)

45.81 26.69 44.20 28.14

Ignore this difficulty. (sc) 24.87 22.59 30.54 26.78Look for something else that is positive in theSituation. (sc)

61.21 23.08 58.63 22.61

Bolded strategies have highest frequencies

4.7.2 Hypothesis Two- Adaptiveness of the Control Strategies for

Controllable and Uncontrollable Difficulties

Hypothesis two proposes that primary control is more positively related to job

satisfaction than secondary control for controllable difficulties and that secondary

control is more positively related to job satisfaction than primary control for

uncontrollable difficulties. In order to test this hypothesis, two standard multiple

regression analyses were conducted.

294


were met for both analyses, and there was no evidence of multicollinearity. For the

controllable situation, R was significantly different from zero, R = 0.26,

F (3, 199) = 4.63, p = 0.004. As demonstrated in Table 46, only primary control

predicted job satisfaction, accounting for 5% of the variance. Thus, consistent with

the first part of hypothesis two, primary control was more positively correlated with

job satisfaction than secondary control for controllable difficulties.

For the uncontrollable difficulties, R was also significantly different from

zero, R = 0.29, F (3, 190) = 5.87, p = 0.001. As with the controllable difficulty, only

primary control predicted job satisfaction, accounting for almost 6% of the variance.

Thus, inconsistent with the second part of hypothesis two, primary control was more

positively correlated with job satisfaction than secondary control for uncontrollable

difficulties.

295

Table 46- Standard Multiple Regression Analysis Predicting Job Satisfaction

From Primary and Secondary Control

Difficulty Var JS PC SC-A B sr2(unique)Control

PC 0.24** 0.33 0.24 5.38**SC- A 0.03 0.12 0.007 0.006SC- P -0.01 -0.01 0.09 -0.08 -0.08

R =0.26** R2=0.07 AdjR2=0.05

UncontrolPC 0.26** 0.28 0.25 5.95**SC-A 0.13 0.18 0.11 0.01SC-P -0.08 0.05 0.14 -0.11 0.08

R =0.29** R2=0.09 AdjR2=0.07

**p<0.01Control - controllable difficulty; Uncontrol - uncontrollable difficulty; JS- job satisfaction; PC - primary control; SC-A - self-affirmative secondary control; SC-P- self-protective secondary control.

In order to determine the most adaptive control strategies for controllable and

uncontrollable difficulties, the correlations between each individual strategy and job

satisfaction are presented in Table 47. This table demonstrates that the most adaptive

primary control strategy for both types of difficulties is “keep trying” and the least

adaptive is “work harder.” The most adaptive secondary control strategy for a

controllable difficulty is “think that I am better off than many other people” and in an

uncontrollable situation is “look for something else that is positive in the situation.”

The least adaptive secondary control strategy in both situations is “ignore this

difficulty.”

296

Table 47- Correlations between Individual Control Strategies and Job

Satisfaction for Controllable and Uncontrollable Difficulties

Strategy Controllable Uncontrollable

Discuss solutions with the people involved. (pc) 0.11 0.19**Choose a solution and act on it. (pc) 0.23** 0.15Work harder. (pc) 0.09 0.14Keep trying. (pc) 0.24** 0.29**Think that the difficulty doesn’t matter. (sc) -0.22** -0.03Think that this difficulty will work out okay in the end. (sc)

0.13 0.12

Think that I knew this difficulty would happen.(sc)

-0.09 -0.08

Think that I can’t always get what I want. (sc) -0.02 0.006Think that I am better off than many other people. (sc)

0.22** 0.17

Think that this difficulty is not my fault. (sc) -0.10 -0.09Tell someone about this difficulty to make me feelbetter. (sc)

-0.08 -0.03

Think of the success of my family/friends. (sc) -0.04 0.004Think about my success in other areas. (sc) 0.12 0.19**Do something different, like going for a walk. (sc) -0.02 -0.009Ignore this difficulty. (sc) -0.25** -0.19Look for something else that is positive in the situation. (sc)

0.15 0.31**

**p<0.01

4.7.3 Hypothesis Three- The Moderating Role of Primary and Secondary

Control

In order to test hypothesis three proposing that primary control moderates the

effect of controllable difficulties on job satisfaction and secondary control moderates

the effect of uncontrollable difficulties on job satisfaction, two hierarchical multiple

regression analyses were conducted.

297


were met for both analyses, and there was no evidence of multicollinearity. As

demonstrated in Figure 9, difficulties (i.e., controllable or uncontrollable) were

entered in the first step. In the second step the moderator variable was entered (i.e.,

primary or secondary control), and in the third step the interaction term was entered.

298

Controllable difficulties

Primary control

Controllable difficulties x primary control

Uncontrollable difficulties

Secondary control

Uncontrollable Difficulties x secondary control

Job Satisfaction

Figure 9 – Primary and Secondary Control Moderate the Relationship between

Work Difficulties and Job Satisfaction

Order of Variable Entry

a)

Step 1

Step 2

Step 3

b)

Step 1

Step 2

Step 3

In the primary control analysis, R was significantly different from zero after

controllable difficulties were entered, R = 0.32, F (1, 199) = 22.91, p = 0.00. As

demonstrated in Table 48, R increased after primary control was added, R = 0.39,

Finc (1, 198) = 12.05, p = 0.001. The addition of the interaction term however did

not increase R, where R = 0.41, Finc (1, 197) = 2.99, p = 0.09. Thus, inconsistent

with hypothesis three, primary control did not moderate the effect of controllable

work difficulties on job satisfaction.

In the secondary control analysis, R was significantly different from zero

after uncontrollable difficulties were entered, R = 0.35, F (1, 189) = 26.68, p = 0.00.

Job Satisfaction

299

As demonstrated in Table 48, R did not increase when secondary control was added,

R = 0.36, Finc (1, 188) = 1.73, p = 0.19, or when the interaction term was added,

R = 0.37, F (1, 187) = 1.35, p = 0.25. As such, inconsistent with hypothesis three,

secondary control did not moderate the effect of uncontrollable work difficulties on

job satisfaction.

300

Table 48- Hierarchical Multiple Regression testing the Moderating Role of

Control Strategies on the Relationship between Work Difficulties and Job

Satisfaction

Step IV DV B sr2(unique)1 Controllable difficulties JS -0.29 -0.32 10.30**

R =0.32** R2=0.10 AdjR2=0.10

2 Controllable difficulties JS -0.28 -0.31 9.24**Primary control 0.32 0.23 5.15**

R =0.39** R2=0.16 AdjR2=0.15

3 Controllable difficulties JS -0.81 -0.89 2.82**Primary Control -0.08 -0.06Controllable difficulties xPrimary control

0.007 0.65

R =0.41 R2=0.17 AdjR2=0.16

1 Uncontrollable difficulties JS -0.29 -0.35 12.39**

R =0.35** R2=0.12 AdjR2=0.12

2 Uncontrollable difficulties JS -0.30 -0.36 12.89**Secondary Control

R =0.36 R2=0.13 AdjR2=0.12

3 Uncontrollable difficulties JS -0.56 -0.68Secondary control -0.13 -0.09Uncontrollable difficultiesx Secondary control

0.004 0.39

R =0.37 R2=0.14 AdjR2=0.12

**p<0.01; JS – Job satisfaction

301

4.7.3.1 Summary

Employees reported using more primary than secondary control strategies for

both controllable and uncontrollable difficulties. Primary control was more

positively related to job satisfaction than secondary control for both types of

difficulties. Although primary control was positively related to job satisfaction, it

did not moderate the effect of controllable difficulties on job satisfaction. Similarly,

secondary control did not moderate the effect of uncontrollable work difficulties on

job satisfaction.

4.7.4 Hypothesis Four - Moderating Role of Instrumental Support

To test hypothesis four, proposing that instrumental support moderates the

relationship between controllable difficulties and job satisfaction, hierarchical

regression analyses were conducted for co-workers and supervisors. The

assumptions of normality, linearity, and homoscedasticity of residuals were met for

both analyses, and there was no evidence of multicollinearity. In the first step

difficulties were entered (i.e., controllable or uncontrollable). In the second step the

moderator variable was entered (i.e., co-worker instrumental or supervisor

instrumental), and in the third step the interaction term was entered.

For co-workers instrumental support, R was significantly different from zero

after step one, R = 0.32, F (1, 198) = 22.74, p = 0.00, where controllable difficulties

accounted for 10% of the variance in job satisfaction. R increased after step two,

302

R = 0.49, Finc (1, 197) = 30.72, p = 0.00, where difficulties and co-worker

instrumental support accounted for 8% and 14% of the variance, respectively. R did

significantly increase when the interaction term was added in step 3, R = 0.50,

Finc (1, 196) = 4.30, p = 0.03, where the interaction term accounted for 1.64% of the

variance. This analysis, displayed in Table 49, is consistent with hypothesis four.

Although not significant at 0.01, this analysis, as in study two, will be

examined further as only a few studies have examined the moderating role of social

support. Furthermore, as discussed in study two, it is difficult to achieve statistical

significance in moderation analyses as the power is low (Bobko, 2001).

Controllable work difficulties were regressed on job satisfaction separately

for those with low co-worker support, and those with high co-worker support. As

proposed by Cohen and Cohen (1983), the low and high distinction was defined as

scores that fell one standard deviation below or above the mean for supervisor

support. As demonstrated in Figure 10, the regression lines were consistent with the

hypothesis, where the slope of the regression line of controllable work difficulties on

job satisfaction was steeper for high co-worker instrumental support than for low co-

worker instrumental support.

303

Figure 10 - Regression of Controllable Work Difficulties on Job Satisfaction for

Employees with Low Instrumental Co-Worker Support and Employees with

High Instrumental Co-Worker Support

304

For supervisors instrumental support, R was significantly different from zero

after step one, R =0.36, F (1, 176) = 25.92, p = 0.00. As demonstrated in Table 49, R

increased after supervisors instrumental support was added, R = 0.42,

Finc (1, 175) = 10.73, p = 0.001, however it did not increase further when the

interaction term was added, R = 0.42, F (1, 174) = 0.06, p = 0.80. Thus, inconsistent

with hypothesis four, supervisor instrumental support did not moderate the effect of

controllable work difficulties on job satisfaction.

305

Table 49- Hierarchical Regression Analyses Testing the Moderating Role of


Step IV DV B sr2(unique)1 Controllable difficulties JS -0.28 -0.32 10.30**

R =0.32** R2=0.10 AdjR2=0.10

2 Controllable difficulties JS -0.26 -0.29 8.49**Co-workers instrumental 0.31 0.37 14.56**

R =0.49** R2=0.24 AdjR2=0.23

3 Controllable difficulties JS -0.64 -0.72 4.20Co-workers instrumental 0.01 0.02 .Controllable difficulties xCo-workers instrumental

0.005 0.55 1.64*

R =0.50* R2=0.25 AdjR2=0.24

1 Controllable difficulties JS -0.33 0.17 12.81**

R =0.36** R2=0.13 AdjR2=0.12

2 Controllable difficulties JS -0.28 -0.31 8.94**Supervisors instrumental 0.16 0.23 5.02**

R =0.42** R2=0.18 AdjR2=0.17

3 Controllable difficulties JS -0.32 -0.35Supervisors instrumental 0.13 0.18Controllable difficulties xSupervisors instrumental

0.0005 0.06

R =0.42 R2=0.18 AdjR2=0.17

*p<0.05, **p<0.01; JS- Job satisfaction

4.7.5 Hypothesis Five- Moderating Role of Emotional Support

In order to test hypothesis five, proposing that emotional support moderates

the effect of uncontrollable work difficulties on job satisfaction, two hierarchical

306

multiple regression analyses were conducted. The assumptions of normality,

linearity, and homoscedasticity of residuals were met for both analyses, and there

was no evidence of multicollinearity.

For co-workers, R was significantly different from zero after uncontrollable

difficulties had been entered, R = 0.33, F (1, 192) = 24.48, p = 0.00. R significantly

increased after co-worker emotional support was entered, R = 0.55,

Finc (1, 191) = 51.32, p = 0.00. However, as demonstrated in Table 50, the addition

of the interaction term in step three was not significant, R = 0.56,

Finc (1, 190) = 1.85, p = 0.18. Thus, inconsistent with hypothesis five, co-worker

emotional support did not moderate the effect of uncontrollable work difficulties on

job satisfaction.

For supervisors, R was significantly different from zero after step one,

R = 0.36, F (1, 173) = 24.95, p = 0.00. As demonstrated in Table 50, R did

significantly increase after supervisors emotional support was entered, R = 0.45,

Finc (1, 172) = 16.90, p = 0.00, but did not increase further when the interaction term

was added, R =0.46, F (1, 171) = 0.76, p = 0.39. Inconsistent with hypothesis five,

co-worker and supervisor emotional support did not moderate the effect of

uncontrollable work difficulties on job satisfaction.

307

Table 50- Hierarchical Regression Analyses Testing the Moderating Role of

Emotional Support

Step IV DV B sr2(unique)1 Uncontrollable difficulties JS -0.28 -0.34 11.29**

R =0.34** R2=0.11 AdjR2=0.11

2 Uncontrollable difficulties JS -0.23 -0.28 7.67**Co-workers emotional 0.38 0.43 18.75**

R =0.55** R2=0.30 AdjR2=0.29

3 Uncontrollable difficulties JS -0.45 -0.55 2.58**Co-workers emotional 0.20 0.23Uncontrollable difficultiesx Co-workers emotional

0.003 0.33

R =0.56 R2=0.31 AdjR2=0.30

1 Uncontrollable difficulties JS -0.30 -0.36 11.22**

R =0.36** R2=0.13 AdjR2=0.12

2 Uncontrollable difficulties JS -0.24 -0.29 7.95**Supervisors emotional 0.20 0.29 7.84**

R =0.45** R2=0.20 AdjR2=0.20

3 Uncontrollable difficulties JS -0.37 -0.44 2.62*Supervisors emotional 0.09 0.13Uncontrollable difficulties x Supervisors emotional

0.002 0.20

R =0.46 R2=0.21 AdjR2=0.19

*p<0.05, **p<0.01; JS – Job satisfaction

308

4.7.5.1 Summary

In summary, only co-worker instrumental support moderated the effect of

difficulties on job satisfaction. Supervisor instrumental support, supervisor

emotional support and co-worker emotional support did not act as moderators.

4.7.6 Hypothesis Six- Major Predictors of Job Satisfaction

In order to test hypothesis six, proposing that job autonomy, difficulties at

work, control strategies, social support at work, and life satisfaction predict job

satisfaction, a standard regression analysis was conducted. The assumptions of

normality, linearity, and homoscedasticity of residuals were met for both analyses,

and there was no evidence of multicollinearity.

When all of the variables were entered, R was significantly different from

zero, R = 0.74, F (12, 142) = 13.92, p = 0.00. As demonstrated in Table 51,

controllable difficulties, job autonomy, life satisfaction, and co-workers emotional

support uniquely predicted job satisfaction. Inconsistent with hypothesis six

however, control strategies, uncontrollable difficulties, and supervisor social support

did not uniquely predict job satisfaction.

309

Table 51- Standard Multiple Regression Predicting Job Satisfaction

Predictor B sr2(unique) Controllable difficulties -0.18 -0.18 2.62**Uncontrollable difficulties -0.11 -0.13Job autonomy 0.24 0.27 4.08**Secondary control- Controllable situation 1.30 0.03Secondary control- UncontrollableSituation

-0.49 -0.01

Primary control- Controllable situation 3.15 0.09Primary control-Uncontrollable situation 0.72 0.03Life Satisfaction 0.31 0.19 4.41**Co-workers emotional support 0.17 0.19 1.49*Co-workers instrumental support 0.14 0.16Supervisors emotional support -0.04 -0.06Supervisors instrumental support -0.02 -0.02

R =0.74** R2=0.54 AdjR2=0.50

*p<0.05, **p<0.01; R is composed of 15.52% unique variance and 84.48% shared variance.

4.7.6.1 Summary

When all of the variables in the hypothesised model of job satisfaction were

entered into a regression equation, they accounted for 54% of the variance.

However, only controllable difficulties, job autonomy, life satisfaction, and

co-worker emotional support uniquely predicted job satisfaction.

4.7.7 Conclusion

Employees reported using more primary than secondary control in both

controllable and uncontrollable situations. Primary control was more adaptive than

secondary control in both situations and was positively correlated with job

satisfaction. However, primary and secondary control did not moderate the effect of

310

work difficulties on job satisfaction. Co-worker instrumental support did moderate

the effect of controllable work difficulties on job satisfaction, however supervisor

instrumental support did not. Furthermore, emotional support did not moderate the

effect of uncontrollable difficulties on job satisfaction. The major predictors of job

satisfaction were controllable difficulties, job autonomy, life satisfaction and

co-worker instrumental support. These findings will now be discussed.

311

4.8 Discussion

The study proposed that the controllability of a work difficulty influences the

use and adaptiveness of the control strategies used to handle that difficulty. The

findings demonstrated, however, that for both controllable and uncontrollable

difficulties, primary control strategies were used more than secondary control

strategies, and primary control strategies were more adaptive than secondary control

strategies. These findings, which are inconsistent with the discrimination model,

suggest that trait control strategies may exist. The proposal that employees use

similar control strategies in all situations questions the assumption that employees

using primary control in uncontrollable situations will experience primary control

failure.

The results from this study also question the importance of the control

strategies, as they, along with social support, did not moderate the effect of work

difficulties on job satisfaction. These findings must be regarded with caution

however as limitations have now been identified in the operationalisation of work

difficulties. These hypotheses will now be examined.

4.9 Hypotheses Testing

The hypotheses can be grouped into three major proposals. The first proposal

is that the controllability of a difficulty influences the use and adaptiveness of the

control strategies used to handle that difficulty. The second proposal is that the

control strategies and social support at work moderate the effects of work difficulties

on job satisfaction. The third proposal is that general job autonomy, difficulties at

312

work, control strategies, social support at work and life satisfaction predict job

satisfaction. Before these proposals are discussed, the conceptualisation of the

control strategies requires explanation.

4.9.1 Primary Control, Self-Protective Secondary Control, and Self-

Affirmative Secondary Control

Factor analyses of the control strategies demonstrated that employees were

using three types of control strategies, namely primary control, self-protective

secondary control and self-affirmative secondary control. Although two types of

secondary control were identified in chapter 2, it was not known whether the

differences between them would be great enough to form separate factors. As this

was the case however, the two types of secondary control require further exploration.

All of the secondary control strategies involve people changing themselves to

fit in with their situation, however there are two ways that this can be done. Self-

protective secondary control strategies reduce negative feelings about the situation.

The strategies that loaded on the self-protective factor were attribution (“Think that

this difficulty is not my fault”), predictive negative (“Think that I knew this difficulty

would happen”), and wisdom (“Think that I can’t always get what I want”). These

strategies make the situation less concerning, and people conclude that a situation is

not as bad as it seems.

The second type of secondary control, self-affirmative, promotes positive

feelings. The strategies that were identified as being self-affirmative were support

(“Tell someone about this difficulty to make me feel better”), vicarious (“Think of

313

the success of my family or friends”), present success (i.e., “Think about my success

in other areas”), active avoidance (“Do something different, like going for a walk”)

and positive re-interpretation (“Look for something else that is positive in the

situation”). These strategies make people feel good about themselves and their lives.

It must be noted that this conceptualisation of self-protective and self-

affirmative secondary control was not completely supported. Four items did not load

on the expected factors. Specifically, downward social comparison (“Think that I am

better off than many other people”) did not load on the self-affirmative factor.

Furthermore, goal disengagement (“Think that the difficulty doesn’t matter”),

illusory optimism (“Think that this difficulty will work out okay in the end”) and

denial (“Ignore this difficulty”) did not load on the self-protective factor. There is no

ready explanation as to why these items did not load on the expected factors. Clearly

however, the majority of items were consistent with the conceptualisation of self-

protective and self-affirmative secondary control.

Although this is a novel approach to secondary control strategies, it must be

noted that the conceptualisation of these three strategies is still consistent with

Heckhausen and Schulz’s (1995) proposals. Specifically, primary control strategies

involve attempts to change the environment to fit in with the self, and both types of

secondary control strategies involve attempts to change the self to fit in with the

environment. The new idea however is that some secondary control strategies

reduce negative feelings, whilst others promote positive feelings.

Although factor analyses have not been conducted on other primary and

secondary control scales as they generally contain only one item (i.e., Thompson et

314

al., 1996, 1994), they have been conducted on coping scales. The factors emerging

from these analyses can be compared to the three factors found in this study. As the

most common coping scale is the Ways of Coping Questionnaire (Folkman &

Lazarus, 1985; Folkman, Chesney, Cooke, Boccellari & Collette, 1994), factor

analyses of this scale will be examined.

Unlike the Situation Specific Primary and Secondary Control Scale (Maher et

al., 2002), which uses one item for each strategy, the Ways of Coping Questionnaire

(Folkman & Lazarus, 1985) uses multiple items for each strategy. As such, when

factor analyses are conducted on the scale, the items cluster according to the type of

strategy. For example, Folkman et al., (1986) demonstrated that a factor analysis,

averaged across several samples yielded eight factors, including confrontative

coping, distancing, self-controlling, seeking social support, accepting responsibility,

escape-avoidance, planful problem solving and positive reappraisal. It must be noted

however, that as discussed in chapter 1, factor analyses conducted on this scale are

far from consistent (Edwards & O’Neill, 1998).

Factor analyses of the Ways of Coping Questionnaire are not comparable to

those conducted on the Situation Specific Primary and Secondary Control Scale.

Factor analyses of the Ways of Coping Questionnaire identify which items measure a

particular strategy, whereas factor analyses of the Situation Specific Primary and

Secondary Control Scale (Maher et al., 2002) identify which strategies cluster

together. As such, the factor analyses in the current study are theoretically different

from previous analyses. Rather than just examining whether items measure a

315

strategy, they demonstrate how the strategies are related to each other. This means

that the underlying purpose of the strategies can be examined.

The development of three types of control strategies is exploratory, and as

such, further research is required. However, this conceptualisation may be useful in

determining the best type of secondary control. It could be hypothesised that self-

affirmative secondary control would be more positively correlated with job and life

satisfaction than self-protective secondary control, as rather than just decreasing

negative feelings, they increase positive feelings. This proposal is not supported in

the current study as both self-protective and self-affirmative secondary control

strategies were not related to job satisfaction. Despite this however, further research

may benefit from recognising there may be two types of secondary control. The

three major proposals of this study will now be examined.

4.9.2 Proposal One: The Controllability of the Difficulty Influences the

Amount and Adaptiveness of the Control Strategies Used to Manage that

Difficulty

4.9.2.1 The Amount of Control Strategies Used for Controllable and


It was hypothesised that employees would use more primary than secondary

control for controllable difficulties, and more secondary than primary control for

uncontrollable difficulties. Support was found for the former part of the hypothesis,

316

however no support was found for the latter as employees reported using more

primary than secondary control for uncontrollable difficulties.

These finding are partially consistent with the life span theory of control

(Heckhausen & Schulz, 1995). This theory proposes that people prefer primary

control and that it has primacy over secondary control. Consistently, employees

reported more primary than secondary control for controllable difficulties.

However, the theory also proposes that when people are faced with

uncontrollable situations, the probability of primary control failure increases, and

control strategies focus on changing oneself rather than changing ones environment

(Heckhausen & Schulz, 1995). This does not appear to be the case for the employees

in this study however, as they report more primary than secondary control in

uncontrollable situations.

Only one other study has reported the amount of control strategies used in an

uncontrollable situation. Thompson et al., (1996) demonstrated that HIV positive

men in prison (i.e., low-control situation) reported slightly more primary control (M

= 48.5%SM) than secondary control (M = 45%SM). This study, which is also

inconsistent with the life span theory of control, was criticised in chapter 1 for

measuring primary control using perceived control and secondary control using

acceptance. However, it now appears that even when a new measure of primary and

secondary control is used, employees report using more primary than secondary

control for uncontrollable difficulties. Three explanations have been developed for

this finding.

317

4.9.2.2 Why is Primary Control Used more than Secondary Control for

Uncontrollable Difficulties?

There are three possible explanations for the employees reporting more

primary than secondary control for uncontrollable difficulties. First, it may be that

when completing the questionnaire, the respondents were unable to conceptualise

uncontrollable difficulties. Second, primary control may be used first for all

difficulties and secondary control may only be used if primary control fails. Third,

the controllability of the situation may not influence the control strategies people use,

and rather people may have trait control strategies. These explanations will be

discussed.

4.9.2.3 a) Conceptualisation of Controllable and Uncontrollable Difficulties

One reason why the employees may have reported higher primary control than

secondary control in uncontrollable situations is that the employees were unable to

conceptualise the distinction between controllable and uncontrollable difficulties.

The terms “controllable” and “uncontrollable” were used because, although being

abstract in nature, they did not bias the respondents as much as other constructs such

as change, influence, do something about, or accept.

Despite being abstract, it appears that the participants generally did understand

these terms and the distinction between them. The results demonstrated that the

majority of participants indicated that difficulties with time management, motivation

and co-workers were controllable and difficulties with pay, amount of work,

318

work-place rules and promotion were uncontrollable. As such, it appears that the

participants understood what constituted a controllable and an uncontrollable

difficulty, and hence this proposal does not explain why employees reported higher

primary than secondary control in uncontrollable situations.

4.9.2.4 b) Primary control is Implemented First for Controllable and


Another explanation for the finding that primary is used more than secondary

control for uncontrollable difficulties is that primary control is always implemented

first. It was assumed that employees would rely on secondary control for

uncontrollable difficulties in an attempt to avoid primary control failure. However, it

must be noted that Heckhausen and Schulz (1995) proposed that primary control

strategies are used first and it is possible that this applies in controllable and

uncontrollable situations. Perhaps people attempt to change all situations using

primary control, and if they fail, they then rely on secondary control strategies. If

this is the case, it would be expected that people would use comparable amounts of

primary control in controllable and uncontrollable situations, but that they would use

more secondary control in uncontrollable situations.

As demonstrated by the mean levels of primary and secondary control

however, this does not appear to be case. The primary control levels were similar for

controllable situations (M = 71.98) and uncontrollable situations (M = 64.56),

however there was no difference in their levels of secondary control

319

(controllable, M = 53.74, uncontrollable, M = 53.10). Hence, the proposal that

employees report more primary control than secondary control in uncontrollable

situations because they use primary control first in such situations and only use

secondary control when primary control fails, does not appear to be supported.

4.9.2.5 c) Trait Control Strategies

Another explanation for the finding that the employees reported more

primary control than secondary control for uncontrollable difficulties is that trait

control strategies may exist. People may have a set of strategies that they

consistently use to handle their difficulties, and they may not consider the usefulness

of the strategy within that situation. The correlations between the control strategies

used in controllable situations with the control strategies used in uncontrollable

situations supports this proposal. Primary control for a controllable difficulty was

strongly correlated with primary control for an uncontrollable difficulty (r = 0.69).

Furthermore, secondary control for a controllable difficulty was strongly correlated

with secondary control for an uncontrollable difficulty (r = 0.64). The correlations

between primary and secondary control strategies were much weaker. Primary and

secondary control strategies for controllable difficulties were not correlated. Primary

and secondary control for uncontrollable difficulties were only weakly correlated

(r = 0.20).

The idea that peoples’ responses to difficulties are stable has been discussed

in the coping literature. It is proposed that people have coping “styles”,

“dispositions”, or “traits” that they bring to the situation (Carver et al., 1989).

320

Accordingly, “people do not approach each coping context anew, but rather bring to

bear a preferred set of coping strategies that remains relatively fixed across time and

circumstances” (Carver et al., 1989, p. 270).

Few researchers have examined trait coping, perhaps because Folkman and

Lazarus (1986) disputed the idea, proposing that coping is contextual, and that it is

influenced by the person’s appraisal of the situation. However, other studies besides

the current research dispute this proposition. A study conducted by Schwartz, Neale,

Marco, Shiffman and Stone (1999) assessed trait coping using the Daily Coping

Questionnaire (Stone & Neale, 1984) and the Ways of Coping Questionnaire

(Folkman & Lazarus, 1984). The question at the beginning of each scale was

changed to “how do you typically deal with stressful situations.” They also

measured coping using a momentary measure where participants were given a

programmable palm-top computer. Participants would type in their stressful events

and indicate how they coped with them immediately after the event.

They examined how much of the variance in the momentary scales was due

to differences among individuals. For example, for escape coping, they examined

how much of the variance was due to the tendency for some individuals to report

escape coping more than others. The results demonstrated that 42% of the variability

in the momentary assessments was due to individual differences in escape coping.

The other coping strategies accounted for less of the variance, ranging between

20-30% for the Ways of Coping Scale, and for 15-19% of the Daily Coping Scale.

These findings suggest that a person’s coping response could be partially predicted

from a general coping scale, and thus supports trait coping.

321

The proposal that coping is a trait or disposition can be used to explain the

current findings. Employees may have reported using primary control for an

uncontrollable difficulty because primary control strategies are within their

disposition. Thus, rather than evaluating the situation, they evaluate the coping

strategies they have in their repertoire.

4.9.2.6 Summary

Consistent with the life span theory of control, employees reported using

more primary than secondary control for controllable difficulties. Inconsistently

however, they also reported using more primary than secondary control for

uncontrollable difficulties. Three explanations were developed to account for these

findings. The first, proposing that employees did not understand uncontrollable

difficulties, was dismissed, as employees seemed to classify their difficulties as

expected. The second explanation proposed that people use more primary control for

uncontrollable difficulties because they implement primary control first for all

difficulties, and only use secondary control if primary control fails. This was not

supported by the data, as the levels of secondary control were the same. The third

explanation proposed that the controllability of the situation did not influence the

control strategies the employees used. Rather, it is proposed that employees have

‘trait’ control strategies. Employees may fail to evaluate the situation and rather

simply use the strategies in their repertoire. The relationship between these control

strategies and job satisfaction will now be examined.

322

4.9.2.7 Adaptiveness of Primary and Secondary Control for Controllable and


It was hypothesised that primary control would be more adaptive than

secondary control for controllable difficulties and that secondary control would be

more adaptive than primary control for uncontrollable difficulties. Partial support

was provided for this hypothesis, as primary control was more positively related to

job satisfaction than secondary control for controllable difficulties. Inconsistently

however, primary control was also more positively related to job satisfaction than

secondary control for uncontrollable difficulties.

These findings are inconsistent with the discrimination model (Thompson et

al., 1998), which proposes that primary control is the most adaptive strategy only for

controllable situations. Rather, the findings support the primacy/back-up model

(Thompson et al., 1998), which proposes that primary control is more adaptive than

secondary control in both controllable and uncontrollable situations.

As with the current study, past empirical studies have supported the

primacy/back up model (Thompson et al., 1996; 1994; 1993; 1998). As limitations

were identified in these studies however, it was thought that when these limitations

were addressed, the discrimination model would be supported. These limitations,

discussed in chapter 1, concern the measurement of perceived control and primary

and secondary control strategies. A more notable flaw however is that these studies

failed to adequately test the discrimination model and the primacy/back-up model.

Rather than correlating the controllability of a situation with the control strategies

used to handle that situation, these studies examined general levels of perceived

323

control and control strategies. Some of the studies did examine the constructs at a

more specific level (e.g., Thompson et al., 1996, 1994), however they then

aggregated the items to obtain an overall measure of perceived control and an overall

measure of primary and secondary control. It appears however that even when all of

the limitations were addressed, the findings still supported the primacy/back up

model.

The current findings, although referring to control strategies, can also be

compared to the empirical studies on the goodness of fit hypothesis for coping

strategies. These studies generally demonstrate that, consistent with the current

findings, problem-focussed strategies are more adaptive than emotion-focussed

strategies in controllable situations. They also demonstrate that emotion-focussed

strategies are not more adaptive than problem-focussed strategies in uncontrollable

situations (e.g., Bowman & Stern, 1995; Conway & Terry, 1992; Osowieki &

Compas, 1998, 1999; Park, Folkman & Bostrom, 2001; Vitaliano et al., 1990).

These studies were criticised for their research designs in the introduction. It appears

however, that even when these problems are addressed, similar results are obtained.

In summary, it appears that consistent with past studies, primary control is

more adaptive than secondary control for both controllable and uncontrollable

difficulties. As many flaws were identified with the past studies, it was expected that

when these flaws were addressed, the results would be more consistent with the

discrimination model. This is not the case however, and as such further exploration

is needed to explain why primary control is more adaptive than secondary control in

uncontrollable situations.

324

4.9.2.8 Why is Primary Control more Adaptive than Secondary Control in

Uncontrollable Situations?

The finding that primary control is adaptive in uncontrollable situations is

contrary to intuition. As such, it is important that this finding can be explained

theoretically. It was expected that if employees tried to change an uncontrollable

situation using primary control, they would fail and this failure would negatively

influence perceived competence, self-efficacy, self-esteem (Heckhausen et al., 1997),

and job satisfaction.

The current findings, which demonstrate that primary control is positively

related to job satisfaction for uncontrollable difficulties, challenge the assumptions

regarding primary control failure. Primary control failure has not been measured in

the past, or in the current study, as it is extremely difficult to assess. It requires the

person to indicate how often they used each of the control strategies and indicate the

successfulness of each strategy. This is cognitively taxing for the respondents, and if

completed for primary and secondary control, would add another 17 items to each

control scale (controllable and uncontrollable). More importantly however, it may

not even be possible for people to recall this information. Whilst they may

remember whether they solved a problem, it is unlikely that they can recall which

strategy was more successful than others. Furthermore, it may actually be a

combination of strategies that contributes to the problem being overcome. For these

reasons, the successfulness of the strategies was not assessed.

As primary control failure was not measured in the current study however, it

is possible that it did not behave as expected. Firstly, it may be that the employees

325

who are implementing primary control are not experiencing primary control failure.

Secondly, employees may be experiencing primary control failure, yet experiencing

few negative consequences. These explanations will be discussed.

4.9.2.9 Primary Control does not lead to Primary Control Failure

In regard to the first explanation, employees who reported high primary

control for “uncontrollable” difficulties may have reported high job satisfaction

because they successfully implemented the strategies. Perhaps people only use

primary control when they know that they will be successful. Indeed, it seems

maladaptive for people to use primary control if they know that it will lead to


If it proposed that employees only use primary control when they know they

will be successful, it must still be questioned how they could successfully change an

uncontrollable situation using primary control. One possibility is that the difficulties

reported by employees as being uncontrollable are only low-control difficulties.

Most of the difficulties reported, such as pay, promotion and workplace-rules may

not be completely uncontrollable. Other people determine them, and it is possible for

the people to be influenced, and thus for primary control to be successful. Perhaps

different results would be obtained if people were given difficulties that are clearly

uncontrollable such as the death of a loved one or a natural disaster.

326

4.9.2.10 Primary Control Failure does not Negatively Influence Job

Satisfaction

In regard to the second explanation, it may be that the employees are

experiencing primary control failure, but that the primary control failure is not

having negative effects. The life span theory of control proposes that primary

control failure will threaten perceived competence, self-efficacy, and self-esteem

(Heckhausen et al., 1997). It must be noted however that these effects have not been

tested. Perhaps it is better to have tried to implement primary control and failed than

to have not tried at all. Employees can tell themselves that there was nothing more

they could do, and thus they may feel better about their own control efforts.

Both of these explanations are speculative, and indeed require empirical

validation. To do this, future studies need to invest time in developing and

measuring the successfulness of primary and secondary control strategies.

4.9.2.11 Summary

Although primary control was more adaptive than secondary control in

controllable situations, it was also more adaptive in uncontrollable situations. These

findings are inconsistent with the discrimination model and the goodness of fit

hypothesis. It is difficult to explain as it was expected that employees who used

primary control for uncontrollable difficulties would experience primary control

failure. However, it may be that the employees only use primary control when they

know that they will be successful. Alternatively, the employees may experience

primary control failure, yet the consequences of primary control failure may be less

327

damaging than not attempting at all. Further empirical research is required to

examine these proposals.

4.9.3 Proposal Two: Moderators of Controllable and Uncontrollable

Difficulties on Job Satisfaction

Moderators of work difficulties were examined, as these variables may be

more amenable to change than work difficulties. Employers may be reluctant to

reduce work difficulties, where both the job and the organisation would need to

undergo a thorough assessment. Furthermore, it may be impossible to reduce some

work difficulties if they are inherent in the nature of the work.

It was hypothesised that the control strategies and social support at work

would moderate the relationship between work difficulties and job satisfaction.

Inconsistently however, primary control did not moderate the effect of controllable

difficulties, and secondary control did not moderate the effect of uncontrollable

difficulties. For the social support variables, only co-worker instrumental support

moderated the effect of controllable work difficulties on job satisfaction. It must be

noted that this finding was significant at 0.05, however it was not significant at the

more stringent alpha level of 0.01. As few studies have examined the moderating

role of social support in the workplace however, this finding was examined further.

The finding that co-worker support played a greater role than supervisor

support is inconsistent with other studies (e.g., Beehr, 1985; Fenlason & Beehr,

1994; Russell, Altmaier & Van Velzen, 1987). It was expected that as co-workers

have less influence over difficulties at work, their support would not be as beneficial

328

as supervisor support (Fenlason & Beehr, 1994). It must be noted however that the

measure of supervisor support used in this scale was exploratory. Although the scale

has face validity, there are no independent psychometric data for the scale. The

findings will be compared to past studies.

4.9.3.1 Past Studies Examining the Moderators of Stress

In regard to the control strategies, no other studies have examined whether

the control strategies moderate the effect of work difficulties. However, a few

studies have demonstrated that coping strategies moderate the effect of stressors on

stress (Aldwin & Revenson, 1987, Ashford, 1988; Parkes, 1990, 1994; Perrewe &

Zellars, 1999; Osipow et al., 1985). These studies are inconsistent with the current

findings, demonstrating that some coping strategies do moderate work stress. There

is no conclusive evidence however, as to which control strategies moderate work

stress.

In regard to social support, a few studies, including study two, have

demonstrated that social support at work has a moderating effect on job satisfaction

(i.e., Karasek et al., 1982; Landsbergis et al., 1992). However, other studies have

failed to find the moderating role of social support (Chay, 1993; de Jonge &

Landeweerd, 1993; Melamed at al., 1991; Parkes & Von Rabenau, 1993). As

discussed in chapter 3, one difference between the supportive and non-supportive

studies is the measure of social support.

Two of the supportive studies (i.e., current study and Landsbergis et al.,

1992) relied on Karasek and Theorell’s (1990) scale. Although some of the items in

329

Karasek and Theorell’s (1990) scale were criticised in this chapter, there is certainly

no agreed way of measuring social support at work (Unden, 1996). The current study

does not shed light on the problem however, as the social support scale did not factor

as expected. The scale was only measuring two variables, supervisor support and

co-worker support. Further research is needed on the operationalisation of social

support to ensure that all four types of social support are assessed.

In general, the findings on the moderating role of the control strategies and

social support are somewhat inconsistent with other similar studies. One major

difference between the current study and the other studies however is the

independent variable. Other studies have relied on job stress or work demands,

whereas this study used work difficulties. This may have been problematic since

work difficulties, controllable and uncontrollable, did not strongly predict job

satisfaction.

4.9.3.2 Limitations in the Moderation Hypotheses

The finding that work difficulties did not strongly predict job satisfaction is a

concern for the robustness of this analysis. A moderation analysis tests whether the

relationship between two variables (i.e., work difficulties and job satisfaction) varies

depending on a moderator variable (i.e., control strategies or social support). A

median split conducted on the moderator produces a low group and a high group.

The relationship between work difficulties and job satisfaction for each group is then

examined. If the relationship between the two variables is not strong for the average

group however, it is unlikely that it will be strong when the moderator is low or high.

330

It was expected that work difficulties would strongly predict job satisfaction,

and as such, two explanations have been developed to account for the weak

relationship. These concern the nature of work difficulties and the operationalisation

of work difficulties.

4.9.3.3 Nature of Work Difficulties

Researchers that have examined the moderating role of social support have

examined job demands rather than work difficulties. Job demands are the

psychological stressors in the work environment (i.e., high pressure of time, high

working pace, difficult and mentally exacting work; Karasek & Theorell, 1990).

Work difficulties are much broader than job demands, and refer to any problems that

employees face at work.

4.9.3.4 Operationalisation of Work Difficulties

Work difficulties were measured by asking the employees to indicate how

often they experienced their most commonly occurring difficulty. This is a difficult

question to answer, as the employee needs to consider all of the difficulties that they

face, think about how often they face each one and identify the one that they face the

most.

This item was useful in that it led people into thinking about how they handle

that difficulty, however it may not have accurately assessed work difficulties. One

person may experience one difficulty all the time yet rarely experience any other

difficulties. Another person may experience ten difficulties all the time. Using the

331

current scale however, these respondents would receive the same score. Thus, the

difficulty at work scale requires revision. Perhaps the primary and secondary control

scale could still include the item assessing the most common difficulty as this helps

respondents to focus on a specific situation, however another measure of work

difficulties is required.

Developing a valid measure of work difficulties for a general sample of

employees is problematic. The obvious solution is to ask respondents on average

how often they face controllable and uncontrollable difficulties at work. These items

may be prone to errors however as they are cognitively taxing, requiring the

employee to mentally average their work difficulties.

Another solution is to ask respondents to indicate how often they experience

each difficulty that they select from a list. Hence, as with the current scale, the

respondents would be given a list of general work difficulties. They would tick

which ones they experience and could control and then indicate how often they

experience each difficulty. They would then do the same for uncontrollable

difficulties. The problem however is that with the addition of the frequency item, the

length of the scale doubles. Furthermore, there are an unlimited number of work

difficulties and as such, some would be omitted.

Another solution is to develop a list of occupational specific difficulties and

ask employees how often they experience them. This solution, although it would

enable the testing of the moderation hypotheses, is discouraged however, as different

occupational groups cannot be compared.

332

One final solution is offered. An open-ended format could be used, where

respondents are asked to list their top five difficulties at work, and for each one,

indicate how often they face it. This solution may be more time-consuming for the

researcher to code, however it is not too cognitively taxing and it can be applied to a

general sample of employees.

4.9.3.5 Summary

There was little support for the moderating role of the control strategies and

social support on the relationship between work difficulties and job satisfaction.

These findings are limited however by the operationalisation of work difficulties.

The scale only examined the most frequently occurring difficulty and as such, did not

provide an accurate assessment of work difficulties. Future researchers may need to

use an open-ended format, where respondents are asked to list their difficulties at

work and indicate how often they face each one.

4.9.4 Proposal Three: Predictors of Job Satisfaction

It was hypothesised that job autonomy, difficulties at work, control strategies,

social support at work, and life satisfaction would predict job satisfaction. Only

controllable difficulties, job autonomy, life satisfaction, and co-worker emotional

support uniquely predicted job satisfaction (15%). Thus, primary and secondary

control strategies, uncontrollable difficulties, co-worker instrumental support, and

supervisor emotional support and instrumental support did not uniquely predict job

satisfaction. Possible explanations for these findings are discussed.

333

The finding that the control strategies did not uniquely predict job satisfaction

is particularly difficult to explain. It is intuitive that the control strategies that

employees use to handle work difficulties influence their level of job satisfaction.

One possibility is that it may not be primary and secondary control alone that predict

job satisfaction, rather the effectiveness of the control strategies. Future studies may

need to assess the control strategies and the effectiveness of them.

The finding that uncontrollable difficulties did not uniquely predict job

satisfaction may reflect the operationalisation of work difficulties. Employees were

asked how often they face their most commonly occurring controllable and

uncontrollable difficulty. As discussed previously, this measure may be flawed and

as such, more research is required to understand the importance of work difficulties

in predicting job satisfaction.

The finding that supervisor support did not uniquely predict job satisfaction

may also be explained by its measurement. As discussed previously, the social

support scale did not factor as expected, and it appeared as though the scale was only

measuring two variables, supervisor support and co-worker support. The scale was a

co-worker scale that was extended to supervisors. Perhaps separate scales are

required for the different roles. As such, further research is needed on the

operationalisation of social support to ensure that all four types of social support are

assessed.

334

4.9.4.1 Summary

Partial support was provided for the proposed predictors of job satisfaction,

as controllable difficulties, job autonomy, life satisfaction and co-worker emotional

support uniquely predicted job satisfaction. The finding that primary and secondary

control strategies, uncontrollable difficulties, co-worker instrumental support and

supervisor emotional and instrumental support did not uniquely predict job

satisfaction may be due to operationalisation issues.

4.9.5 Conclusion

The study tested three major proposals, which centered on job satisfaction,

control strategies and the controllability of the situation. The first proposal that the

controllability of the difficulty influences the use and adaptiveness of the control

strategies used for that difficulty, was not supported. Employees reported using

more primary control than secondary control for controllable and uncontrollable

difficulties. Three explanations were developed to account for these findings,

however the most plausible was that people have trait control strategies.

In addition to being used more than secondary control, primary control was

also more adaptive than secondary control for controllable and uncontrollable

difficulties. These findings, which are inconsistent with the discrimination model,

challenge the assumptions about primary control failure. It is possible that primary

control was adaptive for uncontrollable difficulties because it was being

implemented successfully. If it was being implemented successfully, perhaps the

work difficulties at work were low-control rather than being uncontrollable.

335

Alternatively, the employees may have been experiencing primary control failure,

however that failure may not have negatively affected job satisfaction.

The second major proposal, that the control strategies and social support at

work moderated the effects of work difficulties on job satisfaction, was generally not

supported. The findings tended to be inconsistent with previous studies examining

job stress, and the replacement of job stress with work difficulties was questioned.

Specific problems with the operationalisation of work difficulties were identified that

may have limited the findings.

The third proposal, that general job autonomy, difficulties at work, control

strategies, social support at work and life satisfaction predict job satisfaction was

partially supported. On the basis of these findings, it was clear that measures of

primary and secondary control, work difficulties, and social support require further

exploration.

In summary, these findings suggest that a satisfied worker has high job

autonomy, high social support, high life satisfaction, few work difficulties, and uses

primary control to deal with controllable and uncontrollable difficulties. The

implications of these findings will be discussed in chapter 5.

336

5 Chapter 5 - Final Discussion

337

5.1 Abstract

This thesis tested a model of job satisfaction that includes environmental and

dispositional predictors. The major proposal of the model is that job autonomy

influences the use and adaptiveness of primary and secondary control strategies. The

model also examines other predictors of job satisfaction, including life satisfaction,

work difficulties, and social support at work. Additionally, it proposes that the

control strategies and social support at work moderate the relationship between work

difficulties and job satisfaction. Empirical support offered for these proposals in

chapters two, three and four are reviewed and a revised model of job satisfaction is

presented. This model continues to include job autonomy, primary and secondary

control, life satisfaction and work difficulties, however it also includes the

successfulness of primary and secondary control and re-introduces personality

variables.

338

5.2 The Development of A New Model of Job Satisfaction

This thesis developed a model of job satisfaction that includes environmental

(i.e., job autonomy, social support at work, and work difficulties) and dispositional

predictors (i.e., primary and secondary control, personality and life satisfaction).

This model extended the job demand-control model (Karasek & Theorell, 1990),

offering an alternative explanation for the positive relationship between job

autonomy and job satisfaction. The job demand-control model was selected for

further investigation because, unlike other dominant theories, it is highly applicable

to the workplace and attractive to employers.

The job demand-control model proposes that job demands and job decision

latitude interact to predict job satisfaction, and that the most satisfied workers are

those who have high job decision latitude and high job demands. The implication of

this proposal is that employers can increase job satisfaction without reducing work

demands.

According to Karasek and Theorell (1990), employees with high job decision

latitude can translate the physiological arousal produced from job demands into

action through effective problem solving. They propose that workers with high job

autonomy are “given the freedom to decide what is the most effective course of

action in response to a stressor” (Karasek & Theorell, 1990, p. 36). However, this

explanation has been criticised for being tautological. It proposes that job control, or

the ability to choose at work, increases job satisfaction because it allows people to

choose how they deal with their demands at work.

339

An alternative explanation for the positive link between job autonomy and

job satisfaction is that employees with high job autonomy have higher job

satisfaction because they respond differently to work difficulties. Employees can

respond to work difficulties in two ways; they can either change the situation using

primary control or they can change themselves using secondary control.

It is expected that these primary and secondary control strategies mediate the

relationship between job autonomy and job satisfaction. Job autonomy is expected to

influence the amount of control strategies that employees use and the adaptiveness of

those strategies.

In regard to the amount of control strategies, employees with high job

autonomy are expected to rely on more primary control and less secondary control

than employees with low job autonomy. As primary control strategies are preferred

over secondary control strategies, employees with higher job autonomy have higher

job satisfaction than employees with lower job autonomy.

In regard to the adaptiveness of the strategies, it is expected that primary

control strategies are more adaptive than secondary control only when the situation is

controllable. When the situation is uncontrollable, secondary control is expected to

be the most adaptive strategy.

The major proposal of the model of job satisfaction is thus that: 1) primary

and secondary control mediate the relationship between job autonomy and job

satisfaction. However, several other propositions are also examined, including that;

2) social support at work and life satisfaction are positively related to job satisfaction

and; 3) the control strategies and social support at work moderate the relationship

340

between work difficulties and job satisfaction. Empirical tests of these proposals will

be examined.

1) Primary and Secondary Control Strategies Mediate the Relationship Between

Job Autonomy and Job Satisfaction

It is expected that primary and secondary control strategies explain the

relationship between job autonomy and job satisfaction. Job autonomy is expected to

influence the amount of control strategies than employees report, and the

adaptiveness of the control strategies.

5.2.1.1 Job Autonomy Influences the Use of Primary and Secondary Control

Strategies

It is expected that all employees, with either low or high job autonomy,

implement primary and secondary control strategies. According to the life span

theory of control (Heckhausen & Schulz, 1995), primary control has primacy over

secondary control as it is preferred and is implemented first. If primary control is

implemented successfully, the problem is resolved. If primary control fails however,

the person is expected to implement secondary control strategies to compensate for,

and avoid, future primary control failure.

When these propositions are applied to the workplace, it is expected that job

autonomy influences the likelihood of primary control failure. It is proposed that job

autonomy is inversely related to the probability of primary control failure, which in

341

turn, influences the use of secondary control strategies. Thus, employees with high

job autonomy are expected to experience less primary control failure and as such, use

less secondary control than employees with low job autonomy.

This proposal was tested by comparing the control strategies reported by low

job autonomy workers with those reported by high job autonomy workers. Study

one compared supermarket workers and academics, whilst study two compared

teachers and academics. Both of these studies provided minimal support.

In study one, the supermarket workers reported similar levels of primary

control and more secondary control, than the academics. Although these results

suggest that job autonomy influences the use of secondary control, but not primary

control, it must be noted that these results are based on the levels of job autonomy

inferred from type of occupation. Thus, it is assumed that supermarket workers are

low in job autonomy and academics are high in job autonomy. When the same

analysis was conducted with the reported levels of job autonomy, the results

changed, in that only primary control was related to job autonomy. Thus the findings

from study one suggest that job autonomy influences the use of primary control, but

not secondary control. These findings were limited however, as the primary and

secondary control scale used in this study was flawed. The rating scale did not

assess how much control strategies the person was using, rather how much they

agreed with the strategies presented to them in the scale.

The primary and secondary control scale was revised for study two and

administered to teachers and academics. This study was not supportive of the

proposals however, as the groups reported similar levels of primary and secondary

342

control. When studies one and two are considered together, it appears as though

there is little support for the proposal that job autonomy influences the use of the

control strategies.

These findings were attributed to, in part, the specificity of the hypotheses.

Studies one and two examined the proposal that job autonomy influences the control

strategies at a general level, measuring how much control employees have over their

work environment and how they generally handle work difficulties. It was expected

that this relationship may increase in strength however if the hypotheses were more

specific. In this case, the controllability of one situation would be correlated with the

control strategies used to handle that situation.

As such, study three examined the amount of control strategies that

employees used for controllable and uncontrollable difficulties. It was hypothesised

that employees would use more primary than secondary control for controllable

difficulties and more secondary than primary control for uncontrollable difficulties.

Inconsistently however, employees reported more primary than secondary control

strategies for controllable and uncontrollable difficulties.

One explanation for this finding is that employees have trait control

strategies. The use of primary and secondary control for controllable difficulties was

highly correlated with the use of primary and secondary control for uncontrollable

difficulties. Thus, people may have a set of strategies that they consistently use to

handle their difficulties. Rather than evaluating the controllability of each work

difficulty, employees may simply use the strategies that they know.

343

If the amount of primary and secondary control used by employees remains

stable across situations, then a dispositional factor, such as personality may predict

the control strategies. A few researchers have previously examined how personality

variables relate to coping strategies (Brebner, 2001; Carver et al., 1989; Gunthert,

Armeli & Cohen, 1999; Saklofske & Kelly, 1995; Scheier, Weintraub & Carver,

1986). It is proposed that people with high extroversion use active coping strategies

where they talk out their problems and people with high neuroticism use passive

coping strategies where they tend to blame themselves, and also other people (Costa,

Somerfield & McCrae, 1996). These proposals are extended to the control strategies,

where it is expected that extroversion is positively related to primary control and

neuroticism is positively related to secondary control.

The correlations between extroversion and neuroticism and the control

strategies were examined in studies one and two. In regard to extroversion, study

one demonstrated that primary control was positively correlated with extroversion

for the academics, r = 0.25, and the supermarket workers, r = 0.42. Furthermore, in

study two, primary control was positively related to extroversion for the academics,

r = 0.20. Thus, these findings suggest that people high on extroversion use more

primary control.

In regard to neuroticism, study one demonstrated that there was no

relationship between secondary control and neuroticism. Study two provided some

support, as teachers’ levels of neuroticism were positively related to secondary

control (r = 0.19). Although these findings suggest that neuroticism is at best, only

weakly correlated with secondary control, studies using coping strategies have

344

demonstrated that neuroticism is strongly correlated with emotion-focussed coping

(i.e., Brebner, 2001; Saklofske & Kelly, 1995). These higher correlations may

reflect the difference between emotion-focussed coping strategies and secondary

control strategies. In general emotion-focussed strategies tend to be more negative

than secondary control strategies and thus may be more positively correlated with

neuroticism. As the secondary control strategies developed for this study included

more positive strategies, further research may be required to examine which

personality variables predict secondary control. It might be useful to examine how

the remaining personality variables (i.e., conscientiousness, agreeableness and

openness) relate to secondary control.

In regard to the proposed model of job satisfaction, the finding that

employees reported similar levels of control strategies in controllable and

uncontrollable situations suggests that changes need to be made to the model. As

such, rather than job autonomy, it is proposed that personality predicts the use of the

control strategies.

5.2.1.2 Summary

There was marginal support for the proposal that job autonomy predicts the

use of the control strategies. When this proposal was changed to be more specific,

the controllability of the difficulty did not influence the use of primary and secondary

control strategies. The finding that employees reported similar levels of primary and

secondary control for controllable and uncontrollable difficulties suggests that trait

control strategies may exist. Employees may have a set of control strategies that they

345

regularly use, irrespective of the controllability of the problem. As such, the model of

job satisfaction is changed so that personality predicts the control strategies rather

than job autonomy or the controllability of the situation.

5.2.1.3 Job Autonomy Influences the Adaptiveness of Primary and Secondary

Control

The relationship between the control strategies and job satisfaction is

expected to change depending on the level of job autonomy. This hypothesis is

based on the discrimination model, which proposes that primary control is the more

adaptive strategy in controllable situations and that secondary control is the more

adaptive strategy in uncontrollable situations. This model underlies the philosophy

of the serenity prayer; “Grant me the strength to change what I can, the patience to

accept what I cannot, and the wisdom to know the difference” (Thompson et al.,

1998, p. 587). An alternative model has also been developed, namely the

primacy/back-up model. This model proposes that primary control is more adaptive

than secondary control in controllable and relatively uncontrollable situations.

Previous empirical studies have supported the primacy/back-up model

(Thompson et al., 1996; 1994; 1993; 1998), however these studies were criticised for

their measurement of the controllability of the situation and the control strategies. In

studies one and two, the correlations between the control strategies and job

satisfaction for the low job autonomy group were compared to the correlations for

the high job autonomy group. Study one supported the primacy/back up model,

demonstrating that primary control was the most adaptive strategy for both the

346

academics and the supermarket workers. As the scale of primary control was

subsequently criticised, the proposal was re-tested with a revised scale in study two.

Study two did not support the primacy/back-up model or the discrimination model,

demonstrating that primary and secondary control strategies were not related to job

satisfaction.

As mentioned previously, in both of these studies, the hypotheses were not

consistent with the definition of the discrimination model or the primacy/back-up

model. The hypotheses were too general and as such were made more specific in

study three. In this study, the relationships between the control strategies for

controllable and uncontrollable difficulties and job satisfaction were examined.

The findings from study three refuted the discrimination model and supported

the primacy/back-up model. Primary control was more adaptive than secondary

control for controllable and uncontrollable difficulties. These findings suggest that

employees should use primary control whenever they face a difficulty at work, even

if it is uncontrollable.

The proposal that primary control is adaptive in uncontrollable situations is

difficult to explain as is it is assumed that they are likely to experience primary

control failure. It must be noted however that primary control failure was not

measured, and as such, the assumption that primary control in uncontrollable

situations results in primary control failure may be inaccurate. It is possible that

employees using primary control for uncontrollable difficulties report higher job

satisfaction because they implemented it successfully. As such, the successfulness of

the control strategies must be measured in future studies.

347

This proposal is incorporated in the revised model of job satisfaction. It is

now proposed that primary and secondary control strategies are not directly related to

job satisfaction, rather that they indirectly influence job satisfaction through the

successfulness of the control strategies. For example, suppose two employees report

having primary control strategies in their repertoire, however only one of them

implements primary control successfully. It would be expected that the employee

who is successfully implementing primary control would report higher job

satisfaction than the employee experiencing primary control failure. As such, the

successfulness of primary control may be a better predictor of job satisfaction than

primary control directly. It is expected that if employees successfully implement the

strategies, they will report higher job satisfaction.

5.2.1.4 Summary

The controllability of the difficulty did not influence the relationship between

the control strategies and job satisfaction. Even when the situation was

uncontrollable, primary control was the most adaptive strategy. These findings,

along with previous research, refute the discrimination model and support the

primacy/back-up model. The primacy/back-up model is difficult to explain as people

who use primary control in uncontrollable situations are expected to experience

primary control failure. Primary control failure was not measured however, and as

such, the successfulness of the control strategies must also be measured in future

studies. The model of job satisfaction is revised where it is proposed that the control

348

strategies are indirectly related to job satisfaction through the successfulness of the

control strategies.

5.2.2 Conclusion: Do the Control Strategies Mediate the Relationship

Between Job Autonomy and Job Satisfaction?

The above findings demonstrate that the control strategies do not explain the

relationship between job autonomy and job satisfaction. As such, the question of

why job autonomy is related to job satisfaction remains unanswered. One possibility

is self-determination. According to DeCharms (1968) and Deci and Ryan (1986),

humans have an innate need for competence and self-determination. Individuals

attempt to seek out situations that challenge them. They find these activities

rewarding and experience positive emotions such as enjoyment and excitement (Fay

& Frese, 2001; Ryan & Deci, 2001).

Another possibility is job status. Employees with high job autonomy have

jobs that generally involve more responsibility and job status than employees with

low job autonomy. Although few studies have examined the relationship between

job status and job satisfaction, one study has demonstrated that female employees

with higher job status tend to report higher job satisfaction than females employees

with lower job status (Secret & Green, 1998).

Another possibility is self-esteem. Employees with high job autonomy may

feel that their employer trusts them and thus may value themselves more than

employees with low job autonomy. Self-esteem has been shown to be positively

related to job satisfaction, where the average correlation is r = 0.26 (Judge & Bono,

349

2001). Thus, job status or self-esteem may mediate the relationship between job

autonomy and job satisfaction.

It is important that researchers continue to examine why job autonomy is

related to job satisfaction as the explanation offered by Karasek and Theorell (1990)

in the job demand-control model (Karasek & Theorell, 1990) is tautological and

vague. It is necessary that researchers understand the mechanism underlying the

proposal that job autonomy can reduce the influence of job demands.

5.2.3 2) Social Support at Work and Life Satisfaction Directly Predict Job

Satisfaction

The next major proposal of the model of job satisfaction is that social support

at work and life satisfaction predict job satisfaction.


Social support is expected to be directly related to job satisfaction. In study

two, supervisor support (r = 0.64, r = 0.46), and co-worker support (r = 0.39,

r = 0.42) were positively correlated with job satisfaction for the teachers and

academics, respectively. Furthermore, study three demonstrated that supervisor

support (r = 0.37) and co-worker support (r = 0.48) were moderately correlated with

job satisfaction. In regard to the proposed model of job satisfaction, social support at

work appears to be an important predictor.

350


Life satisfaction is expected to be positively related to job satisfaction. In

study one, life satisfaction was not related to job satisfaction for the supermarket

workers, however it was weakly related for the academics, r = 0.20. The results were

stronger in study two, where life satisfaction was moderately correlated with job

satisfaction for the academics, r = 0.38 and the teachers, r = 0.46. Study three also

demonstrated, using a general sample of employees that r = 0.46. On the basis of

these findings, it is concluded that life satisfaction is a direct predictor of job

satisfaction.

The positive correlations between life satisfaction and job satisfaction support

the spillover model, which proposes that satisfaction in one domain of an

individual’s life extends into other areas. Life satisfaction may spillover into job

satisfaction or job satisfaction may spillover into life satisfaction. Thus, employers

need to ensure that their employees are satisfied with all major areas of their lives,

not just the workplace. Employees also need to be satisfied with their standard of

living, their health, their personal relationships, their safety, and feeling part of their

community.

The levels of life satisfaction reported by the employees are particularly

interesting. According to Cummins (2000b), life satisfaction is held under

homeostatic control. Using two standard deviations to define the normative range, it

is predicted that the mean subjective life satisfaction of Western population samples

lay within the range 70-80%SM (Cummins, 1995). Consistently, all mean levels lay

within the 70-80%SM range, M = 78.31, M = 73.09, M = 74.20, M = 75.61,

351

M = 73.68. The finding that life satisfaction can be predicted within such a small

range is remarkable. Even the employees with low job autonomy (i.e., supermarket

workers and teachers) reported levels of life satisfaction that were within the

normative range.

The mechanisms that underlie this prediction involve personality, perceived

control, optimism, and self-esteem (Cummins, 2000b). More empirical studies are

needed to examine how these predictors are related to life satisfaction. These results

are not only important in developing a theory of life satisfaction, but these predictors

are important for employers attempting to increase job satisfaction.

5.2.3.3 Summary

Social support at work and life satisfaction both directly predicted job

satisfaction and are included in the revised model of job satisfaction. It is proposed

that social support influences job satisfaction and that life satisfaction and job

satisfaction influence each other.

5.2.4 3) The Control Strategies and Social Support at Work Moderate the

Relationship Between Work Difficulties and Job Satisfaction

5.2.4.1 Moderating Role of Control Strategies

Previous researchers have suggested that it is not the stressor that predicts job

satisfaction, but rather how the person deals with the stressor (Aldwin & Revenson,

1987, Ashford, 1988; Parkes, 1990, 1994; Perrewe & Zellars, 1999; Osipow, Doty &

352

Spokane, 1985). Thus, it is expected that if employees match their control strategies

to the situation, the negative influence of work difficulties on job satisfaction is

lessened.

Study three did not support this proposal however, demonstrating that

primary and secondary control did not act as moderators. These findings suggest that

even if employees match their control strategies to the situation, the negative

influence of work difficulties on job satisfaction is not lessened. As such, this part of

the model of job satisfaction requires revision.

An alternative proposal is offered. Rather than the control strategies

moderating the effect of work difficulties on job satisfaction, the successfulness of

the strategies may be important. Thus, it is expected that if employees successfully

implement the matching control strategies, the influence of work difficulties on job

satisfaction decreases. The model of job satisfaction is thus altered, where the

successfulness of primary control moderates the effect of controllable difficulties,

and the successfulness of secondary control moderates the effect of uncontrollable

difficulties.

5.2.4.2 Moderating Role of Social Support

In regard to social support at work, it is expected that co-worker and

supervisor support moderate the effect of work difficulties on job satisfaction. Study

two demonstrated that supervisor support, but not co-worker support moderated work

difficulties. As supervisor support increased, the relationship between work

difficulties and job satisfaction decreased.

353

Study three examined the types of social support, proposing that instrumental

support buffers the effects of controllable difficulties and emotional support buffers

the effects of uncontrollable difficulties. Marginal support was found for this

proposal, as co-worker instrumental support moderated the relationship between

controllable work difficulties and job satisfaction.

These findings in study two and three are somewhat inconsistent. This

inconsistency may be attributed to, in part, the measurement of work difficulties. In

study two, general work difficulties were measured by the item “how often do you

face difficulties at work?” This item is prone to errors as it is cognitively taxing,

requiring the employee to mentally average their work difficulties.

In study three, controllable and uncontrollable work difficulties were

measured by asking employees how often they face their most commonly occurring

controllable and uncontrollable difficulty. This scale was also criticised as it only

focused on one difficulty. One person may experience one difficulty all the time yet

rarely experience any other difficulties. Another person may experience ten

difficulties all the time. Using the current scale however, these respondents would

receive the same score.

It is thus concluded that further research is needed to examine the variables

that moderate the relationship between work difficulties and job satisfaction. This

research needs to measure work difficulties using an open-ended format, where

respondents list their top five difficulties at work and indicate how often they face

each one. This scale is not expected to be excessively taxing and can be

administered to a general sample of employees. Using this scale, it is expected that

354

instrumental support will moderate the effect of controllable difficulties and

emotional support will moderate the effect of uncontrollable difficulties.

5.2.4.3 Summary

There was no support for the moderating role of the control strategies on the

relationship between work difficulties and job satisfaction, however there was some

support for social support at work. As primary and secondary control did not act as

moderators, the model of job satisfaction was revised to examine the successfulness

of the control strategies. Although there was only minimal support for the

moderating role of social support, the indirect relationship is retained in the model of

job satisfaction as the operationalisation of work difficulties was criticised.

355

5.3 Revised Model of Job Satisfaction

This discussion has combined the results from three studies to develop a

revised model of job satisfaction (refer to Figure 11). The revisions are based on the

current results, findings from other research, or are purely speculative. The bolded

arrows and variables represent changes made to the model.

The first proposal that personality influences the use of the control strategies

is based on past studies of coping and personality. Although further research needs

to be conducted to determine which personality variables predict secondary control,

it is proposed that extroversion is positively related to primary control. Past research

has also demonstrated that extroversion is also positively related to job satisfaction

and life satisfaction, and that neuroticism is negatively related to job satisfaction and

life satisfaction. Thus, personality influences the control strategies, job satisfaction,

and life satisfaction.

Primary and secondary control strategies are no longer directly related to job

satisfaction; rather it is speculated that they are indirectly related to job satisfaction

through the successfulness of the control strategies. It is expected that if employees

successfully implement the strategies, they will report higher job satisfaction.

In addition to the successfulness of the control strategies, job autonomy and

social support at work are expected to be positively related to job satisfaction. Job

satisfaction is also expected to be reciprocally related to life satisfaction. These

relationships have all been demonstrated in the current findings.

Based on the findings of study three, controllable and uncontrollable

difficulties are expected to be negatively related to job satisfaction. The relationship

356

between work difficulties and job satisfaction is hypothesised to be moderated by the

successfulness of the control strategies and social support at work.

Specifically, it is expected that the successfulness of primary control

moderates the effect of controllable difficulties, and the successfulness of secondary

control moderates the effect of uncontrollable difficulties. The effect of controllable

work difficulties on job satisfaction is expected to be less when primary control is

successful, and the effect of uncontrollable difficulties on job satisfaction is expected

to be less when secondary control is successful.

Social support at work is also expected to moderate the effect of work

difficulties on job satisfaction. As demonstrated in study three, instrumental support

is expected to moderate the effect of controllable difficulties on job satisfaction. It is

also hypothesised that emotional support will moderate the effect of uncontrollable

difficulties on job satisfaction. Although this proposal was not supported in study

three, it is expected that when a new measure of work difficulties is used, it will be

supported.

357

Figure 11- Revised Model of Job Satisfaction

Controllable Difficulties



Emotional Support

Job Autonomy

Job Satisfaction

Secondary Control

Primary Control

Controllable Diff x Instrumental Support

Uncontrol Diff x EmotionalSupport

ControllableDiff x Success PC

Uncontrol Diff x Success of SC

Life Satisfaction

Success of PC

Personalit

Success of SC

Secondary Control

Primary Control

Success of

Success of

358

5.4 Conclusion

This thesis extended the job demand-control model (Karasek & Theorell,

1990), offering an alternative explanation for the relationship between job autonomy

and job satisfaction. A model of job satisfaction was developed which included job

autonomy, primary and secondary control, life satisfaction, work difficulties and

social support at work. The major proposal of this model was that job autonomy

influences the use and adaptiveness of primary and secondary control strategies.

Empirical testing of the model demonstrated that primary and secondary

control did not mediate the relationship between job autonomy and job satisfaction.

Employees reported using more primary control than secondary control for

controllable and uncontrollable difficulties. Furthermore, primary control was more

adaptive than secondary control for both types of difficulties.

Using these findings, a revised model of job satisfaction was developed. This

model proposes that rather than job autonomy, personality influences the use of the

control strategies. Furthermore, it is proposed that the control strategies do not

directly relate to job satisfaction, rather they are indirectly related through the

successfulness of the control strategies. In addition to these variables, job autonomy,

social support at work, life satisfaction and work difficulties continue to be included

as predictors of job satisfaction.

359

5.5 Final Word

This study developed a model of job satisfaction that offered an alternative

explanation to Karasek and Theorell (1990) for the relationship between job

autonomy and job satisfaction. Based on the life span theory of control (Heckhausen

& Schulz, 1995) and the discrimination model (Thompson et al., 1998), it was

proposed that employees with high job autonomy reported high job satisfaction

because they relied on more primary control and less secondary control strategies

than employees with low job autonomy. These proposals were not supported, as

primary control was the most commonly used and most adaptive strategy for

controllable and uncontrollable difficulties. These findings suggest that the serenity

prayer might best be changed to

“Grant me the strength to change the things I can…. and the things I cannot.”

360

5.6 References

Abouserie, R. (1996). Stress, coping strategies and job satisfaction in

university academic staff. Educational Psychology, 16, 49-57.

Agho, A.O., Price, J.L., & Mueller, C.W. (1992). Discriminant validity of

measures of job satisfaction, positive affectivity and negative affectivity. Journal of

Occupational and Organizational Psychology, 65, 185-196.

Alderfer, C.P. (1969). An empirical test of a new theory of human needs.

Organizational Behavior and Human Performance, 4, 142-175.

Aldwin, C.M., & Revenson, T.A. (1987). Does coping help? A

reexamination of the relation between coping and mental health. Journal of

Personality and Social Psychology, 53, 337-348.

Algera, J.A. (1980). Kenmerken van werk. Unpublished Ph.D. Thesis,

University of Leiden.

Ambrose, M.L., & Kulik, C.T. (1999). Old friends, new faces: Motivation

research in the 1990s. Journal of Management, 25, 231-237.

Armstrong, T.B. (1971). Job content and context factors related to

satisfaction for different occupational levels. Journal of Applied Psychology, 55, 57-

65.

Arnold, H.J., & House, R.J. (1980). Methodological and substantive

extensions to the Job Characteristics Model of motivation. Organizational Behavior

and Human Decision Processes, 25, 161-183.

361

Arvey, R.D., Bouchard, T.J., Segal, N.L., & Abraham, L.M. (1989). Job

satisfaction: Environmental and genetic components. Journal of Applied

Psychology, 74, 187-192.

Arvey, R.D., McCall, B.P., Bouchard, T.J., Taubman, P., & Cavanaugh, M.A.

(1994). Genetic influences on job satisfaction and work values. Personality and

Individual Differences, 17, 21-33.

Ashford, S.J. (1988). Individual strategies for coping with stress during

organisational transitions. Journal of Applied Behavioral Science, 24, 19-36.

Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable

distinction in social psychological research: Conceptual, strategic and statistical

considerations. Journal of Personality and Social Psychology, 51, 1173-1182.

Beehr, T.A. (1985). The role of social support in coping with organizational

stress. In T.A. Beehr., & R.S. Bhagat (Eds.), Human Stress and Cognition in

Organizations: An Integrated Perspective. (pp. 375-398). New York: Wiley

Beutell, N.L., & Wittig-Berman, U. (1999). Predictors of work-family

conflict and satisfaction with family, job, career, and life. Psychological Reports, 85,

893-903.

Bobko, P. (2001). Correlation and Regression: Applications for Industrial-

Organizational Psychology and Management. London: Sage.

Boey, K.W. (1998). Coping and family relationships in stress resistance: A

study of job satisfaction of nurses in Singapore. International Journal of Nursing

Studies, 35, 353-361.

362

Bogg, J., & Cooper, C. (1995). Job satisfaction, mental health, and

occupational stress among senior civil servants. Human Relations, 48, 327-341.

Boomsma, A. (1983). On the robustness of LISREL (maximum likelihood

estimation) against small sample sizes of nonnormality. Unpublished Ph.D. thesis,

University of Groningen, The Netherlands.

Boonzaier, B., Ficker, B., & Rust, B. (2001). A review of research on the Job

Characteristics Model and the attendant job diagnostic survey. South African

Journal of Business Management, 32, 11-35.

Bowman, G.D., & Stern, M. (1995). Adjustment to occupational stress: The

relationship of perceived control to effectiveness of coping strategies. Journal of

Counseling Psychology, 42, 294-303.

Brayfield, A.H., & Rothe, H.F. (1951). An index of job satisfaction. Journal

of Applied Psychology, 35, 307-311.

Breaugh, J.A. (1989). The Work Autonomy Scales: Additional validity

evidence. Human Relations, 42, 1033-1056.

Breaugh, J.A. (1998). The development of a new measure of global work

autonomy. Educational and Psychological Measurement, 58, 119-129.

Brebner, J. (2001). Personality and stress coping. Personality and Individual

Differences, 31, 317-327.

Brenner, V.C., Carmack, C.W., & Weinstein, M.G. (1971). An empirical test

of the motivation-hygiene theory. Journal of Accounting Research, 9, 359-366.

Burger, J.M., & Cooper, H.M. (1979). The desirability of control.

Motivation and Emotion, 3, 381-393.

363

Burke, R.J., & Greenglass, E.R. (2000). Hospital restructuring and nursing

staff well-being: The role of coping. International Journal of Stress Management, 7,

49-59.

Cahill, J. (1998). Stories of success and survival: The role of primary and

secondary control mechanisms in the subjective well-being of people on methadone

maintenance treatment programs. Unpublished Honours thesis, Deakin University,

Melbourne, Australia.

Cammann, C., Fichman, M., Jenkins, D., & Klesh, J. (1979). The Michigan

Organizational Assessment Questionnaire. Unpublished manuscript, University of

Michigan: Ann Arbor.

Campbell, J.P., & Pritchard, R.D. (1976). Motivation theory in Industrial and

Organizational Psychology. In M. Dunnette (Ed.), Handbook of Industrial and

Organizational Psychology. (pp 63-130). Chicago: Rand McNally.

Caplan, R.D., Cobb, S., French, J.R.P. Jr., Harrison, R.U., & Pinneau, S.R. Jr.

(1975). Job Demands and Worker Health. U.S. Department of Health, Education,

and Welfare Publication No. 75-160. Washington D.C.: U.S. Government Printing

Office, The Institute for Social Research.

Carson, K.D., Lanier, P.A., & Carson, P.P. (2001). A glimpse inside the

ivory tower: A cross-sectional comparison of work orientations in academia.

International Journal of Public Administration, 24, 479-492.

Carver, C.S., Scheier, M.F., & Weintraub, J.K. (1989). Assessing coping

strategies: A theoretically based approach. Journal of Personality and Social


364

Champoux, J.E. (1980). A three sample test of some extensions to the Job

Characteristics Model of work motivation. Academy of Management Journal, 23,

466-478.

Chay, Y.W. (1993). Social support, individual differences and well-being: a

study of small business entrepreneurs and employees. Journal of Occupational and

Organizational Psychology, 66, 285-302.

Chipperfield, J.G., Perry, R.P., & Menec, V.H. (1999). Primary and

secondary control-enhancing strategies. Journal of Aging and Health, 11, 517-539.

Clark, A.E. (1996). Job satisfaction in Britain. British Journal of Industrial

Relations, 34, 189-217.

Clark, A.E., & Oswald, A.J. (1996). Satisfaction and comparison income.

Journal of Public Economics, 69, 57-81.

Coakes, S.J., & Steed, L.G. (1999). SPSS: Analysis without Anguish.

Brisbane: John Wiley.

Cohen, J., & Cohen, P. (1983). Applied Multiple Regression/Correlation

Analysis for the Behavioral Sciences. Hillsdale, N.J: Lawrence Erlbaum.

Cohen, S., & Kamarack, T., Mermelstein, R., & Hoberman, H.M. (1985).

Measuring the functional components of social support. In I.G. Sarason., & B.R.

Sarason (Eds.), Social Support: Theory, Research and Applications. The Hague,

Netherlands: Nijhoff.

Cole, M. (1989). The politics of stress in teaching. In M.Cole., & S. Walker

(Eds.), Teaching and Stress. (pp.161-170). Milton Keynes: Open University Press.

365

Constantinople, A. (1967). Perceived instrumentality of the college as a

measure of attitudes towards college. Journal of Personality and Social Psychology,

5, 196-201.

Conway, V.J., & Terry, D.J. (1992). Appraised controllability as a moderator

of the effectiveness of different coping strategies: A test of the goodness of fit

hypothesis. Australian Journal of Psychology, 44, 1-7.

Cooper, C.L., Sloan, S., & Williams, S. (1988). Occupational Stress

Indicator. Windsor: NFER-Nelson.

Costa, P.T. Jr., & McCrae, R.R. (1985). The NEO Personality Inventory

Manual. Odessa, FL: Psychological Assessment Resources.

Costa, P.T. Jr., & McCrae, R.R. (1989). Personality as a life long

determinant of wellbeing. In L.Z. Malatesta., & C.E. Izzard (Eds.), Emotion in Adult

Development. (p.141-157). Beverly Hills, CA: Sage.

Costa, P.T. Jr., & McCrae, R.R. (1992). NEO PI-R professional manual.

Odessa, FL: Psychological Assessment Resources.

Costa, P.T. Jr., Somerfield, M.R., & McCrae, R.R. (1996). Personality and

coping: A reconceptualisation. In M. Zeidner., & N.S. Endler (Eds.), Handbook of

Coping: Theory, Research, Applications. New York: Wiley

Cousins, R. (2001). Predicting subjective quality of life: The contributions of

personality and perceived control. Unpublished Doctorate thesis, Deakin University,

Melbourne, Australia.

Cramer, D. (1995). Life and job satisfaction: A two-wave panel study. The

Journal of Psychology, 129, 261-268.

366

Cummins, R.A. (1995). On the trail of the gold standard for subjective well-

being. Social Indicators Research, 35, 179-200.

Cummins, R.A. (1996). The domain of life satisfaction: An attempt to order

chaos. Social Indicators Research, 38, 303-328.

Cummins, R.A. (1997). Comprehensive Quality of Life Scale-Adult Manual

(ComQol-A5). Melbourne, Australia: Deakin University.

Cummins, R.A. (2000a). Objective and subjective quality of life: An

interactive model. Social Indicators Research, 52, 55-72.

Cummins, R.A. (2000b). Normative life satisfaction: Measurement issues

and a homeostatic model. In B. Zumbo (Ed.), Social Indicators and Quality of Life

Research Methods: Methodological Development and Issues. Amsterdam: Klumer.

Cummins, R.A. (2002). Caveats to the Comprehensive Quality of Life Scale.

http://acqol.deakin.edu.au.

Cummins, R.A., Eckersley, R., Pallant, J., Van Vugt, J., Shelley, J., Pusey,

M., & Misajon, R. (2001). The Australian Unity Index of Well-Being: Survey 1:

Report 1. (on-line). Available: http://acqol.deakin.edu.au.

Cummins, R.A., & Nistico, H. (in press). Maintaining life satisfaction: The

role of positive cognitive bias. Journal of Happiness Studies

DeCharms, R. (1968). Personal Causation. New York: Academic Press.

Deci, E.L., & Ryan, R.M. (1985). Intrinsic Motivation and Self-

Determination in Human Behaviour. New York: Plenum Press.

de Jonge, J., Breukelen, G.J.P., Landeweerd, J.A., & Nijhuis, F.J.N. (1999a).

Comparing group and individual level assessments of job characteristics in testing

367

http://acqol.deakin.edu.au/

the Job Demand-Control Model: A multilevel approach. Human Relations, 52, 95-

97.

de Jonge, J., Mulder, M.J.G.P., & Nijhuis, F.J.N. (1999b). The incorporation

of different demand concepts in the job demand-control model: Effects of health care

professionals. Social Science and Medicine, 48, 1149-1160.

DeNeve, K.M. (1999). Happy as an extroverted clam? The role of

personality for subjective well-being. Psychological Science, 8, 141-144.

De Rijk, A.E., Le Blanc, P., Schaufeli, W.B., & de Jonge, J. (1998). Active

coping and need for control as moderators of the job demand-control model: Effects

on burnout. Journal of Occupational and Organizational Psychology, 71, 1-18.

Diener, E., Suh, E.M., Lucas, R.E., & Smith, H.L. (1999). Subjective well-

being: Three decades of progress. Psychological Bulletin, 125, 276-302.

Dollard, M.F., Winefield, H.R., Winefield, A.H., & Jonge, J.D. (2000).

Psychosocial service workers: A test of the demand control-support model. Journal

of Occupational and Organizational Psychology, 73. 501-510.

Doran, L.I., Stone, V.K., Brief, A.P., & George, J.M. (1991). Behavioral

intentions as predictors of job attitudes: The role of economic choice. Journal of

Applied Psychology, 76, 40-45.

Ducharme, L.J., & Martin, J. (2000). Unrewarding work, co-worker support,

and job satisfaction. Work and Occupations, 27, 223-244.

Dwyer, D.H., & Ganster, D.C. (1991). The effects of job demands and

control in employee attendance and satisfaction. Journal of Organizational Behavior,

12, 595-608.

368

Edwards, A.L. (1959). Edwards Personal Preference Schedule. New York:

Psychological Corporation.

Edwards, J.R., & O’Neill, R. (1999). The construct validity of scores on the

Ways of Coping Questionnaire: Confirmatory analysis of alternative factor

structures. Educational and Psychological Measurement, 58, 955-983.

Evans, B.K., & Fischer, D.G. (1992). A hierarchical model of participatory

decision-making, job autonomy and perceived control. Human Relations, 45, 1169-

1190.

Evans, M.G. (1991). The problem of analyzing multiplicative composites:

interactions revisited. American Psychologist, 46, 6-15.

Ewen, R.B. (1964). Some determinants of job satisfaction: A study of the

generalisability of Herzberg’s theory. Journal of Applied Psychology, 48, 161-163.

Fay, D., & Frese, M. (2000). Self-starting behavior at work: Toward a theory

of personal initiative. In J. Heckhausen (Ed.), Motivational Psychology of Human

Development. Amsterdam: Elservier Science.

Felce, D., & Perry, J. (1995). Quality of life: Its definition and measurement.

Research in Developmental Disabilities, 16, 51-74.

Fenlason, K.J., & Beehr, T.A. (1994). Social support and occupational stress:

Effects of talking to others. Journal of Organizational Behavior, 15, 157-175.

Ferris, K.R. (1977). A test of the expectancy theory of motivation in an

accounting environment. The Accounting Review, 3, 605-615.

369

Finlay, W., Martin, J.K., Roman, P.M., & Blum, T.C. (1995). Organizational

structure and job satisfaction: Do bureaucratic organizations produce more satisfied

employees? Administration and Society, 27, 427-451.

Fisher, C.D. (2000). Mood and emotions while working: missing pieces of

job satisfaction? Journal of Organizational Behavior, 21, 185-202.

Fisher, S. (1994). Stress in Academic Life: The Mental Assembly Line.

Buckingham: Open University Press.

Fletcher, B.C., & Jones, F. (1993). A refutation of Karasek’s Demand-

Discretion Model of Occupational Stress with a range of dependent measures.

Journal of Organizational Behavior, 14, 319-330.

Folkman, S., Chesney, M.A., Cooke, M., & Bocccellari, A. (1994).

Caregiver burden in the HIV-positive and HIV-negative partners of men with AIDS.

Journal of Consulting and Clinical Psychology, 62, 746-756.

Folkman, S., & Lazarus, R.S. (1980). An analysis of coping in a middle-aged

community sample. Journal of Health and Social Behavior, 21, 219-239.

Folkman, S., & Lazarus, R.S. (1985). If it changes, it must be a process:

Study of emotion and coping during three stages of a college examination. Journal

of Personality and Social Psychology, 48, 150-170.

Folkman, S., Lazarus, R.S., Dunkel-Schetter, C., DeLongis, A., & Gruen, R.J.

(1986). Dynamics of a stressful encounter: Cognitive appraisal, coping and

encounter outcomes. Journal of Personality and Social Psychology, 50, 992-1003.

370

Forsythe, C.J., & Compas, B.E. (1987). Interaction of cognitive appraisals of

stressful events and coping: Testing the goodness of fit hypothesis. Cognitive

Therapy and Research, 11, 473-485.

Fox, M.L., Dwyer, D.J., & Ganster, D.C. (1993). Effects of stressful job

demands and control on physiological and attitudinal outcomes in a hospital setting.

Academy of Management Journal, 36, 289-319.

Fox, S., & Feldman, G. (1988). Attention state and critical psychological

states as mediators between job dimensions and job outcomes. Human Relations, 41,

229-245.

Fried, Y. (1991). Meta-analytic comparisons of the Job Diagnostic Survey

and Job Characteristics Inventory as correlates of work satisfaction and performance.

Journal of Applied Psychology, 76, 690-697.

Fried, Y., & Ferris, G.R. (1987). The validity of the Job Characteristics

Model: A review and meta-analysis. Personnel Psychology, 40, 287-322.

Fried, Y., & Ferris, G.R. (1991). The dimensionality of job characteristics:

Some neglected issues. Journal of Applied Psychology, 71, 419-426.

Friedman, L.C., Nelson, D.V., Baer, P.E., Lane, M., Smith, F.E., & Dworkin,

R.J. (1992). The relationship of dispositional optimism, daily life stress, and

domestic environment to coping methods used by cancer patients. Journal of

Behavioral Medicine, 15, 127-141.

Friesen, D., Holdaway, E.A., & Rice, A.W. (1983). Satisfaction of school

principals with their work. Educational Administration Quarterly, 19, 35-58.

371

Frone, M.R., Russell, M., & Cooper, M.L. (1994). Relationship between job

and family satisfaction: Causal or noncausal covariation? Journal of Management,

20, 565-579.

Fung-kam, L. (1998). Job satisfaction and autonomy of Hong Kong

registered nurses. Journal of Advanced Nursing, 27, 355-364.

Ganster, D.C. (1989). Worker control and well-being: A review of research

in the workplace. In S. Sauter., J. Hurrell., & C. Cooper (Eds.), Job Control and

Worker Health. (pp. 3-24). Chichester, U.K: Wiley.

Ganster, D.C., Dwyer, D.J., & Fox, M.L. (2001). Explaining employees’

health care costs: A prospective examination of stressful job demands, personal

control, and physiological reactivity. Journal of Applied Psychology, 86, 954-964.

Ganster, D.C., & Fusilier, M.R. (1989). Control in the workplace. In C.L.

Cooper., & I. Robertson (Eds.), International Review of Industrial and

Organizational Psychology. (pp. 235-280). London: John Wiley

Gardner, G. (1977). Is there a valid test of Herzberg’s two-factor theory?

Journal of Occupational Psychology, 50, 197-204.

Gaziel, H.H. (1989). Determinants of perceived deficiency of autonomy

among elementary school administrators. Social Behavior and Personality, 17, 57-

66.

Geyer, P.D., & Daly, J.P. (1998). Predicting job satisfaction for relocated

workers: Interaction of relocating consequences and employee age. The Journal of


372

Gillet, B., & Schwab, D.P. (1975). Convergent and discriminant validities of

corresponding Job Descriptive Index and Minnesota Satisfaction Questionnaire

Scales. Journal of Applied Psychology, 60, 313-317.

Graham, W.K., & Balloun, J. (1973). An empirical test of Maslow’s need

hierarchy theory. Journal of Humanistic Psychology, 13, 97-108.

Grant, G.P., & Murray, C.E. (1999). Teaching in America: The Slow

Revolution. Cambridge, Mass: Harvard University Press.

Gunthert, K.C., Cohen, L.H., & Armeli, S. (1999). The role of neuroticism in

daily stress and coping. Journal of Personality and Social Psychology, 5, 1087-1100.

Hackman, J.R., & Oldham, G.R. (1975). Development of the Job Diagnostic

Survey. Journal of Applied Psychology, 60, 159-170.

Hackman, J.R., & Oldham, G.R. (1976). Motivation through the design of

work: Test of a theory. Organizational Behavior and Human Performance, 16, 250-

279.

Hackman, J.R., & Oldham, G.R. (1980). Work Redesign. Reading, Mass:

Addison-Wesley.

Hall, D.T., & Nougaim, K.E. (1968). An examination of Maslow’s need

hierarchy theory in an organizational setting. Organizational Behavior and Human

Performance, 3, 12-35.

Halliday, C.A., & Graham, S. (2000). “If I get locked up, I get locked up”:

Secondary control and adjustment among juvenile offenders. Personality and Social

Psychology Bulletin, 26, 548-559.

373

Hallqvist, J., Diderichsen, F., Theorell, T., Reuterwall, C., & Ahlbom, A.

(1998). Is the effect of job strain on myocardial infarction risk due to interaction

between high psychological demands and low decision latitude? Results from

Stockholm Heart Epidemiology Program (SHEEP). Social Science and Medicine,

46, 1405-1415.

Hart, P.M. (1999). Predicting employee life satisfaction: A coherent model

of personality, work and non-work experiences, and domain satisfaction. Journal of


Harwood, M.K., & Rice, R.W. (1992). An examination of the referent

selection processes underlying job satisfaction. Social Indicators Research, 27, 1-39.

Hatfield, J.D., Robinson, R.B., & Huseman, R.C. (1985). An empirical

evaluation of a test for assessing job satisfaction. Psychological Reports, 56, 39-45.

Hawkins, M.J., Hawkins, W.E., & Ryan, E.R. (1989). Self-actualisation as

related to age of faculty members at a large Midwestern University. Psychological

Reports, 65, 1120-1122.

Heckhausen, J., & Schulz, R. (1995). A life span theory of control.

Psychological Review, 102, 284-304.

Heckhausen, J., Schulz, R., & Wrosch, C. (1997). Technical Report:

Optimization in primary and secondary control: OPS Scales, a measurement

instrument for general and domain specific strategies of control and development

regulation. Berlin: Max Planck Institute for Human Development.

Heeps, L., Croft, C., & Cummins, R.A. (2000). The Primary and Secondary

Control Scale (2 nd ed) . Melbourne: Deakin University.

374

Herzberg, F. (1966). Work and the Nature of Man. Cleveland: World Publ

Co.

Herzberg, F., Mausner, B., & Snyderman, B.B. (1959). The Motivation to

Work. New York: Wiley.

Herzberg, F., Mausner, B., & Snyderman, B.B. (1993). The Motivation to

Work. New Brunswick, N.J: Transaction Publishers.

Highhouse, S., & Becker, A.S. (1993). Facet measures and global job

satisfaction. Journal of Business and Psychology, 8, 117-127.

Hill, M.D. (1986). A theoretical analysis of faculty job

satisfaction/dissatisfaction. Educational Research Quarterly, 10, 36-44.

Himle, D.P., & Jayaratne, S. (1991). Buffering effects of four social support

types on burnout among social workers. Social Work Research and Abstracts, 27,

22-28.

Hirschfeld, R.R. (2000). Does revising the intrinsic and extrinsic subscales of

the Minnesota Satisfaction Questionnaire short form make a difference? Educational

and Psychological Measurement, 60, 255-271.

Hoppock, R. (1935). Job Satisfaction. New York: Harper.

Howard, J.L., & Frink, D.D. (1996). The effects of organizational restructure

on employee satisfaction. Group and Organization Management, 21, 278-303.

Iaffaldano, M.T., & Muchinsky, P.M. (1985). Job satisfaction and job

performance: A meta-analysis. Human Relations, 44, 287-307.

375

Ironson, G.H., Smith, P.C., Brannick, M.T., Gibson, W.M., & Paul, K.B.

(1989). Construction of a job in general scale: A comparison of global, composite,

and specific measures. Journal of Applied Psychology, 74, 193-200.

Iverson, R.D., & Maguire, C. (2000). The relationship between job and life

satisfaction: Evidence from a remote mining community. Human Relations, 53, 807-

813.

Jansen, P.G.M., Kerkstra, A., Abud-Saad, H.H., & Van der Zee, J. (1996).

The effects of job characteristics and individual characteristics of job satisfaction and

burnout in community nursing. International Journal of Nursing Studies, 33, 407-

421.

Johnson, J.V., & Hall, E.M. (1988). Job strain, work place social support,

and cardiovascular disease: a cross-sectional study of a random sample of the

Swedish working population. American Journal of Public Health, 78, 1336-1342.

Johnson, J.V., & Hall, E.M. (1994). Social support in the work environment

and cardiovascular disease. In S.A. Schumaker., & S.M. Czajkowski (Eds.), Social

Support and Cardiovascular Disease. Plenum Series in Behavioral

Psychophysiology and Medicine. (pp. 145-166.) New York: Plenum Press.

Johnson, J.V., Hall, E.M., & Theorell, T. (1989). Combined effects of job

strain and social isolation on cardiovascular disease morbidity and mortality in a

random sample of the Swedish male working population. Scandinavian Journal of

Work, Environment and Health, 15, 271-279.

376

Johnson, S.M., Smith, P.C., & Tucker, S.M. (1982). Response format of the

Job Descriptive Index: Assessment of reliability and validity by the multitrait-

multimethod matrix. Journal of Applied Psychology, 67, 500-505.

Judge, T.A., & Bono, J.E. (2001). Relationship of core self-evaluations

traits- Self-esteem, generalised self-efficacy, locus of control and emotional stability

with job satisfaction and job performance: A meta-analysis. Journal of Applied


Judge, T.A., Bono, J.E., & Locke, E.A. (2000). Personality and job

satisfaction: The mediating role of job characteristics. Journal of Applied


Judge, T.A., & Locke, E.A. (1993). Effect of dysfunctional thought

processes on subjective well-being and job satisfaction. Journal of Applied


Judge, T.A., Locke, E.A., Durham, C.C., & Kluger, A.N. (1998).

Dispositional effects on job and life satisfaction: The role of cognitive evaluations.


Judge, T.A., & Watanabe, S. (1994). Individual differences in the nature of

the relationship between job and life satisfaction. Journal of Occupational and


Kandel, D.B., Davies, M., & Raveis, V.H. (1985). The stressfulness of daily

social roles for women: Marital, occupational and household roles. Journal of Health

and Social Behavior, 26, 64-78.

377

Karasek, R.A. Jr. (1979). Job demands, job decision latitude, and mental

strain: Implications for job redesign. Administrative Science Quarterly, 24, 335-357.

Karasek, R.A., Brisson, C., Kawakami, N., Houtman, I., Bongers, P., &

Amick, B. (1998). The Job Content Questionnaire (JCQ): An instrument for

internationally comparative assessments of psychosocial job characteristics. Journal

of Occupational Health Psychology, 3, 322-355.

Karasek, R.A., & Theorell, T. (1990). Healthy work: Stress, productivity

and the Reconstruction of Working Life. New York: Basic Books.

Karasek, R.A., Triantis, K.P., & Chaudry, S.S. (1982). Co-worker and

supervisor support as moderators of associations between task characteristics and

mental strain. Journal of Occupational Behaviour, 3, 181-200.

Kelly, J. (1992). Does job re-design theory explain job re-design outcomes?

Human Relations, 45, 753-775.

Keppel, G. (1991). Design and analysis: A Researchers Handbook (3 rd ed.) .

New Jersey: Prentice Hall.

King, N. (1970). Clarification and evaluation of the two factor theory of job

satisfaction. Psychological Bulletin, 74, 18-31.

Klecker, B.M., & Loadman, W.E. (1999). Male elementary school teachers

ratings of job satisfaction by years of teaching experience. Education, 119, 504-604

Koeske, G.F., Kirk, S.A., & Koeske, R.D. (1993). Coping with job stress:

Which strategies work best? Journal of Occupational and Organizational


378

Kohn, P.M., Hay, B.D., & Legere, J.J. (1994). Hassles, coping styles, and

negative well-being. Personality and Individual Differences, 17, 169-179.

Kunin, T. (1955). The construction of a new type of attitude measure.

Personnel Psychology, 8, 65-77.

Lahey, K.E., & Vihtelic, J.L. (2000). Finance faculty demographics, career

history, diversity, and job satisfaction. Financial Practice and Education, 10, 111-

123.

Landeweerd, J.A., & Boumans, N.P.G. (1994). The effect of work

dimensions and need for autonomy on nurses’ work satisfaction and health. Journal

of Occupational and Organizational Psychology, 67, 207-218.

Landry, M.B. (2000). The effects of life satisfaction and job satisfaction on

reference librarians and their work. Reference and user Services Quarterly, 40, 166-

178.

Landsbergis, P.A., Schnall, P.L., Deitz, D., Friedman, R., & Pickering, T.

(1992). The patterning of psychological attributes and distress by ‘job strain’ and

social support in a sample of working men. Journal of Behavioral Medicine, 15,

379-405.

LaRocco, J.M., House, J.S., & French, J.R.P. (1980). Social support,

occupational stress, and helath. Journal of Health and Social Behavior, 21, 202.

Laschinger, H.K.S., Finegan, J., & Shamian, J. (2001). Promoting nurses’

health: Effect of empowerment on job strain and work satisfaction. Nursing

Economics, 19, 42-52.

379

Latack, J.C. (1986). Coping with job stress: Measures and future direction

for scale development. Journal of Applied Psychology, 71, 377-385.

Lawler, E.E., & Suttle, J.L. (1972). A causal correlational test of the need

hierarchy concept. Organizational Behavior and Human Performance, 7, 265-287.

Lazarus, R.S., & Folkman, S. (1984). Stress, Appraisal and Coping. New

York: Springer.

Leong, F.T.L., & Dollinger, S.J. (1991). NEO Personality Inventory. In D.J.

Keyser., & R.C. Sweetland (Eds.), Test Critiques, Vol VIII. (p.527-539). Texas:

ProEd

Leung, K. (1997). Relationships among satisfaction, commitment and

performance: A group level analysis. Applied Psychology: An International Review,

46, 199-205.

Leung, T., Siu, O., & Spector, P.E. (2000). Faculty stressors, job satisfaction

and psychological distress among university teachers in Hong Kong: The role of

locus of control. International Journal of Stress Management, 7, 121-138

Locke, E.A. (1969). What is job satisfaction? Organizational Behavior and

Human Performance, 4, 309-336.

Locke, E.A. (1976). Nature and causes of job satisfaction. In M. Dunnette

(Ed.), Handbook of Industrial and Organizational Psychology. (pp. 1297-1350).

Chicago, IL: Rand McNally.

Loher, B.T., Noe, R.A., Moeller, N.L., & Fitzgerald, M.P. (1985). A meta-

analysis of the relation of job characteristics to job satisfaction. Journal of Applied


380

Lu, L. (1999). Work motivation, job stress and employees’ well-being.

Journal of Applied Management Studies, 8, 61-69.

Lykken, D., & Tellegen, A. (1996). Happiness is a stochastic phenomenon.

Psychological Science, 7, 186-189.

Ma, X., & MacMillan, R.B. (1999). Influences of workplace conditions on

teachers job satisfaction. Journal of Educational Research, 93, 39-47.

Maher, E., & Cummins, R.A. (2001). Subjective quality of life, perceived

control, and dispositional optimism among older people. Australasian Journal on

Aging, 20, 139-146.

Maher, E., & Cummins, R.A. (2002). Situation Specific Primary and

Secondary Control Scale. Melbourne: Deakin University.

Maher, E., Misajon, R., Heeps, L., & Cummins, R.A. (2001). Primary

Control and Secondary Control Scale (4 th ed.) . Melbourne: Deakin University.

Mannheim, B., Baruch, Y., & Tal, J. (1997). Alternative models for

antecedents and outcomes of work centrality and job satisfaction of high-tech

personnel. Human Relations, 50, 1537-1562.

Marriott, A., & Sexton, L. (1994). Components of job satisfaction in

psychiatric social workers. Health and Social Work, 19, 199-206.

Maslow, A.H. (1954). Motivation and Personality. New York: Harper.

Maslow, A.H. (1970). Motivation and Personality. New York: Harper &

Row.

McConatha, J.T., & Huba, H.M. (1999). Primary, secondary and emotional

control across adulthood. Current Psychology, 18, 164-170.

381

McCrae, R.R., & Costa, P.T. (1991). Adding Liebe und Arbeit: The full

Five-Factor Model and well-being. Personality and Social Psychology Bulletin, 17,

227-232.

McCrae, R.R., & Costa, P.T. Jr. (1992). Discriminant validity of NEO-PIR

Facet Scale. Educational Psychological Measurement, 52, 229-237.

Melamed, S., Kushnir, T., & Meir, E.I. (1991). Attentuating the impact of

job demands: Additive and interactive effects of perceived control and social

support. Journal of Vocational Behavior, 39, 40-53.

Michalos, A.C. (1985). Multiple Discrepancies Theory (MDT). Social

Indicators Research, 16, 347-413.

Miles, E.W., Patrick, S.L., & King, W.C. (1996). Job level as a systemic

variable in predicting the relationship between supervisory communication and job

satisfaction. Journal of Occupational and Organizational Psychology, 69, 277-285

Misajon, R. (2002). The Homeostatic Mechanism: Subjective Quality of Life

and Chronic Pain. Unpublished Ph.D. Thesis, Deakin University, Melbourne.

Misajon, R., & Cummins, R.A. (in press). Subjective quality of life, control

and the impact of chronic illness. Arthritis Care and Research.

Mitchell, T.R. (1974). Expectancy models of job satisfaction, occupational

preference and effort. Psychological Bulletin, 81, 1053-1077.

Moorman, R.H. (1993). The influence of cognitive and affective based job

satisfaction measures on the relationship between satisfaction and organizational

citizenship behavior. Human Relations, 46, 759-776.

382

Munro, L., Rodwell, J., & Harding, L. (1998). Assessing occupational stress

in psychiatric nurses using the full Job Strain Model: The value of social support to

nurses. International Journal of Nursing Studies, 35, 339-345.

Narayanan, L., Menon, S., & Spector, P.E. (1999). Stress in the workplace: a

comparison of gender and occupation. Journal of Organizational Behavior, 20, 63-

73.

Neher, A. (1991). Maslow’s theory of motivation: A critique. Journal of

Humanistic Psychology, 31, 89-112.

Nicolle, P.G. (1994). The effect of work dimensions and need for autonomy

on nurses’ work satisfaction and health. Journal of Occupational and Organizational


Niemann, Y.F., & Dovidio, J.F. (1998). Relationship of solo status, academic

rank, and perceived distinctiveness to job satisfaction of racial/ethnic minorities.


Norman, P., Collins, S., Conner, M., Martin, R., & Rance, J. (1995).

Attributions, cognitions, and coping styles: Teleworkers’ reactions to work-related

problems. Journal of Applied Social Psychology, 25, 117-128.

O’Driscoll, M.P., & Beehr, T.A. (2000). Moderating effects of perceived

control and need for clarity on the relationship between role stressors and employee

affective reactions. The Journal of Social Psychology, 140, 151-157.

Olsen, D. (1993). Work satisfaction and stress in the first and third year of

academics appointment. Journal of Higher Education, 64, 453-462.

383

Osipow, S.H., Doty, R.E., & Spokane, A.R. (1985). Occupational stress,

strain and coping across the life span. Journal of Vocational Behavior, 27, 98-108.

Osowiecki, D., & Compas, B.E. (1998). Psychological adjustment to cancer:

Control beliefs and coping in adult cancer patients. Cognitive Therapy and

Research, 22, 483-499.

Osowiecki, D.M., & Compas, B.E. (1999). A prospective study of coping,

perceived control and psychological adjustment to breast cancer. Cognitive Therapy

and Research, 23, 169-180.

Parahoo, K., & Barr, O. (1994). Job satisfaction of community nurses

working with people with a mental handicap. Journal of Advanced Nursing, 20,

1046-1055.

Park, C.L., Folkman, S., & Bostrom, A. (2001). Appraisals of controllability

and coping in caregivers and HIV+ men: Testing the goodness-of-fit hypothesis.

Journal of Consulting and Clinical Psychology, 69, 481-488.

Parker, S.K., & Sprigg, C.A. (1999). Minimising strain and maximising

learning: The role of job demands, job control and proactive personality. Journal of


Parkes, K.R. (1984). Locus of control, cognitive appraisal and coping in

stressful episodes. Journal of Personality and Social Psychology, 46, 655-668.

Parkes, K.R. (1989). Personal control in an occupational context. In A.

Steptoe., & A. Appels (Eds.), Stress, Personal Control, and Health (pp. 21-47).

Chichester: Wiley.

384

Parkes, K.R. (1990). Coping, negative affectivity and the work environment:

Additive and interactive predictors of mental health. Journal of Applied Psychology,

75, 399-409.

Parkes, K.R. (1994). Personality and coping as moderators of work stress

process: Models, methods and measures. Work and Stress, 8, 110-129.

Parkes, K.R., & von Rabenau, C. (1993). Work characteristics and wellbeing

among psychiatric health care staff. Journal of Community and Applied Social


Parsons, M.B. (1998). A review of procedural acceptability in organizational

behavior management. Journal of Organizational Behavior Management, 18, 173-

190.

Payne, R., & Fletcher, B.C. (1983). Job demands, supports, and constraints

as predictors of psychological strain among schoolteachers. Journal of Vocational

Behavior, 22, 136-147.

Pearson, C.A.L., & Chong, J. (1997). Contribution of job content and social

information in organizational commitment and job satisfaction: an exploration in a

Malaysian nursing context. Journal of Occupational and Organizational Psychology,

70, 357-

Pelfrene, E., Vlerick, P., Mak, R.P., De Smets, P., Kornitzers, M., & De

Backer, G. (2001). Scale reliability and validity of the Karasek ‘Job Demand

Control Support’ model in the Belstress study. Work and Stress, 15, 297-313.

385

Perrewe, P.L., & Zellars, K.I. (1999). An examination of attributions and

emotions in the transactional approach to the organizational stress process. Journal

of Organizational Behavior, 20, 739-752.

Porter, L.W. (1961). A study of perceived need satisfaction in bottom and

middle management jobs. Journal of Applied Psychology, 45, 1-10.

Porter, L.W. (1963). Job attitudes in management: II. Perceived importance

of needs as a function of job level. Journal of Applied Psychology, 47, 141-148.

Pritchard, R.D., & Sanders, M.S. (1973). The influence of valence,

instrumentality, and expectancy on effort and performance. Journal of Applied


Pulakos, E.D., & Schmitt, N. (1983). A longitudinal study of a valence

model approach for the prediction of job satisfaction of new employees. Journal of


Quinn, R.O., & Staines, G.L. (1979). The 1977 Quality of Employment

Survey: Descriptive statistics with comparison data from the 1960-1970 and the

1972-73 surveys. University of Michigan: Ann Arbor.

Rain, J.S., Lane, I.M., & Steiner, D.D. (1991). A current look at the job

satisfaction/life satisfaction relationship: Review and future considerations. Human

Relations, 44, 287-307.

Reinharth, L., & Wahba, R.A. (1976). A test of alternative models of

expectancy theory. Human Relations, 29, 257-272.

386

Renn, R.W., & Vandenberg, R.J. (1995). The critical psychological states:

An underrepresented component on Job Characteristics Model research. Journal of

Management, 21, 279-303.

Rice, R.W., McFarlin, D.B., & Bennett, D.,E. (1989). Standards of

comparison and job satisfaction. Journal of Applied Psychology, 74, 591-598.

Roberts, K.H., & Glick, W. (1981). The job characteristics approach to task

design: A critical review. Journal of Applied Psychology, 66, 193-217.

Roberts, K.H., Walter, G.A., & Miles, R.E. (1971), A factor analytic study of

job satisfaction items designed to measure Maslow need categories. Personnel


Roberts, S.M. (1995). Applicability of the goodness of fit hypothesis to

coping with daily hassles. Psychological Reports, 77, 943-954.

Roberts, T.B. (1982). “Comments on Mathes’s article”. Journal of

Humanistic Psychology, 22, 97-98.

Rodriguez, I., Bravo, M.J., Peiro, J.M., & Schaufeli, W. (2001). The

demands-control-support model, locus of control and job dissatisfaction: a

longitudinal study. Work and Stress, 15, 97-114.

Rothbaum, F., Weisz, J.R., & Synder, S.S. (1982). Changing the world and

changing the self: A two process model of perceived control. Journal of Personality

and Social Psychology Bulletin, 42, 5-37.

Roznowski, M. (1989). Examination of the measurement properties of the

Job Descriptive Index with experimental items. Journal of Applied Psychology, 74,

805-814.

387

Russell, D.W., Altmaier, E., & Van Velzen, D. (1987). Job-related stress,

social support, and burnout among classroom teachers. Journal of Applied


Ryan, E.M., & Deci, E.L. (2001). On happiness and human potentials: A

review of research on hedonic and eudaimonic well-being. Annual Review of

Psychology, 141-155.

Saklofske, D.H., & Kelly, I.W. (1995). Coping and personality.

Psychological Reports, 77, 481-482.

Salancik, G.R., & Pfeffer, J. (1977). An examination of need satisfaction

models of job attitudes. Administrative Science Quarterly, 22, 427-456.

Schappe, S.P. (1998). The influence of job satisfaction, organizational

commitment, and fairness perceptions on organizational citizenship behavior. The

Journal of Psychology, 132, 277-291.

Schaubroeck, J., & Merritt, D.E. (1997). Divergent effects of job control on

coping with work stressors: The key role of self-efficacy. Academy of Management

Journal, 40, 738-754.

Scheier, M.F., Weintraub, J.K., & Carver, C.S. (1986). Coping with stress:

Divergent strategies of optimists and pessimists. Journal of Personality and Social

Psychology, 51, 1257-1264.

Schmidt, G.L. (1976). Job satisfaction among secondary school

administrators. Educational Administration Quarterly, 12, 68-86.

388

Schonfeld, I.S. (2000). An updated look at depressive symptoms and job

satisfaction in first year women teachers. Journal of Occupational and


Schwab, D.P., Olian-Gottlieb, J.D., & Heneman, H.G. (1979). Between-

subjects expectancy theory research: A statistical review of studies predicting effort

and performance. Psychological Bulletin, 86, 139-147.

Schulz, R., & Heckhausen, J. (1996). A life-span model of successful aging.

American Psychologist, 51, 702-714.

Schulz, R., & Heckhausen, J. (1999). Aging, culture and control. Journal of

Gerontology: Psychological Sciences, 54, 139-145.

Schwartz, J.E., Neale, J., Marco, C., Shiffman, S.S., & Stone, A.A. (1999).

Does trait coping help? A momentary assessment approach to the evaluation of

traits. Journal of Personality and Social Psychology, 77, 360-369.

Secret, M., & Green, R.G. (1998). Occupational status differences among

three groups of married mothers. Journal of Women and Social Work, 13, 47-70.

Senate Employment, Education and Training Referees Committee. (1998). A

Class Act: Inquiry into the Status of the Teaching Profession. Canberra.

Sergiovanni, T. (1967). Factors which affect satisfaction and dissatisfaction

of teachers. Journal of Educational Administration, 5, 66-82

Shoura, M.M., & Singh, A. (1999). Motivation parameters for engineering

managers using Maslow’s theory. Journal of Management in Engineering, 15, 44-55.

Silver, P.F. (1987). Job satisfaction and dissatisfaction revisited. Educational

and Psychological Research, 7, 1-20.

389

Smith, C.A., Organ, D.W., & Near, J.P. (1983). Organizational citizenship

behavior: Its nature and antecedents. Journal of Applied Psychology, 68, 653-663.

Smith, C.S., Tisak, J., Hahn, S.E., & Schmeider, R.A. (1997). The

measurement of job control. Journal of Organizational Behavior, 18, 225-237.

Smith, P.C., Kendall, L.M., & Hulin, C.L. (1969). The Job Descriptive

Index. Bowling Green, OH: Department of Psychology, Bowling Green State

University.

Sobel, R.S. (1971). Tests of preperformance and postperformance models of

satisfaction with outcomes. Journal of Personality and Social Psychology, 19, 213-

321.

Spector, P.E. (1986). Perceived control by employees: A meta-analysis of

studies concerning autonomy and participation at work. Human Relations, 39, 1005-

1017.

Spector, P.E. (1987). Interactive effects of perceived control and job

stressors on affective reaction and health outcomes for clerical workers. Work and

Stress, 1, 155-162.

Spector, P.E. (1997). Job Satisfaction: Applications, Assessment, Causes and

Consequences. London: Sage Publications.

Spector, P.E., Dwyer, D.J., & Jex, S.M. (1988). Relation of job stressors to

affective, health and performance outcomes: A comparison of multiple data sources.


Spector, P.E., & O'Connell, B.J. (1994). The contribution of personality

traits, negative affectivity, locus of control and Type A to the subsequent reports of

390

job stressors and job strains. Journal of Occupational and Organizational


Spokes, A. (1998). Subjective quality of life of people with schizophrenia:

Examination of perceived control and positive illusions. Unpublished Honours

Thesis, Deakin University, Melbourne, Australia.

Staw, B.M., Bell, N.E., & Clausen, J.A. (1986). The dispositional approach

to job attitudes: A lifetime longitudinal test. Administrative Science Quarterly, 31,

56-77.

Staw, B.M., & Ross, J. (1985). Stability in the midst of change: A

dispositional approach to job attitudes. Journal of Applied Psychology, 70, 469-480.

Stone, A.A., & Neale, J.M. (1984). New measure of daily coping:

Development and preliminary results. Journal of Personality and Social Psychology,

46, 892-906.

Tabachnick, B.G., & Fidell, L.S. (1996). Using Multivariate Statistics (3 rd

ed.). Northridge: Harper Collins.

Taber, T.D., & Taylor, E. (1990). A review and evaluation of the

psychometric properties of Job Diagnostic Survey. Personnel Psychology, 43, 467-

512.

Tait, M., Padgett, MY., & Baldwin, T.T. (1989). Job and life satisfaction: A

reevaluation of the strength of the relationship and gender effects as a function of the

date of the study. Journal of Applied Psychology, 74, 502-507.

Teacher Stress in Victoria: A survey of teachers’ views: The Report (1990).

Victoria: Ministry of Education.

391

Teas, R.K. (1981). A within-subject analysis of valence models of job

preference and anticipated satisfaction. Journal of Occupational Psychology, 54,

109-124.

Terry, D.J., & Hynes, G.J. (1998). Adjustment to a low-control situation:

Reexamining the role of coping responses. Journal of Personality and Social

Psychology, 74, 1078-1092.

Terry, D.J., Nielsen, M., & Perchard, L. (1999). Effects of work stress on

psychological well-being and job satisfaction: The stress-buffering role of social

support. Australian Journal of Psychology, 45, 168-175.

Thurstone, L.L., & Jones, L.V. (1957). The rational origin for measuring

subjective values. Journal of the American Statistical Association, 52, 458-471.

Tiegs, R.B., Tetrick L.E., & Fried, Y. (1992). Growth need strength and

context satisfactions as moderators of the relations of the Job Characteristics Model.

Journal of Management, 18, 575-593.

Tinsley, H.E.A. (1994). NEO Personality Inventory-Revised. In D.J.

Keyser., & R.C. Sweetland (Eds.), Test Critiques, Vol X. (p. 443-456). Texas:

ProEd.

Thompson, S.C., Collins, M.A., Newcomb, M.D., & Hunt, W. (1996). On

fighting versus accepting stressful circumstances: Primary and secondary control

among HIV-positive men in prison. Journal of Personality and Social Psychology,

70, 1307-1317.

392

Thompson, S.C., Nanni, C., & Levine, A. (1994). Primary versus secondary

and central versus consequence-related control in HIV-Positive men. Journal of


Thompson, S.C., Sobolew-Shubin, A., Galbraith, M.E., Schwankovsky, L., &

Cruzen, D. (1993). Maintaining perceptions of control: Finding perceived control in

low-control circumstances. Journal of Personality and Social Psychology Bulletin,

64, 293-304.

Thompson, S.C., Thomas, C., Rickabaugh, C.A., Tantamjarik, P., Otsuki, T.,

Pan, D., Garcia, B.F., & Sinar, E. (1998). Primary and secondary control over age-

related changes in physical appearance. Journal of Personality, 66, 583-605.

Thurber, C.A., & Weisz, J.R. (1997). “You can try or you can just give up”:

The impact of perceived control and coping style on childhood homesickness.

Developmental Psychology, 33, 508-517.

Thurstone, L.L., & Jones, L.V. (1957). The rational origin for measuring

subjective values. Journal of the American Statistical Association, 52, 458-471.

Tokar, D.M., & Subich, L.M. (1997). Relative contributions of congruence

and personality dimensions to job satisfaction. Journal of Vocational Behavior, 50,

482-491.

Unden, A. (1996). Social support at work and its relationship to absenteeism.

Work and Stress, 10, 46-61.

Valentiner, D.P., Holahan, C.J., & Moos, R.H. (1994). Social support,

appraisals of event controllability and coping: An integrative model. Journal of


393

Van Der Doef, M., & Maes, S. (1999). The job demand-control (-support)

model and psychological well-being: a review of 20 years of empirical research.


Van Eerde, W., & Thierry, H. (1996). Vroom’s expectancy models and

work-related criteria: A meta-analysis. Journal of Applied Psychology, 81, 575-586.

Vitaliano, P.P., DeWolfe, D., Maiuro, R.D., Russo, J., & Katon, W. (1990).

Appraised changeability of a stressor as a modifier of the relationship between

coping and depression: A test of the hypothesis of fit. Journal of Personality and

Social Psychology, 59, 582-592.

Vitaliano, P.P., Russo, J., Carr, J.E., Maiuro, R.D., & Becker, J. (1985). The

Ways of Coping Checklist: Revision and psychometric properties. Multivariate

Behavioral Research, 20, 3-26.

Vroom, V.H. (1964). Work and motivation. New York: Wiley.

Wahba, M.A., & House, R.J. (1976). Expectancy theory in work and

motivation: Some logical and methodological issues. Human Relations, 27, 121-147.

Wahba, M.A., & Bridwell, L.G. (1976). Maslow reconsidered: A review of

the research on the need hierarchy theory. Organizational Behavior and Human

Performance, 15, 212-240.

Wall, T.D., Clegg, C.W., & Jackson, P.R. (1978). An evaluation of the Job

Characteristics Model. Journal of Occupational Psychology, 51, 183-196.

Wall, T.D., Jackson, P.R., Mullarkey, S., & Parker, S.K. (1996). The

demands-control model of job strain: A more specific test. Journal of Occupational

and Organizational Psychology, 69, 153-166.

394

Wanous, J.P., Reichers, A.E., & Hudy, M.J. (1997). Overall job satisfaction:

How good are single-item measures? Journal of Applied Psychology, 82, 247-252.

Warr, P., Cook, J., & Wall, T. (1979). Scales for the measurement of some

work attitudes and aspects of psychological well-being. Journal of Occupational


Warr, P.B. (1990). Decision latitude, job demands, and employee well-being.


Waters, L.K., & Waters, C.W. (1969). Correlates of job satisfaction and job

dissatisfaction among female clerical workers. Journal of Applied Psychology, 53,

388-391.

Weiss, D.J., Dawis, R.V., England,G.W., & Lofquist, L.H. (1967). Manual

for the Minnesota Satisfaction Questionnaire. Minneapolis: Industrial Relations

Centre, University of Minnesota.

Weiss, H.M., Nicholas, J.P., & Daus, C.S. (1999). An examination of the

joint effect of affective experienced and job beliefs on job satisfaction and variations

in affective experiences over time. Organizational Behavior and Human Decision

Processes, 78, 1-24.

Wernimont, P.F. (1966). Intrinsic and extrinsic factors in job satisfaction.


White, R.W. (1959). Motivation reconsidered: The concept of competence.

Psychological Review, 66, 297-333.

Whitsett, D.A., & Winslow, E.K. (1967). An analysis of studies critical of

the motivation-hygiene theory. Personnel Psychology, 391-414.

395

Wicker, F.W., Brown, G., Wiehe, J.A., Hagen, A.S., & Reed, J.L. (1993). On

reconsidering Maslow: An examination of the Deprivation/Domination proposition.

Journal of Research in Personality, 27, 118-133.

Wicker, F.W., & Wiehe, J.A. (1999). An experimental study of Maslow’s

deprivation-domination proposition. Perceptual and Motor Skills, 88, 1356-1358.

Wilensky, H.L. (1960). Work, careers and social integration. International

Social Science Journal, 12, 543-560.

Williams, S., & Cooper, C.L. (1996). Occupational Stress Indicator, Version

2. Harrogate, North Yorkshire: RAD.

Winefield, A.H. (2000). Stress in academe: Some recent research findings.

In D.T. Kenny., & J.G. Carlson., F.J. McGuigan., J.L. Sheppard (Eds.), Stress and

Health: Research and Clinical Applications. (pp. 437-446). Amsterdam: Harwood

Academic.

Witt, L.A., Andrews, M.C., & Kacmar, K.M. (2000). The role of

participation in decision-making in the organisational politics-job satisfaction

relationship. Human Relations, 53, 341- 350.

Wong, K.S., & Cheuk, W.H., & Rosen, S. (2000). The influence of job stress

and supervisor support on negative affects and job satisfaction in kindergarten

principals. Journal of Social Behavior and Personality, 15, 85-99.

Wright, T.A., Bennett, K.K., & Dun, T. (1999). Life and job satisfaction.

Psychological Reports, 84, 1025-1028.

396

Wyrwich, K.W., Nienaber, N.A., Tierney, W.M., & Wolinsky, F.D. (1999).

Linking clinical relevance and statistical significance in evaluating individual

changes in health-related quality of life. Medical Care, 37, 469-478.

397

5.7 Appendices

398

Appendix A- Plain Language Statement for Study One

Dear Sir/Madam,

My name is Elise Maher, and I am completing my Ph.D. in Psychology at Deakin University. As part of my studies, I am undertaking a research project under the supervision of Professor Robert Cummins, a researcher in the School of Psychology. This study is investigating job satisfaction and control. The study aims to provide useful information about how the amount of choice that an employee has influences job satisfaction. The results will provide information that will enhance programs that increase job satisfaction.

You are invited to participate in this research. If you agree, you will be asked to complete the enclosed questionnaire. Any information you provide will be anonymous and confidential. Only group results will be reported and no individuals will be identified. Upon completion of the study, data will be secured in a locked cabinet in the School of Psychology, Deakin University, for a minimum period of six years from the date of publication.

The questionnaire should take around 30 minutes to complete and your participation would be greatly appreciated. Examples of questions are: "My work is boring", "In my job, I can choose the amount I earn", "I am not a worrier" and "How satisfied are you with your close relationships with family or friends". You are free to withdraw at any time during the study in which event your participation in the research study will immediately cease and any information obtained will not be used. You are free to refuse to answer any questions.

Following the completion of the study, I will provide your employer with a summary of the results. If you would like a copy of the summary sent directly to you, please contact Elise Maher.

If you have any further questions regarding the study, please contact:

Elise Maher on 9251 7153 or Email: [email protected] you can contact Professor Robert Cummins on 9244-6845 or Email: [email protected].

If you are happy to be involved in this study, please complete the enclosed questionnaire and return it in the reply-paid envelope supplied (i.e., NO STAMP NEEDED).________________________________________________________________________

Should you have any concerns about the conduct of this research project, please contact the Secretary, Deakin University Ethics Committee, Research Services, Deakin University, 221 Burwood Highway, BURWOOD, VIC, 3125, Tel (03) 9251 7123

399

mailto:[email protected]


Appendix B- Job Autonomy Scale used in Study One (Revision of Ganster,

1989, cited in Dwyer & Ganster, 1991)

Indicate your agreement with the following 13 statements by ticking () a number ranging from 1 to 10, where 1= Do not agree at all, and 10= Agree completely. All of the statements begin with “In my job, I can choose….”

In my job, I can choose:

1) In my job, I can choose among a variety of tasks or projects to do.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

2) In my job, I can choose the order in which I do my work.0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

3) In my job, I can choose how quickly I work.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

4) In my job, I can choose how I schedule my rest breaks.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

5) In my job, I can choose the physical conditions of my workstation.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

6) In my job, I can choose when I interact with others.0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

7) In my job, I can choose the amount I earn.

400

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

8) In my job, I can choose the number of times I am interrupted at work.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

9) In my job, I can choose how my work is evaluated.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

10) In my job, I can choose the quality of my work.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

11) In my job, I can choose the policies and procedures in my work unit.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

12) In my job, I can choose among a variety of methods to complete my work.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

13) In my job, I can choose how much work I get done.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

14) In general, how much are you able to influence work and work-related matters?0 1 2 3 4 5 6 7 8 9 10

Very Little Very Much

401

Appendix C- Primary and Secondary Control Scale used in Study One

(Revision of Heeps et al., 2000)

Indicate your agreement with the following statements by selecting a number ranging from 1 to 10, where 1=Do not agree at all, and 10= Agree Completely

1) When a goal that I have at work is difficult to reach, I think about different ways to achieve it.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

2) When I want something at work to change, I think I can make it happen.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

3) When a work task really matter to me, I think about it a lot.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

4) When I really want to reach a goal at work, I believe I can achieve it.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

5) When faced with a difficult work situation, I believe I can overcome it.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

402

Indicate your agreement with the following statements by selecting a number ranging from 1 to 10 where 1=Do not agree at all, and 10=Agree completely. All the statements begin with “When something bad happens that I cannot change…”

When something bad happens at work that I cannot change:

When something bad happens at work that I cannot change1) I can see that something good will come of it.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

When something bad happens at work that I cannot change2) I remember you can’t always get what you want.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

When something bad happens at work that I cannot change3) I know things will work out OK in the end.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

When something bad happens at work that I cannot change4) I remember I am better off than many other people.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

When something bad happens at work that I cannot change5) I remember I have already accomplished a lot in life.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

When something bad happens at work that I cannot change6) I remember the success of my family or friends.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

When something bad happens at work that I cannot change

403

7) I think nice thoughts to take my mind off it.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

When something bad happens at work that I cannot change8) I remind myself the situation will change if I am just patient.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

When something bad happens at work that I cannot change9) I tell myself it doesn’t matter.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

When something bad happens at work that I cannot change10) I think about my success in other areas.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

When something bad happens at work that I cannot change11) I don’t feel disappointed because I knew it might happen.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

When something bad happens at work that I cannot change12) I can see it was not my fault.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

When something bad happens at work that I cannot change13) I ignore it by thinking about other things.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

When something bad happens at work that I cannot change

404

14) I realise I didn’t need to control it anyway.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

Appendix D- Job Satisfaction Scale used in Study One (Revision of Roznowski,

1989)

Indicate your agreement with the following 15 statements by ticking () a number ranging from 1 to 10, where 1= Do not agree at all, and 10= Agree completely.

1) My work is boring.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

2) My co-workers are stupid.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

3) My pay is bad.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

4) My supervisors know how to supervise.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

5) There is a good chance for promotion in my job.

0 1 2 3 4 5 6 7 8 9 10

405

Do not agree at all

Agree Completely

6) My co-workers are responsible.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

7) I am well-paid.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

8) My work is dull.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

9) There is a fairly good chance for promotion in my job.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

10) My supervisors are bad.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

11) My work is interesting.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

12) My pay is unfair.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

13) My supervisors are annoying.

0 1 2 3 4 5 6 7 8 9 10

Do not Agree

406

agree at all Completely

14) My co-workers are a waste of time.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

15) There are good opportunities for advancement in my job.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

407

Appendix E- Life Satisfaction Scale used in Study One (Cummins, 1997)

Please tick () the box that best describes how SATISFIED you are with each area. Do not spend too much time on any one question. There are no right or wrong answers.

1) How Satisfied are you with the THINGS YOU OWN ?

0 1 2 3 4 5 6 7 8 9 10

Completely dissatisfied

Completely satisfied

2) How Satisfied are you with your HEALTH?

0 1 2 3 4 5 6 7 8 9 10



3) How Satisfied are you with what you ACHIEVE IN LIFE ?

0 1 2 3 4 5 6 7 8 9 10



4) How Satisfied are you with your CLOSE RELATIONSHIPS with FAMILY or FRIENDS ?

0 1 2 3 4 5 6 7 8 9 10



5) How Satisfied are you with HOW SAFE YOU FEEL ?

0 1 2 3 4 5 6 7 8 9 10



6) How Satisfied are you with feeling part of your COMMUNITY?

0 1 2 3 4 5 6 7 8 9 10



7) How Satisfied are you with YOUR OWN HAPPINESS ?

0 1 2 3 4 5 6 7 8 9 10

408



Appendix F- Personality Scale used in Study One (Costa & McCrae, 1992)

This questionnaire contains 24 statements. Read each statement carefully. For each statement tick () the box with the response that best represents your opinion.

1) I am not a worrier.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

2) I like to have a lot of people around me.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

3) I often feel inferior to others.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

4) I laugh easily.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

5) When I'm under a great deal of stress, sometimes I feel like I'm going to pieces.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

6) I don't consider myself especially light hearted.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

409

7) I rarely feel lonely or blue.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

8) I really enjoy talking to people.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

9) I often feel tense and jittery.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

10) I like to be where the action is.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

11) Sometimes I feel completely worthless.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

12) I usually prefer to do things alone.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

13) I rarely feel fearful or anxious.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

14) I often feel as if I am bursting with energy.

0 1 2 3 4 5 6 7 8 9 10

Do not Agree

410


16) I often get angry at the way people treat me.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

16) I am a cheerful, high-spirited person.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

17) Too often, when things go wrong, I get discouraged and feel like giving up.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

18) I am not a cheerful optimist.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

19) I am seldom sad or depressed.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

20) My life is fast-paced. 0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

21) I often feel helpless and want someone else to solve my problems.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

22) I am a very active person.

0 1 2 3 4 5 6 7 8 9 10

Do not Agree

411


23) At times I have been so ashamed I just want to hide.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

24) I would rather go my own way than be a leader of others.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree Completely

412

Appendix G-Levels of Job Satisfaction reported by various occupational groups

No Author Occupation Scale N Job Satisfaction (%SM)

1 Leung et al., (2000)

Academics Job Satisfaction Scale (OSI-2; Williams & Cooper, 1996)

106 57.6

2 Judge et al., (1998)

Physicians Brayfield & Rothe (1951)

164 72.67

3 Judge et al., (2000)

General Brayfield & Rothe (1951)

384107

69.6, 76.53

4 Renn & Vandenberg (1995)

Management, counseling, administration

Job Diagnostic Survey (Hackman & Oldham, 1975)

188 64.8

5 Hackman & Oldham (1975)

General Job Diagnostic Survey (Hackman & Oldham, 1975)

658 60.3

6 Dollard et al., (2000)

Public sector welfare workers

Global measure 786 64

7 Wall, Jackson, Mullarkey & Parker (1996)

Manufacturingemployees

Warr, Cook & Wall (1979) intrinsic satisfaction

1451 51.33

8 O’Driscoll & Beehr (2000)

Accounting firms

Facet measure (12 facets)

236 68

9 Fletcher & Jones (1993)

Manual and non-manual workers

Global measure 501 (manual)788 (non-manual)

60 58

10 Fox et al., 1993

Nurses Faces scale (Kunin, 1955)

136 76.33

11 Mannheim, Baruch & Tal (1997)

Managerial personnel


39 70

12 Beutell & Wittig-Berman (1999)

MBA students Global measure (4 items)

177 52.5

13 Agho, Price & Mueller (1992)

Employees of medical centre

Brayfield & Rothe (1951)

550 62.04

14 Fisher (2000) General Faces scale (Kunin, 124 87.0

413

1955)

15 Howard & Frink (1996)

Managers, administrators police, firefighters, labourers


248 65.83

16 Jansen, Kerkstra, Abud-Saad & Van der Zee (1996)

Nurses Facet measure (Algera, 1980)

355 nurses, 92 nurse auxiliaries

63.75 (nurses) & 68.0 (nurse auxiliaries)

17 Laschinger et al., (2001)

Nurses Job Diagnostic Survey (Hackman & Oldham, 1975)

600 44.75

18 Schonfeld (2000)

Graduate teachers

Quinnes & Staines (1979)-Global

184 80

19 Finlay, Martin, Romas & Blum (1995)

Administrators Hoppock (1935) 169 78.5

20 Ma & Macmillan (1999)

Teachers Global items (4) 2,202 80.25

21 Geyer & Daly (1998)

Private sector organisation

Global items (3) 174 57

22 Schappe (1998)

Insurance company workers

Minnesota Satisfaction Questionnaire (Weiss et al., 1967)

150 71.50

23 Parsons (1998)

Nurses Brayfield & Rothe (1951)

47 71.36

24 Pearson & Chong (1997)

Nurses Job Diagnostic Survey (Hackman & Oldham, 1975)

286 68.88

25 Miles, Patrick & King (1996)

Manufacturing employees

Job perception scale (Hatfield, Robinson & Huseman, 1985)

713 57.5

26 Witt, Andrews & Kacmar (2000)

Public sector organisation

Hoppock (1935) 1251 63.75

27 Bogg & Copper

Civil servants, executives

OSI (Cooper, Sloan & Williams, 1988)

1051 1056

55.6 62.0

414

(1995)28 Weiss,

Nicholas & Daus (1999)

Managers Faces Scale (Kunin, 1955), and global scale

24 82

29 Moorman (1993)

Manufacturers Brayfield & Rothe (1951)

62.25 71

30 Parahoo & Barr (1994)

Nurses Global measure (1 item)

35 75

31 Spector, Dwyer & Jex (1988)

Secretaries Michigan Orgnizational Assessment (Cammann, Fichman, Jenkins & Klesh, 1979)

155 78

32 Spector & O’Connell (1994)

University Graduates

Michigan Orgnizational Assessment (Cammann et al., 1979)

66.66

33 Wong et al., (2000)

Kindergarten principles

Global item 108 54

34 Klecker & Loadman (1999)

Teachers National follow-up survey of teachers education graduates- 7 facets

1874 68.16

35 Marriott & Sexton (1994)

Social workers Global measure- 1 item

188 66.9

36 Frone, Russell & Cooper (1994)

Random Global Scale (Kandel, Davies & Raveis, 1985)

631 73.33

MEAN (N= 41)

66.75

415

Appendix H- Primary and Secondary Control Scale for Study Two (Maher et

al., 2001)

The following items assess the difficulties that you have at work.

Please tick () the areas in which you experience difficulties in your work.

Time management (making time to do everything) Motivation Interpersonal relationships (colleagues, or supervisors) Nature of the Work Promotions Pay Other

b) How often do you have difficulty doing something at work? (ie., think of the examples given above, or other difficulties you may have had at work)

0 1 2 3 4 5 6 7 8 9 10

Never All the time

416

Here are ways people deal with difficult situations at work.

How often have you had these thoughts when facing a difficulty at work OVER THE PAST WEEK?

I thought………

1) It will work out okay in the end.

0 1 2 3 4 5 6 7 8 9 10

Never Every time

2) I knew it would happen.

0 1 2 3 4 5 6 7 8 9 10

Never Every time

3) I can't always get what I want.

0 1 2 3 4 5 6 7 8 9 10

Never Every time

4) It doesn’t matter.

0 1 2 3 4 5 6 7 8 9 10

Never Every time

5) I am better off than many other people.

0 1 2 3 4 5 6 7 8 9 10

Never Every time

6) It was not my fault.0 1 2 3 4 5 6 7 8 9 10

Never Every time

417

Here are other ways people deal with difficult situations at work.

How often have you done these things when facing a difficulty at workOVER THE PAST WEEK?

7) I looked for different ways to overcome it.

0 1 2 3 4 5 6 7 8 9 10

Never Every time

8) I kept trying. 0 1 2 3 4 5 6 7 8 9 10

Never Every time

9) I told someone about it.

0 1 2 3 4 5 6 7 8 9 10

Never Every time

10) I worked to overcome it.

0 1 2 3 4 5 6 7 8 9 10

Never Every time

11) I thought of the success of my family or friends.

0 1 2 3 4 5 6 7 8 9 10

Never Every time

12) I thought about my success in other areas.

0 1 2 3 4 5 6 7 8 9 10

Never Every time

13) I did something different, like going for a walk.

0 1 2 3 4 5 6 7 8 9 10

Never Every time

418

14) I ignored it. 0 1 2 3 4 5 6 7 8 9 10

Never Every time

15) I worked out how to remove obstacles.

0 1 2 3 4 5 6 7 8 9 10

Never Every time

16) I looked for something else that was positive in the situation.

0 1 2 3 4 5 6 7 8 9 10

Never Every time

419

Appendix I- Plain Language Statement for Study Two

Dear Sir/Madam,

My name is Elise Maher, and I am completing my Ph.D. in Psychology at Deakin University. As part of my studies, I am undertaking a research project under the supervision of Professor Robert Cummins, a researcher in the School of Psychology. This study is investigating job satisfaction and control. The study aims to provide useful information about how the amount of choice that an employee has influences job satisfaction. The results will provide information that will enhance programs that increase job satisfaction.


The questionnaire should take around 30 minutes to complete and your participation would be greatly appreciated. Examples of questions are: "I am satisfied with the praise I get for doing a good job", "How much can you choose the amount that you earn", "I am not a worrier", "How satisfied are you with your close relationships with family or friends", and “Which management style do you prefer”. You are free to withdraw at any time during the study in which event your participation in the research study will immediately cease and any information obtained will not be used. You are free to refuse to answer any questions.

Following the completion of the study, I will provide your employer with a summary of the best coping strategies. If you would like a copy of the summary sent directly to you, please contact Elise Maher. If you have any further questions regarding the study, please contact: Elise Maher on 9251 7153 or Email: [email protected], or you can contact Professor Robert Cummins on 9244-6845 or Email: [email protected]

If you are happy to be involved in this study, please complete the enclosed questionnaire and return it in the reply-paid envelope supplied (i.e., NO STAMP NEEDED).Thank you very much for your time. _________________________________________________________________


420


Appendix J- Job Autonomy Scale for Study Two (Hackman & Oldham, 1975)

The following 3 items assess how much freedom you have at your work. For each item, please tick () a number ranging from 1 to 10, where 1= Do not agree at all, and 10= Agree completely.

1) In my job, I can decide on my own how to go about doing my work.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

2) In my job, I have the chance to use my personal initiative and judgement in carrying out the work.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

3) In my job, I have considerable opportunity for independence and freedom.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

421

Appendix K- Need for Autonomy Scale for Study Two (de Rijk et al., 1998)

The following 4 items assess how important it is for you to do certain things at work. Please tick a box ranging from 1=Not important at all to 10=Could not be more important.

1) How important is it for you to set the pace of your tasks at work?

0 1 2 3 4 5 6 7 8 9 10

Not important at all

Could not be more important

2) How important is it for you to have control over what you do at work and the way that you do it?

0 1 2 3 4 5 6 7 8 9 10



3) How important is it for you to do your own planning at work?

0 1 2 3 4 5 6 7 8 9 10



4) How important is it for you to give orders at work instead of receiving them?

0 1 2 3 4 5 6 7 8 9 10



422

Appendix L- Job Satisfaction Scale for Study Two (Weiss et al., 1967)

Please indicate how satisfied you are with the following aspects of your work. Please tick a box ranging from (1) Very dissatisfied to (10) Very satisfied.

On my present job, this is how I feel about…..

1) Being able to keep busy all the time.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

2) The chance to work alone on the job.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

3) The chance to do different things from time to time.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

4) The chance to be “somebody” in the community.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

5) The way my boss handles his/her work.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

6) The competence of my supervisor in making decisions.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

423

7) Being able to do things that don’t go against my conscience.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

8) The way my job provides for steady employment.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

9) The chance to do things for other people.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

10) The chance to tell people what to do.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

11) The chance to do something that makes use of my abilities.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

12) The way company politics are put into practice.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

13) My pay and the amount of work I do.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

14) The chances for advancement on my job.

424

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

15) The freedom to use my own judgement.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

16) The chance to try my own methods of doing the job.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

17) The working conditions.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

18) The way my co-workers get along with each other.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

19) The praise I get for doing a good job.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

20) The feeling of accomplishment I get from the job.

0 1 2 3 4 5 6 7 8 9 10

Very dissatisfied

Very satisfied

425

Appendix M- Social Support Scale for Study Two (Revision of Karasek &

Theorell, 1990)

The following 8 questions ask about your supervisor and your co-workers. Please circle a number 1= Not true at all to 10= Could not be more true.

1) My supervisor shows concern for me.

0 1 2 3 4 5 6 7 8 9 10

Not true at all

Could not be more true

2) My supervisor pays attention to me.

0 1 2 3 4 5 6 7 8 9 10

Not true at all


3) My supervisor is helpful getting work done.

0 1 2 3 4 5 6 7 8 9 10

Not true at all


4) My supervisor creates a good teamwork environment for me.

0 1 2 3 4 5 6 7 8 9 10

Not true at all


5) My co-workers are friendly to me.

0 1 2 3 4 5 6 7 8 9 10

Not true at all


6) My co-workers are helpful to me.

0 1 2 3 4 5 6 7 8 9 10

Not true at all


7) My co-workers are personally interested in me.

426

0 1 2 3 4 5 6 7 8 9 10

Not true at all


8) My co-workers are competent.

0 1 2 3 4 5 6 7 8 9 10

Not true at all


Appendix N-Plain Language Statement used in Study Three

Dear Sir/Madam,

My name is Elise Maher, and I am completing my Ph.D. in Psychology at Deakin University. As part of my studies, I am undertaking a research project under the supervision of Professor Robert Cummins, a researcher in the School of Psychology. This study is investigating job satisfaction and coping. The study aims to provide useful information about the best type of coping strategies that workers should use. The results will provide information that will enhance programs that increase job satisfaction.


The questionnaire should take around 20 minutes to complete and your participation would be greatly appreciated. Examples of questions are: "I can decide on my own about how to go about doing my work", "Your co-workers really care about you", "What type of difficulties do you face at work?" and "How satisfied are you with your close relationships with family or friends." You are free to withdraw up until you have returned the survey, in which event your participation in the research study will immediately cease and any information obtained will not be used. You are free to refuse to answer any questions.

Following the completion of the study, I am happy to provide you a summary of the best coping strategies. If you would like a copy of the summary or if you have any further questions regarding the study, please contact:

Elise Maher on (03) 9251 7153 or Email: [email protected], or Professor Robert Cummins on (03) 9244-6845 or Email: [email protected].

427



If you are happy to be involved in this study, please complete the enclosed questionnaire and return it in the reply-paid envelope supplied (i.e., NO STAMP NEEDED).

Thank you very much for your time. __________________________________________________________________


428

Appendix O- Primary and Secondary Control Scale for Study Three (Maher &

Cummins, 2002)

People may experience several kinds of difficulties in their work. They can control some of them, but not others.

For example, Worker ‘A’, a teacher, can control difficulties involving students, parents and time management. They cannot however control difficulties involving

school policies and work times.

Another example, Worker ‘B’ a supermarket operator, can control difficulties involving customers and co-workers. They cannot control difficulties involving pay,

promotion and holiday leave.

1) Tick the difficulties you experience at work that you CAN CONTROL.

Difficulties with Supervisor(s) Difficulties with Promotion Difficulties with Co-worker(s) Difficulties with Time Management Difficulties with Kind of work you do Difficulties with Motivation Difficulties with Pay Difficulties with Work Times

Difficulties with Work-place rules Difficulties with Amount of Work Other………………………………………………………………………

2) Consider the difficulty that you experience MOST OFTEN, and which you CAN CONTROL.

How often do you experience this difficulty?

1 2 3 4 Rarely Sometimes Often Always

429

3) When you face this difficulty that you CAN CONTROL, how often do you do the following? a) Discuss solutions with the people involved

0 1 2 3 4Never Rarely Sometimes Often Always

b) Think that the difficulty doesn’t matter


c) Think that this difficulty will work out okay in the end


d) Choose a solution and act on it


e) Think that I knew this difficulty would happen


f) Think that I can’t always get what I want


g) Work harder


h) Think that I am better off than many other people


i) Think that this difficulty is not my fault

430


j) Keep trying


k) Tell someone about this difficulty to make me feel better


l) Think of the success of my family/friends


m) Think about my success in other areas


n) Do something different, like going for a walk


o) Ignore this difficulty


p) Look for something else that is positive in the situation


q) Other …………………………..(please specify)


431

The following questions examine difficulties that you CANNOT CONTROL

4) Tick the difficulties you experience at work that you CANNOT CONTROL.

Difficulties with Supervisor(s) Difficulties with Promotion Difficulties with Co-worker(s) Difficulties with Time Management Difficulties with Kind of work you do Difficulties with Motivation Difficulties with Pay Difficulties with Work Times

Difficulties with Work-place rules Difficulties with Amount of Work Other………………………………………………………………………

5) Consider the difficulty that you experience MOST OFTEN, and which you CANNOT CONTROL. How often do you experience this difficulty?

1 2 3 4 Rarely Sometimes Often Always

6) When you face this difficulty that you CANNOT CONTROL, how often do you do the following?

a) Discuss solutions with the people involved

0 1 2 3 4

432

Never Rarely Sometimes Often Always

b) Think that the difficulty doesn’t matter


c) Think that this difficulty will work out okay in the end


d) Choose a solution and act on it


e) Think that I knew this difficulty would happen


f) Think that I can’t always get what I want


g) Work harder


h) Think that I am better off than many other people


i) Think that this difficulty is not my fault


j) Keep trying

0 1 2 3 4

433

Never Rarely Sometimes Often Always

k) Tell someone about this difficulty to make me feel better


l) Think of the success of my family/friends


m) Think about my success in other areas


n) Do something different, like going for a walk


o) Ignore this difficulty


p) Look for something else that is positive in the situation


q) Other ………………(please specify)


434

Appendix P- Life Satisfaction Scale for Study 3 (Cummins et al., 2001)

1) How satisfied are you with your standard of living?

0 1 2 3 4 5 6 7 8 9 10



2) How satisfied are you with your health?

0 1 2 3 4 5 6 7 8 9 10



3) How satisfied are you with what you achieve in life?

0 1 2 3 4 5 6 7 8 9 10



4) How satisfied are you with your personal relationships?

0 1 2 3 4 5 6 7 8 9 10



5) How satisfied are you with how safe you feel?0 1 2 3 4 5 6 7 8 9 10



6) How satisfied are you with feeling part of your community?0 1 2 3 4 5 6 7 8 9 10



7) How satisfied are you with your future security?

0 1 2 3 4 5 6 7 8 9 10



435

436

Appendix Q- Social Support Scale for Study 3 (Ducharme & Martin, 2000)

The following 6 questions ask about your co-workers. Please circle a number ranging from 0 to 10, where 0= Do not agree at all and 10= Agree completely.

1) My co-workers really care about me.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

2) I feel close to my co-workers.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

3) My co-workers take a personal interest in me.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

4) My co-workers assist with unusual work problems.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

5) My co-workers are helpful in getting the job done.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

6) My co-workers give useful advice on job problems.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

437

The following 6 questions ask about your supervisor(s). For each item, please circle a number ranging from 0 to 10, where 0= Do not agree at all, and 10= Agree

completely

1) My supervisor really cares about me.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

2) I feel close to my supervisor.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

3) My supervisor takes a personal interest in me.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

4) My supervisor assists with unusual work problems.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

5) My supervisor is helpful in getting the job done.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

6) My supervisor gives useful advice on job problems.

0 1 2 3 4 5 6 7 8 9 10

Do not agree at all

Agree completely

438

439

Date post:	22-Oct-2014
Category:	Documents
Upload:	reenakini674717
View:	121 times
Download:	1 times

Thesis Maher e 2

Documents