The Universality of Subjective Wellbeing Indicators: A Multi-disciplinary and Multi-national...

THE UNIVERSALITY OF SUBJECTIVE WELLBEING INDICATORS

Social Indicators Research Series

Volume 16

General Editor:

ALEX C. MICHALOS University of Northern British Columbia,

Prince George, Canada

Editors:

ED DIENER University of Illinois, Champaign, U.S.A.

WOLFGANG GLATZER 1. W. Goethe University, Frankfurt am Main, Germany

TORBJORN MOUM University of Oslo, Norway

MIRJAM A.G. SPRANGERS University of Amsterdam, The Netherlands

JOACHIM VOGEL Central Bureau of Statistics, Stockholm, Sweden

RUUTVEENHOVEN Erasmus University, Rotterdam, The Netherlands

This new series aims to provide a public forum for single treatises and collections of papers on social indicators research that are too long to be published in our journal Social Indicators Research. Like the journal, the book series deals with statistical assessments of the quality of life from a broad perspective. It welcomes the research on a wide variety of substantive areas, including health, crime, housing, education, family life, leisure activities, transportation, mobility, economics, work, religion and environmental issues. These areas of research will focus on the impact of key issues such as health on the overall quality of life and vice versa. An international review board, consisting of Ruut Veenhoven, Joachim Vogel, Ed Diener, Torbjorn Mourn, MirjamA.G. Sprangers and Wolfgang Glatzer, will ensure the high quality of the series as a whole.

THE UNIVERSALITY OF SUBJECTIVE WELLBEING INDICATORS

A Multi-disciplinary and Multi-national Perspective

Editedby

ELEONORA GULLONE School 0/ Psychology, Psychiatry and Psychological Medicine,

Monash University, Australia

and

ROBERT A. CUMMINS School 0/ Psychology,

Deakin University, Australia

SPRINGER SCIENCE+BUSINESS MEDIA, B.V.

A C.I.P. Catalogue record for this book is available from the Library of Congress.

ISBN 978-94-010-3960-4 ISBN 978-94-010-0271-4 (eBook) DOI 10.1007/978-94-010-0271-4

Printed on acid-free paper

All Rights Reserved © 2002 Springer Science+Business Media Dordrecht

Originally published by Kluwer Academic Publishers in 2002 Softcover reprint of the hardcover 18t edition 2002

No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording

or otherwise, without written permission from the Publisher, with the exception of any material supplied specifiCally for the purpose ofbeing entered

and executed on a computer system, for exclusive use by the purchaser of the work.

TABLE OF CONTENTS

Biographical summaries of contributors to this volume

Editorial- The Universality of Subjective Wellbeing 5 Eleanora Gullone and Robert A. Cummins

A model of subjective well-being homeostasis: the role of personality 7 Robert A. Cummins, Eleonora Gullone and Anna L.D. Lau

Subjective quality of life: the affective dimension 47 Paul Bramston

Theoretical basis for the measurement of quality of life 63 Jouko Kajanoja

The concept of life satisfaction across cultures: exploring its diverse meaning and relation to economic wealth 8 I Joar Vittersf'l, Espen Roysamb and Ed Diener

Quality of life in residential care 105 Edward Helmes and Lynne Austin

The universality of quality of life: an empirical approach using the WHOQOL 129 MJ. Power, M Bullinger and the WHOQOL group

Predicting quality of work life: from work conditions to self-regulation 151 Georgia Pomaki and Stan Maes

Substitution, buffers and subjective well-being: a hierarchical approach 175 Anna Nieboer and Siegwart Lindenberg

BIOGRAPHICAL SUMMARIES OF CONTRIBUTORS TO THIS VOLUME

Lynne Austin is currently completing a Masters Degree in Clinical Geropsychology at Edith Cowan University. Her specific area of interest is in the impact of residential care on quality of life.

Paul Bramston, PhD, has been in the Psychology Department at the University of Southern Queensland for 10 years and has published widely in the area of emotions such as stress and depression in aged and disabled populations. More recently he has turned his attention to positive emotions such as happiness and wellbeing in these groups. He has also completed some ground breaking explorations into sense of community and belonging among people who are aged and/or disabled.

Robert A. Cummins, PhD, Professor of Psychology, holds a Personal Chair at Deakin University in Melbourne. He is an international authority on quality of life, most particularly in the area of subjective wellbeing, and has published widely in this area. He is a Fellow of both the Australian Psychological Society and the International Society for Quality of Life Studies. He also directs the Australian Centre on Quality of Life at Deakin University (http://acqol.deakin.edu.au) and is Convenor of the International Wellbeing Group which is attempting to evolve a cross-culturally valid index to measure population subjective wellbeing.

Ed Diener, PhD, is Alumni Professor of Psychology at the University of Illinois. Dr. Diener received his Ph.D. at the University of Washington in Seattle in 1974, and has been a faculty member at the University of Illinois ever since. He is pastpresident of the International Society of Quality of Life Studies, and is pastpresident of the Society of Personality and Social Psychology (and Division 8 of APA). Professor Diener is the editor of the Journal of Personality and Social Psychology (1998-2003) and is also editor of Journal of Happiness Studies. He won the 2000 Distinguished Researcher Award from the International Society of Quality of Life Studies, and a distinguished alumni award from California State University at Fresno. In 200 I, Professor Diener was selected to speak in the American Psychological Association's Distinguished Lecture Series. Diener has about 140 publications, of which about 90 are in the area of subjective well-being (SWB). He was listed as the second most published author in the first 30 years of the Journal of Personality and Social Psychology, and currently has 41 publications in this journal. Professor Diener is the Chair of the Positive Experience Network of the Positive Psychology Initiative. His citation count is approximately 5,000.

Eleonora Gullone, PhD, is Associate Professor in the Department of Psychology at Monash University, Melbourne, Australia. She has published extensively in the area of child and adolescent emotional development with particular a focus on fear, anxiety, and depression. More recently, she has shifted her research attention to psychological well-being. Her specific interests within this broad area of Positive Psychology focus upon the promotion of human well-being through interaction with the natural environment and non-human species.

1

E. Gullone and RA. Cummins (eds.}. The Universality a/Subjective Wellbeing Indicators, 1-3. © 2002 Kluwer Academic Publishers.

2 BIOGRAPHICAL SUMMARIES

Edward Helmes, PhD is an Associate Professor in Psychology. He joined the School of Psychology at James Cook University in 2001 after 5 years with the School of Psychology, Edith Cowan University that in tum fol1owed his earlier career of work with hospitals and the University of Western Ontario in Canada. His publications are in several areas, including ageing, neuropsychology, personality, and psychological assessment. His interests in the area of this book arise from his work on the definition and measurement of psychological constructs, such as quality oflife.

Jouko Kajanoja (PhD, Po1.Sc.) is senior researcher and director of administration in the Government Institute for Economic Research. He is docent at Helsinki University. His research has combined sociology, philosophy and economics. The main topics have been theories of society, measurement of welfare and characteristics of welfare state. More recently, his focus has been on social capital and relations between good life and working life.

Anna Lau, PhD, is Assistant Professor in the Department of Rehabilitation Sciences, at the Hong Kong Polytechnic University. Her main research interests and publications are in the conceptualization and cross-cultural investigation of QOL. She is the Hong Kong coordinator and project investigator for the International Wel1-Being group. She is also involved in several other collaborative projects on QOL that include a comparison of subjective wellbeing between Hong Kong and Australia, and the study of factors or interventions that influence the QOL of elderly people in Hong Kong.

Siegwart Lindenberg (PhD Harvard 1971) is Professor of Theoretical Sociology at the University of Groningen. He is member of the Royal Netherlands Academy of Arts and Sciences. His research interests prominently include informal group processes (he is principle investigator of the Groningen research project "The Future of Community") and the development of theories that underlie human action in the context of groups and social networks (the so-called theory of "social rationality" which includes social production function theory of human goals and framing theory). His recent publications include "Social Rationality Versus Rational Egoism". Pp. 635-668 in: J. Turner (ed.) Handbook of Sociological Theory. New York: Kluwer AcademiclPlenum, 2001. More detailed information is available from his homepage.

Stan Maes, PhD, is Professor of Health Psychology at Leiden University, the Netherlands. He produced over 200 scientific publications including 5 books, in various languages, concerning health promotion in school and work settings, doctor patient communication and psychological aspects of/and interventions in patients with chronic diseases. His current work focuses on the development of a new model for the prediction of health behaviour (the health behaviour goal model).

Anna Nieboer, PhD, studied Sociology at the University of Groningen in the Netherlands. After her graduation in 1992 she worked at the Interuniversity Center for Social Science and Methodology (lCS). In 1997 she finished her dissertation on the consequences of major life-events in the elderly, after which she worked as a coordinator of the research program "The Future of Community", also conducted at the ICS. Since January 2001 she is an assistant professor at the department of Health Policy, Economics and Organization of Care at Maastricht University. Her main

BIOGRAPHICAL SUMMARIES 3

research interests concern the organisation of health care, quality of life and the international comparison of health system performance.

Georgia Pomaki is completing her PhD in Health Psychology at Leiden University, the Netherlands. Her thesis concerns predictors of employee health and wellness, and considers both the work environment as well as individual differences. Her theoretical perspective is based in goal theory and her focus is on employee personal goals. To date, her empirical work has centered on teachers and health care employees.

Mick Power, PhD, is Professor of Clinical Psychology at the University of Edinburgh, where he directs the doctoral training programme in clinical psychology, and he is a Consultant Clinical Psychologist at the Royal Edinburgh Hospital. He has been one of the co-ordinators of the development of the World Health Organization's measure of quality of life, the WHOQOL. He is the current coordinator of a major cross-cultural programme to adapt the WHOQOL for use with older adults and to study active ageing and its impact on quality of life.

Espen RllJysamb, PhD, is currently a senior researcher at the Norwegian Institute of Public Health, and Associate Professor at the University of Oslo. He received his PhD in health psychology in 1997, and has been a visiting researcher at University of Michigan (USA), Virginia Commonwealth University (USA) and Birzeit University (Palestine). His research interests include adolescent risk behaviour and mental health, subjective well-being and emotions, heritability, and structural equation modelling.

Joar VittersllJ, PhD, Associate Professor in Psychology, has been in the Department of Psychology at the University of Troms0, Norway, since 1999. His research interests are in the area of outdoor recreation and subjective well-being.

Acknowledgements: The editors would like to thank Ms Kathy Tempini for her assistance informatting the final document.

ELEONORA GULLONE AND ROBERT A. CUMMINS

EDITORIAL

THE UNIVERSALITY OF SUBJECTIVE WELLBEING

When researchers study subjective wellbeing they implicitly make assumptions of universality. In order to embed their research within the current literature, they often ignore the fact, that the reports they cite for this purpose, have employed samples heterogeneous on such variables as culture, age, and wealth, all of which are known to exert powerful influences on sUbjective wellbeing in their own right. Researchers also often ignore the fact that the scales they employ have no demonstrated validity, reliability or sensitivity for the specific populations being sampled for their study. And, often, they have little direct evidence that the construct of subjective wellbeing, as described in the literature, is shared by their research participants.

So, to what extent is it valid to combine the findings from different studies in order to create conceptual models? The answer is, we do not know.

We assume that there must be a hierarchy of importance. Clearly some sample characteristics are more likely to disrupt universality than others, and some outcome variables are more universal than others. But we are generally uncertain about the relative ordering of such influences for the purpose of combining study findings. This lack of understanding has significant implications for advancing knowledge in this area of study. If study groupings combine highly heterogeneous samples or variables, valid difference variance in criterion variables may be hidden from view, and overly generalised conclusions may result.

This topic of universality is addressed by the chapters that follow. The authors take a variety of approaches, which combine to provide a useful and insightful overview of the issues involved. At the most fundamental level is the problem of definition and, as any researcher in the area of subjective wellbeing knows, the variety of definitions can be a great source of frustration. Clearly authors find it hard to agree on the precise character of this construct. However, Kajanoja in his chapter, argues this is not only beneficial but also inevitable. It is beneficial because it extends our understanding of what is possible, and is inevitable since a universal definition is, in his view, an unattainable goa\.

There are, however, some points of broad agreement emerging from the literature in regard to the construction and character of subjective wellbeing. One of these is that it can be considered a composite of affective and cognitive processes. Bramston explores and elucidates the evidence for these dual processes as they interact with one another to create our perception of wellbeing. A related area of emerging consensus, and likely universality, is that the affective and cognitive

5

E. Gullone and R.A. Cummins (em.). The Universality o/Subjective Wellbeing Indicators. 5-6. © 2002 Kluwer Academic Publishers.

6 EDITORIAL

processes are determinedly managed by the brain to normally create a stable, and predictably positive, outlook on life. How these processes might operate, is addressed by three chapters. Cummins, Gullone and Lau argue for a homeostatic system comprising personality, a set of cognitive buffers, and needs. Nieboer and Lindenberg, on the other hand, while also invoking the concept of buffers, argue for their Social Production Function Theory. Here, subjective wellbeing is maintained through the utilisation of substituted means when the normal means of wellbeing maintenance fail through loss. The final chapter in this set addresses self-regulation of wellbeing in the workplace. Pomaki and Maes use Motivational Systems Theory as the basis for describing the strategies and processes involved in the pursuit of work goals. The achievement of such personal goals are then proposed as core predictors of general wellness.

It is interesting to note that the assumption of active wellbeing maintenance, implicit in each of these three chapters, points to another aspect of universality. This is the generally poor relationship between publicly observed objective life quality and subjective wellbeing. Such understanding is highly relevant to researchers or practitioners using subjective wellbeing as a dependent variable, and the chapter by Helmes and Austin exemplifies the difference in the context of old age and the onset of disability. Due to the processes of adaptation, the effects of changed life circumstances on wellbeing are perceived very differently by people who have experienced the change and others who are passive observers.

All such measurement, of course, depends on instrumentation, and the extent to which two well-known scales provide universally valid and reliable data is addressed by two chapters. Power, Bulinger, and The WHOQOL Group discuss the WHOQOL instruments, while Vitterso, Roysamb and Diener discuss the Satisfaction with Life Scale. Both groups of authors take a cross-cultural approach and both report aspects of universality within this context. Both also find limitations, as would be expected, but these reports do provide hope that a truly universal scale is a possible achievement. This, indeed, is the aim of the International Wellbeing Group (ht1;p:llacgol.deakin.edu.au). Their progress can be monitored through the website.

All in all, this collection of essays provides a good deal of conceptual coverage on issues relating to the universality of subjective wellbeing. While it is clear that understanding in this area is still in its infancy, this collection underscores the importance of better understanding the universality of subjective well-being and its underlying mechanisms and processes. This pursuit will undoubtedly contribute to the development of theory and this, in tum, will foster the emergence of a more coherent and comprehensive understanding that will hopefully transcend disciplinary and methodological boundaries. We thank the authors for contributing to this goal by guiding our thoughts to aspects of subjective wellbeing that might indeed emerge as universals in the near future.

ROBERT A. CUMMINS, ELEONORA GULLONE AND ANNA

L.D. LAU

A MODEL OF SUBJECTIVE WELL-BEING HOMEOSTASIS: THE ROLE OF PERSONALITY

A bstraci. A considerable body of data is now available to suggest that subjective well-being (SWB) is not free to vary over the theoretical range offered by measurement scales. Rather, most people experience a moderately positive level of well-being, such that the population average IS normally held at about 75 percent of maximum. This has led to the proposal that SWB is under the influence of a homeostatic system designed to hold its value within a narrow, positive, set-point-range for each individual. Our paper offers a model that could account for such maintenance through an interlocking system of psychological devices as follows: (a) Personality provides a steady affective background that determines the set-pointrange for the whole homeostatic system. (b) A set of cognitive buffers involving perceived control, selfesteem and optimism, absorb the impact of different need states and, together with personality, create subjective well-being (c) Met and unmet needs act directly on the cognitive buffers, with the met needs reinforcing the buffering system and the unmet needs providing motivation. Personality is also presented as having a powerful influence on motivational systems that seek satisfaction and, thus, predispose behavior that is likely to maintain normal levels ofSWB. (d) Finally, at the most fundamental level of the homeostatic system, the processes of habituation and adaptation constitute the first line of defence against the threat of changed extrinsic conditions influencing levels of SWB. Data are Cited in support of all levels of the model and the implications of homeostatic control are discussed. In particular, depression is described in terms of homeostatic failure, and the limitation of using SWB as a measure of intervention outcome is emphasized. That is, mterventions can only raise SWB if its initial levels were below the setpoint-range. It is concluded that the model is consistent with a great deal of the literature and that testing its many propositions should prove a fruitful approach to advancing knowledge in this area.

Measuring quality of life through self-evaluation is a recent idea that has captured the attention of researchers and practitioners alike. Research interest stems from the formalization of subjective well-being (SWB) as a definable, measurable construct whose theoretical characteristics are slowly becoming understood. Reflecting the fact that the quality of life field is still in its infancy is the general lack of agreement relating to definition, measurement and conceptual structure. Therefore, this chapter will commence with a general discussion of nomenclature and instrumentation with a view to defining the descriptive terms and measurement procedures that underpin the subsequent discussion. This will be followed by an introduction to the idea that SWB is not free to vary in response to changing external conditions, but is held within an idiosyncratic range by a system of homeostasis. A model will subsequently be proposed in an effort to explain likely structures, constitutional, or otherwise, that underpin individual levels of subjective well-being and their homeostasis. Literature and data relevant to this model will be discussed along the way.

7

E. Gullone and RA. Cummins (eds.), The Universality of Subjective Wellbeing Indicators, 7-46. © 2002 Kluwer Academic Publishers.

8 A MODEL OF SUBJECTIVE WELL-BEING

ISSUES OF NOMENCLATURE AND MEASUREMENT

Among the most inconsistently used terms within the human sciences is "quality of life". Indeed, the words "quality of life" are used with such abandon that readers must delve into the text to ascertain the intended meaning. Other terms such as "happiness" and "we II-being" are likewise afflicted. It is therefore necessary to make explicit the current use of nomenclature which we propose to be based on majority opinion within the literature at this time.

As has been widely acknowledged since Campbell, Converse and Rodgers (1976), perceived weIl-being comprises both affect and cognition and is otherwise referred to as subjective well-being (SWB). Other terms describe a focus onto one process or the other. In relation to affect, the most general term is happiness and this can be measured either by the use of a simple question, "How happy are you with your life as a whole?", or by the use of more complex scales. Such scales have traditionally been based on a conceptualisation that positive and negative affect are separate and mostly independent bipolar dimensions. However, recent literature makes it clear that this simple dichotomy is inadequate. Using the conceptualisation represented by a circumplex model, affect has emerged as a two-dimensional structure. One dimension is emotional valence (positive or negative) while the other is activation (strong or weak) (Larson & Diener, 1992, RusseIl & CarroIl, 1999; Watson & TeIlegen, 1985). This understanding casts fresh doubt on the multitude of studies that have employed single-dimensional scales, such as the Positive and Negative Affect Schedule (Watson, Clark and TeIlegen, 1988) and the Affect Balance Scale (Bradburn, 1969). As yet, however, no commonly used bidimensional scale has emerged, even though Huelsman et al. (1998) have generated a list of terms that could be used for this purpose.

In terms of the cognitive component, it is generally recognized that this part of SWB involves some form of internal comparison process. The precise nature of such comparisons is not entirely certain but the most complete description of possible contenders has been provided by Multiple Discrepancies Theory (Michalos, 1985). This theory proposes comparisons with the self in the past, other people, etc., and has received considerable support (e.g. MeIlor, Cummins & Loquet, 1999). As one consequence, it is generally accepted that this cognitive component of well-being can be measured through questions of "satisfaction".

At the simplest level, this cognitive component can also be measured by a single question. "How satisfied are you with your life as a whole?" This yields a measure of life satisfaction. In addition, it is now widely recognized that life satisfaction can be divided into a number of "domains", representing the component areas of life experience, and that domain satisfaction, in aggregate, reflects overall life satisfaction (Campbell et aI., 1976; Diener, 1984). How, precisely, to characterize such domains has yielded a wide variety of opinion. However, this situation has become less contentious in recent years, with many authors agreeing on the character of some central domains, such as those involving health, wealth and relationships (see e.g. Felce & Perry, 1995, Flanagan, 1978; Headey & Wearing, 1992). Two recent documents, have consolidated such views. The International

R.A. CUMMINS, E. GULLONE AND A.L.D. LAU 9

Society for Quality of Life Studies (Hagerty, Cummins, Ferriss, Land, Michalos, Peterson, Sharpe, Sirgy & Vogel, 2001) and the International Association for the Scientific Study of Intellectual Disability (Schalock, Bonham, & Marcharnd, 2000) have both reviewed quality of life measurement. They have separately agreed that domains should exhibit a number of defining characteristics. These include being both objectively and subjectively described, being parsimoneous, and being descriptive of generic life areas.

One instrument that is consistent with these views is the Comprehensive Quality of Life Scale (ComQol) (Cummins, 1997a,b) and more recently the Personal Wellbeing Index (Cummins et aI., 2002) which employs seven domains as "Material well-being, health, productivity, relationships, safety, community and emotional well-being (For a more detailed argument justifying these domains see Cummins, 1996, I 997c). The aggregate of satisfaction across life domains yields Subjective Quality of Life (SQOL). For a listing and brief description of available instruments to measure SWB for people in general and other popUlation groupings consult the "Directory of Instruments to Measure Quality of Life (Cummins, 2001), published on the web-site of the Australian Centre on Quality of Life (http://acqol.deakin.edu.au).

GENERIC VERSUS SPECIFIC INSTRUMENTATION

A glance through the above-named directory will reveal over 600 instruments, many of which have been designed to measure the SWB of particular population subgroups. Notably, however, there is an absence of measures designed for sub-groups who are relatively advantaged, such as the economically wealthy. Instead, specific instruments are inevitably designed to measure the SWB of people who are disadvantaged, such as due to some medical condition, low income, or congenital condition such as an intellectual disability. This has a number of unfortunate consequences. One is that such scales have a deficit orientation, such that a high score indicates a relative lack of disability, rather than a high quality of life. For example, the medical literature in this area is dominated by a poorly defined construct called "Health-related quality of life". Depending on the scale that is employed, the measures involve disease symptoms, functional status, standards of care, patient perceptions of their health, etc., often combined into a single scale. But such scales have little in common with SWB, even when they involve patient perceptions. A "Cancer" quality of life scale, for example, may enquire whether the respondent experiences nausea (see, e.g., the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire; Aaronson et aI., 1993). So, the absence of nausea contributes to a high quality of life as recorded by the scale. This brings into sharp focus the difference between such instruments and SWB scales that provide an assessment of quality of life that is relevant and applicable to the general population.


NORMATIVE SQOL VALUES

One of the surprising features of SQOL is that it can be described in terms of an empirical normative standard. This understanding has come about through a series of studies demonstrating that Western population mean scores for SQOL predictably lie within the range 70 to 80 %SM (Cummins, 1995, 1996, 1998,2002). The statistic %SM describes the conversion of Likert scale data to a range from 0 to 100. This conversion is simply made by a two step process which allows the "percentage" calculation to be based on the principle of a ratio scale commencing with zero. The steps are: (a) Re-code the Likert scale scoring to commence with zero. Thus, a scale scored 1-5 is recoded 0-4; (b) a percentage is then calculated against the maximum scale score. For example, a score of3 is calculated as 3/4 x 100 = 75%SM.

Aggregates of mean values from general population surveys when recoded in this manner have been demonstrated to consistently yield a mean of 75 and a standard deviation of 2.5. Hence, the range 70 - 80% SM describes two standard deviations around the mean, and thus approximates the normative range for general population sample mean scores (see Figure I). This distribution has been demonstrated for life satisfaction and SQOL measured by the Comprehensive Quality of Life Scale (Cummins, I 997a,b). While all other SWB scales seem to yield general population mean values that lie between 60-90% SM (personal observation, RAC), the distribution of only one other scale has been systematically studied. The Satisfaction With Life Scale (Diener, Emmons, Larsen, & Griffm, 1985) yields values 5-10%SM below the 70-80%SM range and the reasons for this probably lie within the items comprising the scale (see Pallant & Cummins, 2001).

F r e q u e n c y

Lower threshold for individuals :

I I

~! I I I I I I I I

I

o--tft-'M-'ffi-"-zttr-.;rr-'60 7,

I I I I I I I I I

80 90 100

Lower and upper thresholds

Population Mean

Percentage of Scale Maximum (%SM)

Figure I. The normal distribution a/subjective quality a/life

R.A. CUMMINS, E. GULLONE AND A.L.D. LAU II

The hypothesized range of 70-80%SM for SQOL population mean scores has been confirmed by a variety of data. At the top end of the range, no samples yet discovered have a mean value that lies significantly above 80%SM. For general population samples, the life satisfaction values for the Nordic countries, which are higher than all other countries, do not exceed this value (Cummins, 1995). In terms of population sub-groups, one of the highest recorded levels of life satisfaction has been derived from people who are very wealthy, and their values also average to around 80%SM (Cummins, 2000a). The other sub-group with very high levels of life satisfaction, in an entirely different sphere of comparison, is the "Back-to-theLanders" in the U.S.A. (Jacob & Brinkenhoff, 1999). For this group, the mean value also approximates 80%SM. The implication of such data is that group mean values for SQOL cannot be reliably held above 80%SM. This has received quite explicit confirmation from Groot and VandenBrink (2000) who found that additional income made no difference to life satisfaction for people with minimum value of80%SM.

The lower threshold of 70%SM has been verified through a detailed analysis of sample variance (Cummins, 2002). From Figure 1 it can be predicted that as the mean value of samples descends below the threshold of 70%SM, the variance suddenly becomes greater. This pattern of change in sample variance around 70%SM has been confirmed (Cummins, 2002).

The values so far discussed represent calculations based on whole sample mean scores. The values for individuals within samples show a broader distribution that appears sensitive to the measurement instrument. Life satisfaction as measured by a single question, shows a normative, within sample distribution, of 75± 18%SM or a range of about 40-100%SM (Cummins 2002). SQOL, as measured through the Comprehensive Quality of Life Scale shows a normative distribution of75±12%SM, or a range of about 50-1 OO%SM (Cummins, 1999). This has been depicted in Figure I by the vertical line at approximately 50%SM demarcating two standard deviations (±12%SM) from the mean (75%SM). Thus in approximate terms, individual people generally maintain their SQOL within the positive sector of the %SM range.

From the data that have been presented it can be seen that levels of SWB are predictable within the normative ranges described. It is therefore not surprising that authors have reported a high level of stability in SWB over time. For example Bowling (1996) reported correlations of .46 to .65 in the life satisfaction of elderly people over a 3 year period; Suh and Diener (1996) reported correlations of .56 and .61 respectively, for positive and negative affect in college students over a 2 year period; Headey and Wearing (1989) reported coefficients of .64, .51 and .52 when using their Life Satisfaction Index on a general population sample at 2, 4 and 6 years, respectively; while Costa and McCrae (1989) reported correlations of .47 to .63 using a battery of SWB instruments over a 2 year period. These data, together with the predictable ranges of SQOL values, constitute converging evidence that SWB is not simply free to vary at the whim of personal circumstances, but is managed. This idea, that SWB is under active internal control, we term SWB homeostasis.


HOMEOSTASIS

The general idea of the proposed homeostatic system is that SWB is managed, for each individual, within a "set-point-range" (Cummins, 2000b). That is, each person has an in-built "set-point" for their normal level of SWB, as proposed by Headey and Wearing (1992), and their perceived SWB is normally held within a narrow range around this setting. This idea also involves the concept of threshold, as previously discussed, which we propose exists at the margins of the set-point-range. That is, as SWB approaches these margins the homeostatic system resists further change and, if the threshold is exceeded, the homeostatic system then works to bring SWB levels back to lie within the normal range for the individual person. Empirical evidence for this proposition is available at the level of sample means. As has been described, Cummins (200 I b) found a systematic increase in variance as sample means fall below 70%SM, the hypothesized lower threshold. In addition, 80%SM appears to be the highest value that can be sustained by a representative sample, as has also been described.

At the level of individuals, homeostasis predicts that people who suffer some event that depresses their SWB below threshold should improve their levels of SWB over time. This has been widely reported, for example, by people who have received a diagnosis of cancer (Bloom, Fobair, Spiegel, Cox, Varghese & Hoppe 1991), who have received bums (Andreasen & Norris, 1972), and who have become paraplegic/tetraplegic (Bach & Tilton, 1994). However, all homeostatic systems have their limitations and so it would also be expected that SWB recovery would be contingent on the residual discomfort or lost functional status not being overly severe. This limitation is exemplified by the variable degree of recovery shown by people with paraplegia/tetraplegia reported above. In fact, such recovery was restricted to people left with autonomous breathing. If they remained ventilatorassisted no significant SWB recovery took place (Bach & Tilton, 1994).

This understanding, that the temporal stability of SWB depends on the severity of the challenging agent, allows a further prediction. This is that the stability of SWB, mentioned earlier, should be restricted to two broad groups. The first group comprises people who have levels of SWB within their homeostatic range, and who experience no major event sufficient to disrupt homeostasis. The second comprises people who experience homeostatic defeat due to some chronic condition, like extreme poverty, to which they cannot adapt. Conversely, the least stability should be evidenced by people who, at the time of initial measurement, were experiencing homeostatic defeat due to either a transitory event or to circumstances that could be accommodated over time by the processes of homeostasis.

Data consistent with these predictions have been reported by Landua (1992) in a large, longitudinal population study. He measured life satisfaction on a 0-10 scale, but then created response categories as 0-4, 5-6, 7-8, and 9-10. Measures were made at baseline and then four years later, when it was found that the following percentages of people had remained in their initial category: II % (0-4), 50% (5-6), 63% (7-8), and 61% (9-10). Thus, as predicted, greatest stability was recorded by people with an initial SWB of at least 70% SM and the lowest stability by people


who initially scored less than 40%SM. These data are consistent with the idea that this low scoring group initially comprised a high proportion of people who were suffering homeostatic defeat due to either transitory events or circumstances that were amenable to adaptation. Thus, over the four year interval, many of these people were able to re-establish homeostatic control, with the result that their life satisfaction moved above 40%SM, and so out of the lowest response category.

A further set of predictions arises from a consideration of the thresholds depicted in Figure 1. These allow the prediction that the correlation between SWB and extrinsic indicators will increase as SWB moves outside its normative set-pointrange. Extrinsic in this context refers to perceptions that arise outside the homeostatic system. These may have their origins either external to the person, in terms of objective indices or life events, or within the person, such as in perception of pain.

Provided that the homeostatic system is not overly challenged by such extrinsic influences, and the homeostatic system can adapt to their presence, they will have little discernable influence on SWB. However, as the strength of an extrinsic influence increases, at some value it will exceed the adaptive capacity of the homeostatic system, and the control of SWB will fall under the influence of the extrinsic agency. In other words, the plot of the relationship between the strength of the extrinsic agent and the value of SWB is curvilinear around each threshold, much as depicted in Figure 2.

High

SQOL

Low

Dominant Source of SQOL Control

Extrinsic conditions

Strong negative

i

Homeostasis conditions

Lower

Threshold

Neutral

Extrinsic conditions

1 Unner

Threshold

Strong Positive

Strength of The Extrinsic Conditions

Figure 2. The relationship between subjective and objective quality a/life


Figure 2 describes the changing relationship between the strength of the external agencies and SQOL. It indicates that provided the strength of extrinsic agents remains sub-threshold (Le. above the lower threshold and below the upper threshold), their variation will exert little systematic influence on SWB, which is held within its set-point range. This will change, however, if the strength of the agents exceeds the homeostatic threshold. Once this happens, the extrinsic agents begin to wrest control of SWB away from the homeostatic system, causing SWB to rise or fall. As this occurs, the agencies and SQOL start to co-vary. This theoretical understanding allows the following specific predictions: I. Under maintenance conditions, where no threat to homeostasis can be

recognized, there should be no systematic relationship between the objective circumstances of people's lives and their SWB. This is because homeostasis, not the extrinsic conditions, are controlling levels ofSWB. This lack of relationship has been confirmed in an empirical review (Cummins, 2000c).

2. Under non-maintenance conditions, where the homeostatic system is facing defeat, the relative strength of the relationship between extrinsic conditions and SWB changes. Here, the extrinsic conditions are the dominating force, defeating homeostasis and, thereby, wresting control ofSWB by causing it to rise or fall. Under these conditions, the correlation between SWB and the extrinsic condition is much enhanced. Some evidence for this has also been presented in the review by Cummins (2000c) which demonstrated a generally higher correlation between SWB and objective variables under conditions of extrinsic threat. Further evidence can be deduced from studying the relationship between perceived health and physical health. Perceived health, as a component of SWB, is generally unrelated to physical health in general population samples, presumably due to the influence of homeostasis. Duckitt (1983), on the other hand, reported a high correlation between objective health and perceived health among elderly women. Such data are consistent with these women being under homeostatic threat from the likely compromised state of their physical health associated with aging.

3. There will be a law of diminishing returns in the ability of improved objective conditions to cause an increase in SWB. That is, in conditions of marked deficit, many of the objective indicators will have the power to control SWB. For example, chronic poverty, friendlessness, lack of safety, etc. However, if such circumstances are improved to the point that they are no longer instrumental in causing homeostatic defeat, further improvements are predicted to have little further effect on SWB for two reasons. First, control has been returned to the homeostatic system, and so further improvements will be absorbed by the system, effectively holding the SWB output constant. Second, if a sudden, marked improvement occurs that exceeds the upper threshold, the processes of adaptation will quite rapidly diminish the impact of this new experience and, once again, return control to homeostasis. An example of this latter phenomenon has been provided by (Groot et aI., 2000). They divided a large population sample into deciles on the basis of life satisfaction. True to prediction, they found that in the two deciles above 80%SM, income had lost its ability to further increase life satisfaction.


In summary, a considerable body of data is consistent with the idea of a homeostatic ally controlled level of SWB. Not only does SWB appear to be held within a range characterized by upper and lower thresholds, but also deviations from this narrow range are characterized by instability and a heightened correlation between SWB and the responsible extrinsic agent. What is now required, in order to understand these ideas further, is an indication of the psychological processes that might comprise such a homeostatic system. Our model will now be described.

A MODEL FOR HOMEOSTASIS

The idea that SWB is maintained by the brain in some form of dynamic equilibrium has been proposed by several other authors (Headey & Wearing, 1989; Nieboer, 1997; Ormel, 1983; Ormel & Schanfeli, 1991). However, apart from a shared view that personality must be somewhere involved in such maintenance, these theorists have not attempted to explain the mechanisms that are responsible for such an equilibrium state. We will now attempt to fill this gap by outlining a model for the homeostatic control of SWB. The proposed model is shown in Figure 3 below.

Homeostatic system NOT under challenge Homeostatic system under challenge

Personality & Affectivity

Cognitive Buffer

A versive Extrinsic Conditions

Figure 3. A Homeostatic ModelJor Subjective Well-Being

This model proposes two potential pathways that describe the relationship between objective circumstances and the subjective perception of one's well-being. For each of these pathways, there are three levels of processing between some perceptual input that is extrinsic to the system and SWB, which is depicted as output. The first level of processing constitutes the unconscious processes of adaptation and habituation. The second level involves the conscious awareness of met and unmet needs. The third involves a system of cognitive "buffers" which act to absorb the impact of changing need states in order to maintain a steady-state output, which is SWB. As shown in Figure 3, the model also predicts that the second and third levels


of processing are strongly influenced by personality. [n other words, personality (including stable cognitions and affect) moderates the relationship between external experiences and subjective welJ-being. This is true in both depicted pathways.

The central distinguishing characteristic between the two pathways is whether or not the environmental circumstances challenge the homeostatic threshold. If environmental experiences do not challenge the system (i.e. habituation or adaptation occurs), the needs will be perceived as having been met. Personality is proposed to play an important part in determining these perceptions which, consequently, determine SWB level. In such circumstances, cognitive buffers are not strongly implicated in SWB management as they are largely unnecessary. Alternatively, if environmental experiences do challenge the system and the individual experiences difficulty in adapting to the "stressor", related needs will be perceived as not having been met. Such a state will be strongly influenced by, and will also provide chalJenge to, the individual's personality (e.g. high neuroticism, high negative affect) and resources (i.e. the cognitive buffers). In other words, the system's component parts will experience change (some parts are more resistant to change, particularly personality) in an attempt to achieve equilibrium, and consequently to maintain SWB at homeostatic levels.

Of course, if the threshold has been exceeded, a third pathway (not depicted in Figure 3) is implicated in available data (described above). This will involve a direct relationship between objective circumstances and subjective well-being. We wilJ now proceed to describe each of the model constructs and paths in more detail.

PERSONALITY VARIABLES

NEUROTICISM AND EXTRA VERSION, POSITIVE AND NEGATIVE

AFFECTIVITY

There is within the literature strong, convergent evidence that personality is a major determinant of how happy or sad people feel with their lives. As stated by esteemed ethnographer Robert Edgerton (1990):

"The pattern that emerges again and agam is that people who were happy and hopeful 10, 20, or even 30 years ago remain so no matter what ill-fortune they suffer; and those who were sad or negative about life do not change even though their environment improves significantly. The data clearly indicate that major life stressors or major gratifications can bring about changes in affect and expressed life satisfactIOn, but these changes are short lived.

Counter-mtuitive as this finding may seem to those like myself who believe in the causal power of environmental factors. these data support that internal dispositions - call them temperament for want of a better term - are better predictors of people's satisfaction with the quality oftheir lives than are objective environmental variables" (p. 156-7).

Quantitative data have overwhelmingly confirmed this view, most particularly with respect to the two personality dimensions of extraversion and neuroticism. PracticalJy all of the many studies that have looked at the relationship between these dimensions and SWB have found a robust correlation. This has led to the "topdown" hypothesis, such that SWB is substantially influenced by personality (see,


e.g. Mallard, Lance & Michalos 1997). Despite this strong relationship, other factors (as noted above and to be discussed in detail below) are also implicated since differences in personality do not completely predict differences in SWB except under extreme circumstances. In other words, personality and SWB are different constructs.

Two dimensions of personality in particular are intimately linked with affect. These are neuroticism which has consistently been shown to be related with negative affectivity, and extraversion which has consistently been shown to be related with positive affectivity (Fogarty, Machin, Albion, Sutherland, Lalor & Revitt, 1999; Watson & Clark, 1992; Wilson & Gullone, 1999). A good description of negative affectivity has been provided by Brief, Butcher, George and Link (1993) as "a mood-dispositional dimension that reflects pervasive individual differences in negative emotionality and seif-concept"(p. 647). In particular, people high on negative affectivity are nervous, apprehensive, irritable, overly sensitive and emotionally labile (Watson & Pennebaker, 1989), as well as having a more negative worldview. They rate peers less favourably, and experience a wide variety of negative emotions even in the absence of known stressors (Elliott, Marmarosh and Pickelman, 1994). They also have a tendency to experience anxiety, dysthymia and depression.

In confirmation of the uni-dimensional nature of these negative affective states, they all tend to co-vary (see, e.g., Abbey & Andrews, 1985; Depue & Montoe, 1986; Hunt, Singer & Cobb, 1967; Watson, Clark, Weber, Assenheimer, Strauss & McCormick, 1995; Watson & Pennebaker, 1989). Thus, for the purpose of this discussion, neuroticism will be considered as the source of constitutional negative affectivity. Extraversion, as the source of constitutional positive affectivity, appears as the natural opposing force to neuroticism and it seems reasonable to suggest that it is the balance between these two personality dimensions that provides the setpoint-range for SWB.

Further supporting the associations between extraversion and positive affectivity, and neuroticism and negative affectivity, are their relationships with happiness. There is general agreement in the literature that extraversion is positively correlated with happiness (e.g. Argyle & Lu, 1990; Diener, Sandvik, Pavot & Fujita, 1992; Francis, 1999) and life satisfaction (e.g. Doyle & Youn, 2000). There is also agreement that neuroticism is negatively correlated with happiness (e.g. Francis, 1999; Lu & Shih, 1997) and life satisfaction (e.g. Brief et aI., 1993).

However, for the pathway in Figure 3, determined by extrinsic circumstances that challenge threshold levels, an indirect link between personality and SWB is also depicted. This is mediated by the buffering systems (e.g. perceived control and optimism). Support for this indirect link can be found in research demonstrating that extraversion acts on the buffers and the perception of met needs, thereby constituting what Flynn and Cappeliez (1993) term "protective factors" for the person's well-being. These are high self-perceived social competence, high levels of learned resourcefulness, a perceived high frequency of pleasant events, and the perceived availability of a close and immediate confidant. Thus, it might be expected that such protective factors would reinforce one another, and so co-vary, in much the same manner as described for the components of negative affectivity. This


is, indeed, the case. For example, internal control and social performance correlate positively (Abbey & Andrews, 1985) while in a meta-analysis, DeNeve and Cooper (J 998) report an average correlation >.3 between the following traits: Trust, emotional stability, desire for control, hardiness (the tendency to cope positively with life events), and positive affectivity. In summary, and as also argued by Fyrand, Wichstom, Mourn, Glennas and Kvien (1997) on the basis of structural modelling, the personality traits of neuroticism and extraversion seem likely to be causally related to levels of SWB through both direct and indirect links involving the cognitive buffers and met needs.

THE OTHER DIMENSIONS OF PERSONALITY

How, precisely, to define personality in terms of operationalized variables, remains a matter of debate. Literally hundreds of variables have been created which purport to represent such measurement. However, over the past decade a great simplification has taken place with the advent of the "Big Five" personality factors as measured by the NEO-Revised (Costa & McCrae, 1988). While these factors are obviously not a perfect representation of personality, they are clearly robust and widely accepted as the best collective approximation to such measurement so far available. For this reason our attention will be restricted to these five factors which, in addition to extraversion and neuroticism, include openness, agreeableness, and conscientiousness.

Table I comprises a synthesis of findings from the eight studies known to us that have provided bi-variate correlations between all five personality factors and variables relevant to the homeostatic model. The footnote to this Table indicates the wide variety of scales that have been employed. Clearly, any coherent patterns in these results are going to be very robust.

This Table presents, for each homeostatic variable, the number of values (number of sample means available for each calculation), the number of studies that contributed data, and the number of independent samples that contributed data. In order to combine the bi-variate correlations for each homeostatic variable, each coefficient has been converted to Fisher's z, prior to the calculation of each mean and standard deviation. The resultant values have then been converted back to correlation coefficients for reporting in the Table.

R.A. CUMMINS, E. GULLONE ANO A.L.D. LAU 19

Table 1. Personality related to other variables

., ., '" ... ... ...

OIl = :;; "E. = '; .:: 'is. e =

, ., .. <= <= 0 '" ... :z z <= N E 0 A C Av () :z

12 mean - 3424 2527 .1403 .1529 2173 .22 TOTAL 62 8 SO .1958 .1340 .0962 .0954 .1303 SWB 18 6 9 mean -.3450 .3194 .1233 .1667 .2389 .24

SO .1312 .1490 .1001 .1084 .1129 SQOL 8 5 8 mean - 4450 3588 1325 .1800 2413 .27

SO 0883 1057 .1005 .0795 0685 SELF 9 4 7 mean -.5333 .3711 1300 1689 .3411 31 ESTEEM SO .1592 .0790 .1057 .0925 0761 CONTROL 8 1 2 mean -.2963 .2114 .1475 .1288 .2588 .21

SO .1256 .0648 .0734 .0869 .1641 JWB II 1 2 mean - 1982 .1245 1600 .1218 0745 .14

SO .0950 .0755 .0578 0634 .0472 MOTlV 4 1 2 mean -.0875 1800 1075 1325 .1575 .\3

SO .0741 0616 .0685 .0403 .0793 ILL BEING 6 3 4 mean .6567 .2383 .1500 .\300 1667 .27

SO .1093 .1196 .1343 .0825 .1056

Notes

I N = Neuroticism, E = Extraversion, 0 = Openness, A = Agreeableness, C = ConscientIOusness, Av = Average correlation coefficient, SWB = Subjective Well-being, SQOL = Subjective Quality of LIfe, JWB = Just World Beliefs, Motiv = Motivation.

2. The studies and variables comprising Table I are as follows:

Studies: Robins et al. (200 I); Campbell et aI., (\996); Lipkus et al. (1996); Tokar & Subich (1997); Courneya et al. (2000): Kwan et al. (1997); Pychyl & Little (\998); Staudinger et al. (1999).

Personality Scales: NEO (Costa & McCrae, \992); (Goldberg, \992); eO/60

representative items from the short version of NEO-FFI, Costa & McCrae, \989); Own scale (Staudinger et aI., 1999).

Subjective Well-Being (SWB) Scales: Job Satisfaction Blank (Hoppock, 1935); Positive and Negative Affect Scale (Diener & Emmons, 1985); Affect Balance Scale (Bradburn, \969); Positive relationships, from the short form of the Ryff Inventory (Ryff & Keyes, \995); Four items measuring perceived health (Staudinger et aI., \999); Relationship harmony, own scale (K wan et aI., 1997); Satisfaction with academic life, own scale (Pychyl & Little, 1998).

Subjective Quality of Life (SQOL) Scales: Satisfaction with Life Scale (SWLS; Diener et ai, \985); Combined index of SWLS plus satisfaction with life as a whole; Three items (Staudinger et a\., 1999).


Self-Esteem Scales: Rosenberg Self-Esteem Scale (Rosenberg, 1965); SelfAssessed uniqueness (Dwan et aI., 1997); Self-concept clarity (Campbell et aI., 1996); self-acceptance (short form, Ryff Inventory (Ryff & Keyes, 1995).

Control Scales: Autonomy, Environmental Mastery, Control over own health, Control over work. (Short form, Ryff Inventory (Ryff & Keyes, 1995).

Just World Beliefs Scales: Beliefs in a just world scales/items (World Assumptions Scale, Janoff-Bulman, 1989; Lipkus, 1991; Rubin & Peplau, 1975; Lipkus et aI., 1996).

Motivation Scales: Personal life investment: Thought and effort put into health and work (Staudinger et aI., 1999).

Ill-Being Scales: Center for Epidemiology Studies Depression Scale (Radloff, 1977); Perceived Stress Scale (Cohen, et aI., 1983); Negative affect: Positive and Negative Affect Scale (Emmons & Diener, 1985); Negative affect: Bradburn Affect Balance Scale (Bradburn, 1969).

The following points can be noted in respect of the relative strength of relationship between the personality and homeostatic variables displayed in Table 1. 1. The whole Table is dominated by Neuroticism. In all but one category

(Motivation) it yields the highest correlation. 2. The highest correlations with neuroticism are with III-being (.66), followed by

Self-esteem (-.53) and SQOL (-.45). 3. The second most dominant factor is Extroversion. This appears highest in the

case of Motivation and second highest in 517 of the other categories. 4. The third most dominant factor is Conscientiousness. It appeared second highest

in Control and third highest in 617 ofthe remaining categories. 5. The two factors of Openness and Agreeableness appear to have very weak

connections with the homeostatic categories. The highest single value was .17 for Self-esteem.

When the correlations are averaged across the five factors, the highest loading category is Self-esteem (.31) followed by SQOL and III-being, both on .27. The lowest are Just World Beliefs (.14) and Motivation (.13).

It can be concluded that the factors of neuroticism and extraversion, and to a lesser extent conscientiousness, have a robust relationship with subjective wellbeing conceptualised either as SWB or SQOL. In addition, these three factors also have a reliable influence on the two putative "buffer" variables of self-esteem and control (see later). Indeed, the -.53 correlation between neuroticism and self-esteem is the highest value in the Table.

The magnitude of this correlation draws attention to the nature ofthe relationship between neuroticism and self-esteem. In particular it questions the extent to which these constructs are separable. Certainly there is abundant evidence that neuroticism accounts for most of the variance in self-esteem as measured by the Rosenberg scale. For example, Roberts and Kendler (l999) found that the power of self-esteem to predict depression was eliminated when neuroticism was used as a co-variate. There are, however, many different ways to measure self-esteem, and these different instruments may yield very different results.


There is general agreement that Rosenberg's scale measures a general, global conception of self-esteem. Other scales, however, often target more specific aspects of the construct. It is interesting in this framework to partition the three values from the Rosenberg scale (-.673 ± .038) from the six values derived from other scales (.463 ± .149). This difference is significant [t(7) = 2.211, p<.05] and gives rise to two observations. First, as noted by many other authors (see later), self-esteem scales are not equivalent. Second, and most important for the homeostatic model, the Rosenberg scale is sharing some 50 percent of its variance with neuroticism, which raises the question of whether global self-esteem should be considered to be personality.

This presents something of a difficulty for the depiction of self-esteem within the homeostatic model. Another difficulty is how to represent personality in the model since the analyses in Table I indicate no clear separation between the five factors in their degree of influence on other variables. So, for the sake of simplicity, the following scheme will be followed. The influence of personality in the model will be restricted to that exerted by neuroticism and extraversion. Moreover, self-esteem will be considered as one of the three putative buffers, even though its close link with neuroticism is acknowledged.

STRENGTH OF RELATIONSHIP BETWEEN THE PERSONALITY VARIABLES

ANDSWB

Estimates of the relationships between personality traits and SWB range from very low and ambiguous degrees of relationship (Diener & Larsen, 1984) to strong correlations. For example, Sandvik, Erikssen, Thaulow, Erikssen, Mundal and Rodahl (1993) concluded that, among their sample of college students, about half of the variance in SWB could be attributed to personality and stable environmental conditions. The reason for such wide differences of opinion is that such an estimation is likely dependent on many other considerations as follows: 1. The similarity of the SWB measure to trait affect (see, for example, the review by

De Neve, 1999). Given the previously described relationships between affect and personality, it is not surprising that, in studies where extraversion and neuroticism have been used as covariates, they appear to negate the relationship between positive and negative affect with some other measure of well-being.

2. The degree to which the personality variables reflect extraversion or neuroticism as opposed to the other factors of personality. For example, Cooper (J 998) reported that within a meta-analysis involving a very heterogeneous collection of 137 "personality" traits, they found an average correlation with SWB of .19. Clearly this is less than would be expected using either measures of extraversion or neuroticism as the correlates. In a similar vein, Fogarty et al. (1999) found strong correlations between job satisfaction and both extraversion and neuroticism, but weak correlations with the personality factors of openness, agreeableness and conscientiousness (see also, however, the discussion in conjunction with Table I).


3. Homeostasis theory predicts that the correlations should be maximal in samples with normal levels of SWB where no obvious source of threat to well-being is in operation. This is because, under such conditions, the level of SWB should be minimally affected by extrinsic influences, and so its value should most closely reflect the set-point-range managed by personality. Under extrinsic threat conditions, on the other hand, the control ofSWB is predicted to shift, first, to the homeostatic system of buffers, and then when homeostasis is defeated, to the threatening agent itself, as it causes SWB to fall below its normal range (see Figure 2).

Evidence for such a shift in the control of SWB can be deduced from the study by Duckitt (1983) on elderly women. Consistent with the expectation that this sample would be under homeostatic threat due to poor medical health, a high correlation was found between objective and perceived health. Moreover, personality did not correlate with perceived health after objective health had been partialled out, thus further emphasizing the dominance of objective health over SWB. It was also found that life satisfaction and life events remained correlated with perceived health even after objective health had been partialled out. This indicates the continued association between measures of SWB (life satisfaction and perceived health) as expected from a "bottom-up" model. Moreover, the continual influence of life events on perceived health after partialling out objective health, can also be explained in terms of the model as follows: Because the homeostatic buffers had been defeated by poor objective health, the system had little capacity to absorb the impact of other negative events so that they, also, exerted a direct effect to reduce life satisfaction.

Thus, given the above review, it seems that the strength of the relationship between personality and SWB will be situationally variable and, apart from being dependent on the nature of the measures employed, it will be determined by the degree of challenge that objective circumstances bring to the homeostatic system. In corroboration of this view, Diener, Suh, Lucas and Smith, (1999) conclude from their review of this area, that while it is clear personality and SWB are linked, no simple general estimate of the strength of this linkage can be made. However, it is possible that if samples were separated on the basis of their relative degree of homeostatic threat, more consistent estimates would emerge.

PERSONALITY, NEEDS AND MOTIVATION

"Personality and motivational self-system processes are the fulcrum around whIch all other psychological, educatIonal, and self-regulatory processes rotate to energize behavior and performance" (Switzky, 1999, p 70).

This quotation, which leads Switzky into a major review of the topic, clearly places personality at the center of operations. This view is consistent with the role of personality depicted in Figure 3 and, most importantly from a motivational viewpoint, it is consistent with the proposed influence of personality on the strength of met and unmet needs. The evidence for such a genetic link appears to be strongest


in relation to perceived social support (see e.g. Bergeman, Plomin, Pedersen & McClearn, 1991). This link has special significance in terms of the model since it may represent the simplest connection between personality and motivation, as can be seen from the powerful study by Lucas et al. (2000).

These authors studied extraversion with particular attention to the facet of "sociability", as the enjoyment of social activities, preference for being with others, and "reward sensitivity". This latter term depicts an underlying motivational system characterized by a strong Behavioural Activating System (Gray, 1970) which regulates reactions to positive stimuli. According to Dupue and Collins (1999) the operation of this system can be described in the following terms:

"Exposure to ...... mcentlve stimuli (or activation of theIr central representation) elicits an incentive emotional state that facilttates and guides approach behaviour to a goal. In humans, incentive motivational states are associated with strong positIve affect characterized by feelings of desire, wanting, excitement, energy, potency, and self-efficacy" (p. 495).

Thus it can be seen that the role of the Behavioural Activating System is to motivate and guide goal-directed behaviour. This is consistent with the general view, offered by many authors (e.g. A via, 1997; Izard, 1993) that one of the main functions of the emotional system is to motivate specific response patterns which are related to feelings of both positive affect and negative affect (e.g. anger: fighting, fear: fleeing).

Lucas et al. (2000) provide compelling evidence that the sociability facets of extraversion form a higher-order factor of reward sensitivity. The facets they identify are: Affiliation (enjoying and valuing close interpersonal bonds, being warm and affectionate); Ascendance (feeling dominant or being an exhibitionist); Venturesome (feelings of excitement seeking and desire for change); and Social Interaction (preference for social interaction). It is thus proposed that these facets of extraversion form an aspect of motivation (reward sensitivity) which reflects the degree to which people are motivated by the prospect of a reward.

They then ask why reward sensitivity is linked to the enhanced sociability of extraverts, and provide an interesting answer. This is that social situations involving warmth, affection and close emotional bonds, are especially rewarding. Thus, the increased sociability of extraverts (see e.g. Argyle & Lu., 1990) is simply a byproduct of greater sensitivity to rewards. This is not to say, however, that such increased sensitivity is confined to social situations. As these authors also note, extraverts tend to feel more positive affect even when they are alone (see e.g. Diener et aI., 1992), but there does appear to be some special motivational power attributable to social rewards.

They also offer an interesting piece of confirmatory evidence which draws on cross-cultural data. As Lucas et al. (2000) note, social contact may serve different purposes in different cultures. For example, while within individualistic cultures people are more likely to seek social interaction because it is rewarding, in collectivistic cultures a more central goal may be group harmony. In confirmation of social reward sensitivity, while extraversion and positive affect correlated in samples from both cultures, the relationship was less in collectivist samples. Thus they conclude "sensitivity to rewards, rather than sociability, forms the core of extraversion" (p. 466).


It seems possible that this differential reward sensitivity may also explain why extraverts appear to experience more positive events (Headey, Holmstrom & Wearing, 1985) and more adverse events (Headey & Wearing, 1989). That is, people who are sensitive to rewards seek more interaction with their environment and, subsequently, experience more "events". The fact that these authors found favourable and adverse events kept happening to the same people over time is further evidence that reward sensitivity represents a stable personality characteristic, linked to met and unmet needs. The influence of reward sensitivity can also be linked, via personality, to SWB. Assuming that such sensitivity facilitates a search for affiliation, then such activity has been reported to produce a positive feedback cycle in which the searching behaviour and SWB are mutually reinforcing to one another (Filipp & Klauer, 1991). Thus, as also concluded by De Neve (1999), relationship-enhancing traits are important for SWB maintenance.

The linkage between neuroticism and needs is less established. However, such a connection has been supported by Fyrand et at. (1997), who found a strong correlation between depression and the lack of social support (an unmet need) among people with arthritis, However, the relationship virtually disappeared after the effects of neuroticism were removed. The authors conclude that neuroticism may, therefore, have caused the perceived lack of social support.

In summary, a case can certainly be made that the two personality dimensions of extraversion and neuroticism are intimately and centrally related to levels of SWB. The nature of the relationship has been argued to be both direct and indirect. The direct link involves a constant background level of affect. The indirect links with SWB are made through the cognitive buffers (discussed in the next section) and metlunmet needs. These links, together with the direct supply of affect, are argued to create the stable set-point-range for individual levels of well-being. The case has also been argued that extraversion is strongly implicated in the creation of motivation to increase positive affect, most particularly by way of generating a sensitivity to social rewards. Thus, personality has a powerful role in the maintenance of SWB through supplying of a steady level of affect and the generation of reward motivation.

COGNITIVE BUFFERS

In a recent review, De Neve (1999) found overwhelming evidence that the way people think about and explain what happens in their lives is intimately tied to their level of SWB. Among the most researched cognitive protective factors are selfesteem, perceived control and optimism (Cummins & Nistico, 2002). Each of these components will now be considered.

Self-Esteem

The construct of self-esteem must be one of the most researched areas of psychology. Even by 1970 over 2,000 articles had been published on the topic (Rosenberg, 1979) and its popularity among researchers remains high to this day. Because of this extraordinary degree of attention, the construct has been dissected


and examined in minute detail, with substantial argument being devoted to the distinction between self-concept and self-esteem (e.g. Hattie, 1992) and the character of putative sub-domains (e.g. James, 1892; Luk & Bond, 1992). However, such distinctions are very marginally relevant to this review. Here, for the sake of simplicity as argued in the previous section, we will consider self-concept and selfesteem as synonyms, as also suggested by Josephs, Markus and Tafarodi, (1992) and Marsh (1994), and we will regard the construct as a single factor, consistent with the most commonly used scale (Rosenberg, 1968).

It is generally agreed that self-esteem is a cognition (Sherwood, 1965) employed to evaluate our overall worthiness by comparing our perceived self against our ideal self (Endo, 1992). Included in this self-evaluation is personal dignity and merit (Chrzanowski, 1981), social confidence and ability (Fleming & Courtney, 1984), capability and success (Coopersmith, 1981), and personal respect (Rosenberg, 1968). Thus, self-esteem is our own evaluation of self (Gaulin & McBurney, 2001).

The link to the social world is obvious from these descriptions. Thus, for example, Leary and colleagues (Leary & Downs, 1995; Leary, Schreindorfer & Haupt, 1995) consider self-esteem to be a "sociometer" or gauge of our social standing within important social groups. According to these authors, rather than being motivated to enhance our self-esteem, we are motivated to avoid exclusion from important social groups. Consequently, evolutionary psychologists argue that self-esteem would be better conceptualised as our self-evaluation of our status in the hierarchy, particularly in relation to skills that are valued in the hierarchy (Gaulin and McBurney, 2001). Importantly, such a conceptualisation has universal applicability (see Heine, Lehman, Markus, & Kitayama, 1999 for review).

In terms of its description as a homeostatic device, Kitayama, Markus, Matsumoto and Norasakkunkit (1997) have noted the "numerous studies" that demonstrate a robust and pervasive motivation to maintain and enhance self-esteem. The centrality of self-esteem as a cognitive buffer has also been reinforced through its link to Terror Management Theory (see Harmon-Jones, Simon, Greenberg, Soloman, Pyszcynski & McGregor, 1997). According to this theory, the juxtaposition of the instinctive drive for self-preservation and awareness of the inevitability of death has the potential to cause paralysing terror. Obviously, therefore, this needs to be managed, and Greenberg et ai. (\ 993) accord self-esteem a central role. They proposed a dual-component anxiety buffer that comprises (a) a cultural world-view, which imbues life with meaning, permanence, stability, and provides a set of values by which individuals can be evaluated, and (b) self-esteem, which is achieved by living up to the standards derived from the worldview to which one subscribes. Thus, the perception that one is good (high self-esteem) is associated with security, while the perception that one is bad (low self-esteem) is associated with intense anxiety. This is further supported by the predictable relationships between self-esteem and the personality traits of neuroticism and extraversion as shown in Table I.

From this line of argument it seems intuitive that people with high self-esteem will show SWB resilience in the face of negative life events, and this has indeed been found to be the case (see, e.g. DuBois et ai, 1998: Long & Spears, 1998). Moreover, Brown and Mankowski (\993) have provided further understanding of


the cognitive processes involved. They argue that people with low self-esteem react to events in an even-handed fashion. That is, positive events produce positive reactions and negative events produce negative reactions. This is not so, however, for people with high self-esteem. Such people embrace positive events but reject, limit or offset negative events. As a consequence their reactions to negative events are less severe.

It is not surprising, therefore, that several previous authors have considered SWB to be a determinant of self-esteem (e.g. Pugliesi, 1988). Certainly the link between self-esteem and SWB is legendary, with most authors reporting correlations that are so high (e.g .. 77, Boschen, 1996; .62, Coyle, Lesnik-Emas & Kinney, 1994; .78, Ralph et ai., 1995) that Diener and Diener (1995) were moved to report a study establishing that the constructs were distinguishable from one another.

Given the aforesaid connections and consequences of self-esteem, it is not difficult to imagine that high self-esteem is self-perpetuating through the kinds of behaviours such people are motivated to perform. Consider, for example, the following paraphrased from Coopersmith (1967, pp. 70-71). He concluded "people with high self-esteem approach tasks and persons with the expectation that they will be well received and successful, they follow their own judgement and consider novel ideas, they hold a conviction they are correct and the courage to express those convictions, they have high social independence and creativity, are assertive and likely to take "vigorous social actions". They are more likely to be participants then listeners in group discussions, report less difficulty in forming friendships, have a lack of self-consciousness and a lack of preoccupation with personal problems." The tight links between self-esteem, extraversion, neuroticism, and SWB mentioned earlier seems easy to understand in the light of such attributes.

Self-Esteem and Motivation

Consistent with the argument that self-esteem has a high level of determination from personality, are the proposals of Self-Consistency Theory (Lecky, 1945). According to this theory, individuals employ different buffering strategies dependent on their set-paint-range of self-esteem. Thus, as argued by Brown, Bayer and Brown, (1988), both high and low self-esteem people engage in the process of self-enhancement, but they do so in different ways. People with high self-esteem engage in direct forms of self-enhancement. The self is linked to positive identities and outcomes, and such people over-evaluate their own product rather than devaluing the out-group product. People with low self-esteem, on the other hand, engage in indirect forms of selfenhancement. The self is indirectly linked to positive identities and outcomes by virtue of one's association with others. These people show little evaluation of their own product but derogate the out-group product (see e.g. Hogg & Sunderland, 1991).

In summary, both high and low self-esteem is associated with motivation to engage in particular behaviours and cognitive strategies. The precise behaviours, however, differ between the two groups and, most importantly, are directed to maintain the status quo. In other words, self-esteem has a strong personality component (Campbell et ai., 1966; Hattie, 1992) and, as with neuroticism and


extraversion, is highly stable over time (e.g. Block & Robins, 1993; O'Malley & Bachman, 1979).

However, self-esteem does exhibit short-term variation due to either intentional intervention (e.g. Omizo, Omizo & D'Andrea, 1992) or life events, and several authors have suggested self-esteem lability as a predictive index for depression (e.g. Butler, Hokanson & Flynn, 1994; Roberts & Gotlib, 1997). This is interesting and consistent with the homeostatic model. Low self-esteem would be predicted to correlate with low values of the other cognitive buffers (see later section), reflecting a weak-buffering system that is susceptible to homeostatic defeat by negative life events. It is thus rendered labile and, as a consequence, self-esteem should be highly negatively correlated with depressive symptoms. This has been found to be the case (e.g. Lau, Chi, & McKenna, 1998; Penland et aI., 2000; Rawson, 1992). Indeed, as previously noted, the constellation of high anxiety, high neuroticism, high depression, and low self-esteem has been found to co-vary so predictably in Western cultures as to be referred to as "negative affectivity" (see review by Watson & Clark, 1984).

In summary, research evidence strongly supports relationships between selfesteem and SWB as depicted in Figure 3. Specifically, self-esteem appears to be a powerful determinant of SWB, is strongly influenced by personality, and acts as a cognitive buffer.

Perceived Control

There is a broadly held view that perceived control is central to life quality. The most common classification system for perceived control is its division into internal and external locus of control (Rotter, 1966). Indeed, along with self-esteem, the locus of control construct, as originally conceptualised by Rotter is among the most researched constructs, having been incorporated into literally thousands of studies (Rotter, 1990).

Internal control describes a perception that control over an event rests with the person. External control attributes such control to an external agency, such as luck or some powerful other person or force. At least within Western culture, it is generally assumed that people prefer internal control, and so the presence of external control is indicative that the person is attempting to cope with events that lie outside their sphere of actual control. Kaplan De-Nour (1981), for example, reported higher levels of external control among people on dialysis than the general population.

Perceived control is proposed to operate as a cognitive buffer in the following manner. Under normal conditions, where the person believes their environment is under their control, they will evidence internal control. This, in tum, reinforces SWB. Thus, Phares (\ 978) in a review on this topic describes "the typical internal to be one who actively comes to grips with the world. Compared to the external, the internal is resistant to social pressure and dedicated to the pursuit of excellence". (p. 295).

However, the extrinsic environment cannot always be perceived as under one's personal control, such as when negative life events must be endured. This is when the perception of external control can act as a buffer. That is, if negative life events


were simply accepted as evidence of a complete loss of control this would be indicative of helplessness and very damaging to SWB. If, on the other hand, such events can be understood as simply bad luck (i.e. that the event is unlikely to recur) or the will of God (i.e. that there is a higher, but not understood purpose underlying the event), then the negative feelings associated with the event can be reduced, and the impact on SWB diminished. This is the proposed role of external control, not to directly enhance SWB but to buffer the potential negative impact of negative life events.

The capacity of the buffer is, however, limited in this regard, and this capacity will be exceeded by a strong and protracted negative event. As a consequence, high levels of external control will most commonly be linked to a measurable reduction in SWB, even though the model predicts such a reduction to be less than would be the case in the absence of the external control buffer. These predictions can be readily confirmed. For example, a high internal locus of control has been linked to happiness (Kopp & Ruzicka, 1993; Mullis, 1992), life satisfaction (Cvetanovski & Jex, 1994; Lewinsohn, Redner & Seeley, 1991; Sloper, Knussen, Turner & Cunningham, 1991), and better capacity to cope with various stressful situations, including marital problems (Miller, Lefcourt, Holmes, Ware, & Saleh, 1986).

Conversely, a low internal locus of control has been linked to anxiety (Cvetanovski & Jex, 1994; Lefcourt, 1976; Rawson, 1992) and general malaise (Sloper et aI., 1991). A low internal locus of control has also been linked to depression (e.g. Cvetanovski & Jex, 1994; Lefcourt, 1976; Rawson, 1992). This is consistent with the pattern of relationships between the variables that have been described, and also the view of a fragile buffering system that has lost much of its resilience to external challenge, a view also proposed by Lefcourt (1976) in his review.

Control and Personality

According to our model, perceived control is directly influenced by personality variables whilst remaining distinct from them. Evidence for this comes from the review by Lefcourt (1976) who concluded that, while locus of control can be changed by intervention or circumstances, there is little evidence for the persistence of such changes. This is consistent with a role for personality in setting the resting level of perceived internal control, and returning control to such levels following homeostatic threat; an analogous argument to that mounted with respect to personality and the set-point-range of SWB. Our model reflects this. When under threat, SWB is proposed to be a direct outcome of the cognitive buffers, which in turn are related to both personality factors and perception of whether or not needs have been met (see Figure 3).

It would also be expected that high internal control would be related to behaviours consistent with extraversion, since both are being proposed as links to the maintenance of SWB. This has been supported by Abbey and Andrews (1985) who found that high internal locus of control related positively to social performance.


Optimism

Optimism is defined as a perception that the future will be to the perceiver's advantage or for their pleasure (for an elaboration see Peterson, 2000; Tiger, 1979). General population studies have consistently shown that people view themselves on an upward path of "life getting better", with the past remembered as less good then the present, and the future anticipated as being better still (Bortner & Hultsch, 1970; Cantril, 1965; Gallup, 1998), even though the strength of this effect may diminish in elderly people (Ryff, 1991; Schmotkin, 1991). The effect may also apply to domainlevel satisfactions. For example, Vaillant and Vaillant (1993) demonstrated present marital satisfaction greater than that of the past.

The most substantial empirical demonstration of optimism as a positive cognitive bias has been provided by Glatzer (1991) who reported the results of a large general population, longitudinal study conducted in West Germany between 1978 and 1984. On each of three occasions people were asked to rate their life satisfaction as they remembered it to be 5 years ago, in the present, and as they anticipated it would be 5 years hence. The mean scores, in %SM units, were as follows: past (73, 71, 74), present (78, 77, 77), future (80, 75, 76). These data do not confirm the future bias, perhaps due to the prevailing uncertainty in the country at that time. However, the "present greater than the past" bias is clearly evident and is consistent with the idea that the optimism buffer helps to maintain SWB through a downward comparison with the past.

Measures of optimism, usually made through the Life Orientation Test (Scheier & Carver, 1985), behave in very similar ways to the control buffer, as previously described. In the first place optimism shows a high and robust relationship with internal control. It also shows a strong link with the personality traits and selfesteem, as argued by Tiger (1979) and Peterson (2000) in their major reviews of this topic. Moreover, as expected, it is predictive of life satisfaction (e.g. Fitzgerald, Tennen, Affleck & Pransky, 1993) and relates inversely to depression (e.g. Pyszczynski, Greenberg & Holt, 1987; Scheier & Carver, 1985).

The buffering aspect of optimism lies within the global expectation that things are going to get better with time and in the tendency to perceive things in a more positive light than may be objectively true (cf. high self-esteem and high extraversion). Thus, the impact of negative events on SWB is reduced by the prospect that the difficulties that are being experienced will not last. For example, O'Brien, VanEgeren and Mumby, (1995) found optimists to both underestimate their susceptibility to health problems and to report lower levels of stress and physical symptoms. Interestingly, however, their frequency of preventative health behaviours was no different from non-optimists. Thus, even though they underreport negatively valanced events, due to the effectiveness of their buffering system, this did not prevent them processing and acting on relevant information in an appropriate and adaptive manner. It must also be acknowledged, however, that optimism may be a risk factor for failure in some situations when it combines with a lack of motivation (e.g. Klaczynski & Fauth, 1996).

In summary, the self-esteem, control and optimism buffers appear to display the necessary characteristics to justify their role as depicted in Figure 3. This is


consistent with SWB being a product of both the butTers and personality, as has been previously described. In addition, the butTers are highly correlated, consistent with the proposal that they constitute a single butTering system for the purpose of SWB output. While it has also been argued that these butTers, through the influence of personality, can be linked to motivation, the most central source of motivation is likely to be the cognitive appraisal of needs.

MET AND UNMET NEEDS

A seemingly obvious determinant of SWB involves met and unmet needs. This connection has been made by many authors (e.g. Avia, 1997), who proposes that well-being arises when events contribute to the meeting of needs and the realization of goals. However, in the proposed scheme of Figure 3, well-being is depicted as having only an indirect relationship to met (or unmet) needs, being mediated by personality and the butTering system. Our model depicts that, when SWB homeostasis is under threat, butTering comes into playas influenced by the nature of the unmet needs and the personality variables. Thus, in association with personality, the butTers act to absorb the impact of the environmental stressor/so In order to explain these predictions which involve an indirect link between needs and SWB, the theories proposing a direct link between needs and SWB will be first examined.

It is generally accepted that the main purpose of needs is to provide motivation, and this idea has been commonly described within the quality of life literature as "Person-Environment Fit" (e.g. Andrews & Withey, 1976; Edwards & Rothbard, 1999; French, Rodgers & Cobb, 1974). Within such schemes, satisfaction stems from the degree of congruence between the environment, as the person perceives it, and the person's needs or aspirations. Such theories link with the idea that satisfaction is a product ofthe congruence between multiple potential needs within a person's life, an idea that has its most elegant form as Multiple Discrepancies Theory (Michalos, 1985). According to this theory, the net satisfaction is a positive linear function of the perceived differences between what one has versus (I) what one wants, (2) what others have, (3) the best one has had in the past, (4) what one expected to have in the past, (5) what one expects to have in the future, (6) what one deserves, and (7) what one needs. The resultant net satisfaction, emanating from the meeting of such multiple needs, has been tied to motivation through the assumption that the desire for satisfaction motivates people to act (Mallard et aI., 1997). These authors state "This applies to satisfaction with income, health, education, and other life facets, as well as with satisfaction with life as a whole." (p. 260).

While it seems eminently reasonable to propose that unmet needs provide the basis for motivation (see also Ryan & Deci, 2000), the idea that they are directly linked to SWB levels is problematic. Assume, for example that needs exist in some approximately hierarchical form, such as proposed by Maslow (1954). Then it is assumed that, as most basic needs are met, higher level needs arise which, in Maslow's hierarchy, culminate in needs ofself-actualisation.

In such a scheme, because needs can be consciously recognized and met, SWB would show marked fluctuations as unmet needs arise, and are then met. But this does not occur. SWB evidences a remarkable level of stability. So in order to


account for such stability it must be assumed either that some needs cannot be met or that met needs are seamlessly replaced by unmet needs, so as to maintain an average SWB level of75%SM.

Both of these ideas seem implausible. First, it is difficult to imagine how such a scheme could work to manage SWB with the required precision. Second, since the majority of people in Western society do not experience chronic unmet needs at the lower levels of Maslow's hierarchy, and have their relationship needs reasonably well met, the assumption of "unmet" needs, required to fulfil the average 25%SM deficit from complete life satisfaction, must be maintained via unmet "selfactualisation" needs. There is no evidence of which we are aware to support such an idea.

There is an alternative, and this is depicted in Figure 3. In this scheme, SWB is under the direct control of personality and the indirect control of the buffers, which together act to maintain SWB within its normal range. The varying needs in this model, act directly on the buffers. As a consequence, the buffers have the role of absorbing the impact of unmet needs and maintaining a constant level of SWB. In this scheme, chronic met needs will normally have little influence on SWB levels and, in such circumstances, the buffers will typically not be activated. While such circumstances do not necessarily actively involve the buffers, they nevertheless, may cause the buffers to be more resilient to negative extrinsic influences. Consequently, the level of SWB will remain determined by the dispositional set-point-range.

What may, however, affect the levels of SWB is the presence of a strong, chronic, unmet need that acts to compromise the buffers, and thereby to reduce SWB. For example, a person who is chronically very hungry or insecure is likely to have a level of SWB that lies below the normal range. So, what is being described are two distinct systems that interact in the control of SWB. The motivational system, which has as its prime purpose the initiation of behaviour to meet needs, interacts with the SWB homeostatic system, which has as its prime purpose the maintenance of SWB.

To summarize, it is proposed that, in the absence of strong unmet needs, the SWB homeostatic system will deliver a level of SWB that is within its set-pointrange. The presence of strong unmet needs can defeat this system and, therefore, cause SWB to fall below its normal range. However, the meeting of needs, as a long term state, will not cause SWB to rise above its set-point-range except for brief periods prior to adaptation.

What, then, is the nature of the interaction between these two systems? As depicted in Figure 3, the motivational system (perception of whether or not needs have been met) has a direct link with the buffers. This link is stronger when needs are perceived as not being met. The differential strength of the buffers combines to absorb the impact of the homeostatic threat arising from the unmet needs, and so act to maintain SWB within its set-point range.

MET NEEDS AND THE BUFFERS

Figure 3 indicates a link between met needs and the buffers. This is intended to imply that met needs strengthen the cognitive buffers. For example, a strong


intimate relationship or an above average income, each constituting a powerful met need, is likely to engender a sense of perceived control and optimism.

Indirect evidence for this proposition emerged from a review of personal wealth and SWB (Cummins, 2000a). It was found that while, as expected, very low income increased the probability of homeostatic defeat and therefore low SWB, a very high income induced levels of SWB to approximately 80%SM on a group basis. Such levels were shown to be significantly higher than the population average.

The explanation offered rested on differential access to general resources, not only to meet needs but also to avoid or attenuate negative events. In essence, people who are very poor are highly vulnerable to their environment. For example, they may be required to perform tasks, such as house cleaning, which they may not wish to perform. Also, they may be less likely to effectively counteract the influence of major negative events, such as the theft of property. People who are very wealthy, on the other hand, are able to use their money to acquire resources that allow them to be minimally affected by such concerns. They can employ a cleaning person and can afford insurance against property loss. Thus, while such wealth cannot induce SWB above the set-point-range, it can allow wealthy people to approximate their SWB potential.

Additional supportive evidence has been provided by Wolinski, Coe, Miller & Prendergast, (1985). Over the course of a one year longitudinal study they found that people with high or low levels of SWB at the time of initial measurement scored lower and higher, respectively, five or 12 months later. Such a result would be expected as a consequence of regression to the mean. Interestingly, however, they also found that socio-economic status (SES) over-rode this effect, in that there was a direct relationship between socio-economic status and SWB improvement over the course of the study. The authors' conclusion is consistent with that emerging from the aforementioned review. They state that their results "suggest a vicious cycle wherein lower SES elderly people will increasingly face declining SWB, while their higher SES counterparts will somehow be able to avoid and/or compensate for the circumstances and problems that would otherwise result in reducing their subjective well-being" (p. 102).

In summary, unmet needs are seen as an important motivational component within the SWB homeostatic system, having their origin in the desire for satisfaction. However, the meeting of needs is not regarded a direct predictor of SWB. Rather, the status of needs (met or unmet) is a predictor of buffer activation. Importantly, the perception of whether or not important needs are being met is strongly influenced by personality (see Figure 3) as argued above. This perception is also influenced by adaptation to extrinsic circumstances.

ADAPTATION

The most basic process in the homeostatic model is adaptation, suggested by Andrews and Withey (1976) to be essential to any full explanation ofSWB. This is a psychological process that allows people, over some period of time, to experience a reduced reaction to some changed life circumstance that comes to represent their extrinsic experience. The most popular explanation of how adaptation occurs in


response to major life events involves Adaptation Level Theory (Helson, 1964). Excellent descriptions of this process have been supplied by Brickman and Campbell (1971), Brickman, Coates and Janoff-Bulman, (1978), and Zautra and Goodhart (\ 979), which should be consulted for a detailed account. Excellent recent reviews have been provided by Frederick and Loewenstein (1999), and Kahneman (1999).

In brief terms, the basic principle of Adaptation Level Theory is that people's judgements of their current level of stimulation (positive or negative) depends upon whether this stimulation is higher or lower than the level to which they are accustomed. If it is higher, for example, then the person immediately experiences change. However, two processes then act to make the future experience of this new level of stimulation seem less remarkable. The first is an upward shift in adaptation level because the novel stimulation has been added to the "accustomed" level of stimulation. The second is habituation, which describes the idea that events are judged by the extent to which they deviate from adaptation level.

The most widely cited study to exemplify the operation of these processes is Brickman et al. (1978). The popularity of this citation continues unabated despite the methodological limitations documented by Headey and Wearing (1989). Brickman et aI., compared the reported happiness of major lottery winners, people who had been rendered paraplegic, and a general non-event control group. The data from the former two groups were collected some I to 18 months following the event and, remarkably, when the three groups were compared the differences in SWB were less than might be expected.

The explanation, in terms of Adaptation Level Theory is as follows: The lottery winners' initial euphoria and changed life-style caused a massive upward shift in adaptation level. As a consequence, the new pleasures lost their capacity to excite, and the ordinary events of their previous lives, that had been sources of pleasure, were now unable to do so because they fell below the new adaptation level. In contrast, the reduced level of positive stimulation experienced by people who had acquired paraplegia, caused their adaptation level to fall. As a consequence, these people increasingly experienced pleasure from minor positive events that would have previously gone unnoticed. Thus, the changed adaptation levels of these groups had assisted SWB recovery.

Authors have offered several extensions to this theme in terms of what, precisely, adapts. Thus, Zautra and Goodhart (1979) incorporate comparisons made with others, while Headey and Wearing (\ 986) incorporate role performance. No doubt there are many other psychological processes that could be nominated as being susceptible to adaptation, but it is interesting to note that the theory has not gone unchallenged, and some empirical studies even appear to report data inconsistent with predictions based on Adaptation Level Theory.

The earliest such data were reported by Andrews and Withey (1976) who asked people to rate their SQOL five years ago, in the present, and five years hence. They found the past and future estimations to be virtually independent, but that the present estimation correlated much more strongly with the future (.44) then with the past (.25) estimations; a three-fold difference in shared variance. This could be interpreted as evidence against Adaptation Level Theory. That is, assuming that


current adaptation level had incorporated past levels of SWB, this should make past and present estimates co-vary. Future estimates, on the other hand, could not influence current adaptation level, and so would be expected to have lower level of shared variance. Such a proposition, however, fails to take account of the optimism buffer which predicts the opposite pattern. That is, present SWB and levels of optimism are normally highly correlated. Therefore, since future estimates of SWB and optimism must be considered virtually synonymous, present and future estimates of SWB should be strongly correlated. So, the fact that Andrews and Withey found the pattern they did, more likely reflects the influence of optimism than adaptation level.

A second study that provided data seemingly against adaptation level was that by Zautra and Reich (1980). They reported that people who experienced a high incidence of self-originated positive events actually rated the pleasantness of mundane events higher than people with low incidence. The authors interpreted this as evidence against Adaptation Level Theory since the experience of positive selfgenerated life events appeared to enhance the pleasure gained from mundane events. However, there are problems with such an interpretation. Perhaps most important is the understanding, as argued above, that people with a high set-point-range for SWB are predicted to experience more positive events and also to rate the events in their lives as more pleasurable. In this context, the findings by Zautra and Reich can be interpreted as confirmation of the high personality contribution to SWB.

Other evidence apparently against Adaptation Level Theory is the finding by Hunskaar and Vinsnes (1991) that the QOL of women with incontinence was not related to the duration of their condition. However, two matters make such data insubstantial for this purpose. First, the Sickness Impact Profile (Bergner, Bobbitt, Carter & Gilson, 1981), which they used as a measure of outcome, comprises mainly functional status and objective QOL indices. It is certainly not a measure of SWB. Second, some aversive circumstances do not readily permit adaptation, and the social and practical considerations imposed by incontinence may well fall into this category.

While this cannot be a complete list of relevant studies, those above are the only ones known to the authors that could be cited against the theory. On the other hand, many others can be cited in support of the theory. In addition to the Brickman et al. (1978) report, which has already been cited, Rigby, McCarron and Rigby (1990) who studied people who were both disabled and living in an institution. In support of Adaptation Level Theory they found less negative affect among the longer-stay residents. However, they found no such difference in positive affect.

It is worth noting, however, that this is a difficult area to research for a number of reasons. One is that evidence for adaptation cannot be reliably sought from the presentation of non-significant data. Take, for example the study by Borgen, Amundson and Tench (1996). These authors followed-up high school graduates for two years and found that neither employment, unemployment, nor study continuation differentially influenced either self-esteem or depression. However, several factors other than the success of adaptation could have caused this result, including the fact that the authors used a relatively insensitive form of analysis.


It also needs to be acknowledged that, given the homeostatic processes, it is going to be difficult to disentangle adaptation to extrinsic events from other forms of adaptation occurring elsewhere in the model, such as at the level of needs. But for the purpose of this paper it is clear that adaptation to altered circumstances of living does occur through one means or another, and that such processes are involved in maintaining the set-point-range ofSWB, as has been described.

Just how long the system takes to adapt to changed circumstances will, no doubt, depend on such parameters as the perceived magnitude of the event, individual differences, and the resilience of the system. Nevertheless, some indication of the time-span can be derived from the study by Suh et al. (1996) on college students. This combined longitudinal and cross-sectional study found that only events within the previous six months exerted an influence on SWB. This appears a reasonable starting point for such estimations.

In summary, Adaptation Level Theory has a strong level of acceptance within the literature. It thus seems reasonable to incorporate it into our Homeostatic model.

HOMEOSTATIC MAINTENANCE AND FAILURE

The maintenance of SWB within its set-point-range requires some kind of feedback loop. While, the nature of such a device is speculative at this stage, a recent suggestion by Carver (2000) provides one explanation for how the set-point-range might be maintained. Carver's idea involves the assumption that action is goaldirected and managed by feedback loops. Some of the loops are goal seekers, others are "anti-goal" avoiders. These loops make comparisons between current conditions and the goal (or reference point) and make adjustments as necessary. As a consequence, in the absence of distractions or other forces, people tend to do what they intend to do (goal seeking and matching) and avoid doing the things they regard as unpleasant (anti-goal avoiding). According to Carver, "The affect portion of the theory uses much the same logic but with the incorporation of time as a parameter. It proposes a loop that monitors the rate at which the behavioral systems (goal seeking and anti-goal avoiding) are doing what they are trying to do. . ..... The loop takes sensed velocity and compares it to a velocity goal. If the sensed velocity is less than the reference point, the result is negative affect; if it exceeds the reference point, the result is positive affect." (p. 339).

In the context of the homeostatic model, it could be supposed that SWB lying either above or below the set-point-range constitutes a deviation from the velocity goal. Such deviations will then lead to internal adjustments that will, eventually, restore SWB to a level within its set-point-range. Thus, in terms of the goal-seeking loop, which Carver (2000) presumes to be the affect loop, levels of SWB lying above the set-point-range should lead to "coasting". That is, high levels of SWB should lead to withdrawal of some effort, basking, or turning to some other behavioural domain. Levels of SWB lying below the set-point-range should lead to the more intense goal-seeking motivation or behaviour. Thus, this system acts to prevent the maintenance ofSWB levels outside the set-point-range.

While this proposition is speculative at this stage, it does represent an attempt to explain the nature of the feedback loops underpinning the homeostatic set-point-


range. Such considerations, however, concentrate on the chronic maintenance of SWB within its set-point-range. Consideration will now be given to variations of SWB that lie outside this range.

JOY

The buffers do not normally allow SWB to reach its absolute maximum (IOO%SM). The buffers are constrained by their own set-point which has, as its cognitive manifestation, the understanding that we do not have absolute control, that we are not perfect, and that the future will not be uniformly positive. However, higher than normal SWB can be induced in the short-term by either meeting a need (e.g., having a satisfying meal) or experiencing a highly positive life event (e.g., being promoted at work). Such situations represent homeostatic failure at the level of the cognitive buffers. In support of this idea, Mikulincer and Peer-Golden (199 I) found that shortterm happiness reflected situational validation of the ideal self ("the person you would like ideally to be"). They found that happiness in a given event increased with the congruence between the ideal self and the perceived self in that event. We therefore hypothesize that other forms of situational validation, involving ideals based on control or optimism, can also yield above-range levels of SWB. Such deviations will, however, be short-lived due to the influence of adaptation.

DEPRESSION

At the other end of the spectrum, there is depression. If depression is conceptualised as the loss of SWB homeostasis, then the relationship between dysthymia and SWB should approximate the relationship between SWB and extrinsic factors as shown in Figure 2. That is, provided that some negative influence lacks the power to defeat homeostasis, the symptoms of depression should be virtually absent. As Kammann and Flett (1983) commented, "an absence of depression can occur with many different levels of positive well-being." (p. 261). This changes, however, as the negative influence starts to induce dysthymia by defeating homeostasis and forcing SWB to lower levels. Under such conditions, the lower values for SWB and the symptoms of depression start to covary in the manner depicted by Figure 2. Thus, the model predicts a curvilinear relationship between extrinsic conditions and depression, precisely as has been reported in relation to depression and marital satisfaction (Fincham & Bradbury, 1993), life satisfaction (Coyle et aI., 1994; Lewinsohn et aI., 199 I), and positive affect (Kammann & Flett, 1983).

A further prediction from the model is that, under conditions of homeostatic defeat, the symptoms of depression (i.e. scores on depression indices) are coincident with low values within the buffer systems and related personality trait constellations (e.g., low self-esteem, high neuroticism). This pattern of relationships has been verified in relation to self-esteem, both within a general population sample (Fincham and Bradbury, 1993) and for people with an intellectual disability (Benson and Ivins, 1992). Thus, as the buffer systems fail to successfully absorb the impact of negative events (which will, to some extent be influenced by personality), their own values


decrease and the consequential fall in SWB is mirrored by a rise in the indices of depression.

The model also predicts that the level of the set-point-range should predict the extent to which the homeostatic system is robust to negative life events. A relatively high setting should indicate a high level of resilience, while a low setting predicts a fragile system that is prone to failure and, therefore, to depression. This prediction has already been discussed and further support comes from Lewinsohn et al. (1991) who found low levels of life satisfaction to be a significant risk factor for future depression.

It might also be expected that people with a high level of neuroticism would be predisposed to depression. This is consistent with the logic that their personality structure would not only give them a low SWB set-point-range but would also predispose them to reward insensitivity and relative inability to counteract the negative influence of stressful situations (Flynn & CappeJiez, 1993). Just such a finding has been reported by Roberts and Kendler (1999) in a large study of twins. They found neuroticism to be a far stronger predictor of risk for major depression than self-esteem. In fact, when considered together, the predictive power of neuroticism remained while the predictive power of self-esteem disappeared. This is also consistent with the idea that the power of the buffering system is determined by personality.

Finally, it might also be anticipated that people with high neuroticism could be considered to have a "depressogenic" attributional style, characterized by a fragile buffering system. For such people, negative life events are very likely to be experienced as stressful, as they are more likely than normal to exceed the threshold of the buffers. So, in this sense, such people "generate" negative life events (see Simons, Angell, Monroe & Thase, 1993) with the consequence that their lives are characterized by a predictably high frequency of such events.

In summary, both joy and depression can be regarded as consequences of homeostatic failure. However, whereas joy is inevitably an acute state, quickly defeated by homeostatic processes, depression can be chronic, representing the failure of homeostasis in the face of powerful and persistent negative experience. This depiction of depression, as representing the loss of SWB through homeostatic failure, is novel.

CONCLUSIONS

The idea that it is normal and adaptive to view one's life positively is revolutionary. As has been pointed out by other authors in this area (e.g. Peterson, 2000), such an idea fundamentally challenges the ideas of many influential psychologists and psychiatrists from the last century as to what constitutes desirable psychological functioning. Most essentially, the idea that the accurate perception of reality equates to optimal psychological functioning is clearly wrong. People who see their lives accurately are generally depressed. Thus, since moderately high levels of SWB are normal, there must be some robust psychological machinery to ensure that such levels are maintained.


This chapter has attempted to describe a model that could account for such maintenance. It draws together several major psychological constructs into an interlocking system and proposes the idea of a homeostatic system that manages SWB. In this system, personality provides a stable affective background that sets SWB within a narrow range for each individual. This is achieved through the supply of a direct affective component to SWB and through the influence that personality exerts on other components of the homeostatic system.

The cognitive buffers are proposed to be under the influence of personality, which acts to provide them with a determined range of operation such that SWB is held within its set-point-range. Personality is also envisaged as having a powerful influence on motivational systems that seek satisfaction and, thus, predispose behaviour that is likely to maintain normal levels of SWB. This is mainly through the link between extraversion and reward sensitivity, but also in relation to the general influence of personality on the level of perceived needs. Finally, at the most fundamental level of the homeostatic system, are the processes of adaptation and habituation, which constitute the first line of defence against the threat of changed extrinsic conditions influencing levels ofSWB.

In the construction of this chapter, all accessed and relevant publication have been incorporated into the description. While these appear to overwhelmingly offer support for the model, an unconscious bias to generate such fit must be considered as a limitation to our judgement. We also acknowledge that only a fraction of available evidence has been cited, given the breadth of the psychological processes under consideration. It is, therefore, certain that alternative views can be formed on the interpretation of the cited publications in respect of the model, and that other data may cause a re-evaluation of the structure that has been described. We look forward to such developments and offer this homeostatic model only as a first step in our understanding of how it is that SWB is so effectively maintained.

Address for Correspondence: Professor R.A. Cummins, PhD, School of Psychology, Deakin University, Burwood, Victoria 3125, Australia, Email: robert. cumm ins@deakin. edu. au

REFERENCES

Aaronson, N.K., Ahmedzai, S., Bergman, B et al. (1993). The European OrganizatIon for Research and Treatment of Cancer QLQ-C30: A quality ofhfe instrument for use in international clinical trials in oncology. Journal of the NatIOnal Cancer Instllute. 85. 365.

Abbey, A, & Andrews, F. M. (1985) Modeling the psychological determinants oflife quality. SocIal IndIcators Research. 16. 1-34.

Andreasen, N. J. c., & Norris, AS. (1972) Long-term adjustment and adaptation mechanisms in severely burned adults. The Journal of Nervous and Mental DIsease. 154. 352-362.

Andrews, F.M., & Withey, S.B. (1976). SOCial md,cators ofwell-bemg: American's perceptions of life quality. New York: Plenum Press.

Argyle, M , & Lu, L. (1990). The happmess of extraverts Personality and IndIVIdual Differences. II. 1011-1017.

Avia, M.D. (1997). Personality and positive emotions. European Journal of Personality. II. 33-56.

R.A. CUMMINS, E. GULLONE AND A.L.D. LAU

Bach, J.R, & Tilton, M.e. (1994). Life satisfaction and well-being measures in ventilator assisted individuals with traumatic tetraplegia. ArchIves of PhysIcal Medlcme and RehabilItatIOn, 75, 626-632.

Benson, B.A. & Ivins, J. (I992). Anger, depression and self-concept In adults with mental retardation Journal of Intellectual DIsabIlity Research, 36, 169-175.

Bergeman, e.S., Plomin, R., Pedersen, N.L & McClearn, G.E. (1991) Genetic mediation of the relationship between social support and psychological well-being Psychology & Agmg, 6, 640-6.

Bergner, M., Bobbitt, R. A., Carter, W. B., & Gilson, B. S. (I981). The Sickness Impact Profile' Development and final revIsion of a health status measure. MedIcal Care, 19, 787-805

Block, 1., & Robins, R.W. (1993). A longitudinal study of consistency and change In self-esteem from early adolescence to early adulthood. Ch,ld Development, 64, 909-923.

Bloom, J.R., Fobair, P, Spiegel, D, Cox, R. S, Varghese, A., & Hoppe, R. (1991). Social supports and the social well-being of cancer survivors Advances m MedIcal SOCIOlogy, 2. 95-1\4.

39

Borgen, W A., Amundson, N.E., & Tench, E. (\996). Psychological well-being throughout the transition from adolescence to adulthood. Career Development Quarterly, 45, 189-199.

Bortner, R.W., & Hultsch, D.F. (1970). A multivariate analysis of correlates ofhfe satisfaction in adulthood. Journal of Gerontology, 25, 41-47.

Boschen, K.A. (\996). Correlates of life satisfaction, residential satisfaction, and locus of control among adults with spinal cord injuries. RehabIlitation Counselmg Bul/etm, 39, 230-243.

Bowling, A. (1996). Associations with changes in life satisfaction among three samples of elderly people hving at home InternatIonal Journal ofGenatnc PsychIatry, II, 1077-1087.

Bradburn. N.M. (1969) The structure of psychological well-being. Chicago' University of Chicago Press. Brickman, P., & Campbell, D. T. (197\) Hedonic relativism and planning the good society. In M. H.

Appley (Ed.) Adaptaflon-leveltheory: A sympOSIUm, (pp. 287-302). Academic Press, New York. Brickman, P., Coates, D, & Janoff-Bulman, R. (1978) Lottery winners and accident victims: Is happiness

relative? Journal of Personality and SOCIal Psychology, 36, 917-927. Brief, A.P., Butcher, A.H., George, J.M., & Link, K.E (1993}.lntegrating bottom-up and top-down

theories of subjective well-being: The case of health. Journal of personalIty and SOCIal Psychology, 64,646-653

Brown, R.I., Bayer, M.B & Brown, P.M. (1988). Quality ofhfe: A challenge for rehabilitation agencies. Australian and New Zealand Journal of Developmental Dlsab,ltfles, 14, 189-199 ..

Brown, J.D., & Mankowski, T.A. (l993). Self-esteem, mood, and self-evaluation: Changes in mood and the way you see you. Journal of Personality and SOCIal Psychology. 64, 421-430

Butler, A. C., Hokanson, J. E., & Flynn, H. A. (1994) A comparison of self-esteem lability and low trait self-esteem as vulnerability factors for depression. Journal of Personality and SOCIal Psychology, 66, 166-177.

Campbell, A., Converse, P. E., & Rodgers, W. L. (1976). The quality of American life: PerceptIOns, EvaluatIOns, and SatlsfactlOns._New York: Russell Sage FoundatIOn.

Campbell, J.D., Trapnell, P.D., Heine, SJ., Katz, I.M., Lavallee, L.F., & Lehman, D.R (1996). Selfconcept clarity: Measurement, personality correlates, and cultural boundaries. Journal of Personality and SOCIal Psychology, 70, 141-156.

Cantril, H (1965). The Pattern of Human Concerns. New Jersey: Rutgers University Press. Carver, C.S. (2000). On the continuous calibration of happiness. Amertcan Journal of Mental

RetardatIOn, 105, 336-341. Chrzanowski, G. (1981) The genesis and nature of self-esteem Amertcan Journal of Psychotherapy, 35,

38-46. Cohen, S, Kamark, T., Mermelstein, R. (1983) A global measure of perceived stress Journal of Health

and SOCIal BehaVIor, 24, 385-396. Cooper, J. (I998). Unlearning cogmtive dissonance: Toward an understanding of the development of

dissonance, Journal-of-Experlmental-Soclal-Psychology, 34, 562-575. Coopersmith, S. (1967) Some expressions of self-esteem. In W H. Freeman (Ed ), The antecedents of

self-esteem. San Francisco Coopersmith, S. (1981). Manualfor the Coopersmllh Self-Esteem Inventory (SE1). Consulting

Psychologists Press. Palo Alto, CA" Costa, PT., Jr, & McCrae, R. R. (1988) Personality In adulthood: A six-year longitudinal study of self

reports and spouse ratings on the NEO personality inventory. Journal of PersonalIty and SOCIal Psychology, 54, 853-863.


Costa, P. T., Jr., & McCrae, R. R. (1989). Personality as a lifelong determinant of wellbeing. In L. Z. Malatesta & c. E. Izzard (Eds.). EmotIOn m adult development (pp. 141-157). Beverly Hills, CA: Sage.

Costa, P.T. Jr, & McCrae, R.R. (1992). Revised NEO Personality Inventory (NEO PI-R) and NEO FlveFactor Inventory (NEO-FFI) professIOnal manual. Florida' Psychological Assessment Resources, Inc.

Courneya, K.S., Bobick, T.M., Rhodes, R.E., Jones, L.W., Friedenreich, C.M., & Arthur, K. (2000) Personality correlates of patients' subjective well-being after surgery for colorectal cancer: An application of the five-factor model. Journal of Psychosocial Oncology, 18, 61-73.

Coyle, C. P., Lesnik-Emas, S., & Kinney, W. B. (1994). Predicting life satisfaction among adults with spinal cord injuries. Rehabilitation Psychology, 39. 95-112

Cummins, R.A. (1995). On the trail of the gold standard for life satisfaction. SOCial Indicators Research. 35. 179-200

Cummins, R.A. (1996). The domams of life satisfaction. An attempt to order chaos SOCial Indicators Research. 38. 303-332.

Cummins, R.A. (1997a) Comprehensive Quality of Life Scale - Adult._Manual: Fifth EdItion. Melbourne: School of Psychology, Deakin University. (ISBN 0 7300 2804 6)

Cummins, R.A. (1997b). ComprehenSive Quality of Life Scale - Intellectual!cogmtlve disability. Manual: Fifth Edition. Melbourne: School of Psychology, Deakin University. (ISBN 0 7300 2725 2)

Cummins, R.A. (1997c). Assessing quality oflife for people with disabilities. In R.1. Brown (Ed.). Quality of life for handicapped people. Second edition. Cheltenham, England: Stanley Thomas (p 116-150).

Cummins, R.A. (1998). The second approximatIOn to an international standard oflife satisfaction. SOCial Indicators Research. 43, 307-334

Cummins, R.A .. (1999). A psychometric evaluation of the ComprehenSive quality of Life Scale - fifth Edition. In: L.L. Yran, B. Yuen, & c. Low (Eds ), Urban Quality of Life: Cntical Issues and Options (p p 32-46). Singapore: NatIOnal University of Singapore.

Cummins, R.A. (2000a). Personal income and subjective well-being: A review. Journal of Happmess Studies. 1, 133-158

Cummins, R.A. (2000b). A homeostatic model for subjective quality oflife. Proceedmgs. Second Conference of Quality of Life m CWes (pp. 51-59). Singapore: National University of Singapore

Cummins, R.A. (2000c) Objective and SUbjective qualIty oflife: An interactive model. SOCial Indicators Research, 52, 55-72.

Cummins, R.A. (200 I). Directory of instruments to measure quality of life and cognate areas. Sixth Edition. Melbourne: School of Psychology, Deakin University (ISSN 1325-0752).

Cummins, R A. (2002). NormatIve life satisfaction: Measurement issues and a homeostatic model. Social Indicators Research (in press).

Cummins, R.A., Eckersley, R., Pall ant, J., van Vugt, J., Misajon, R. (2002). Developing a national index of subjective wellbeing: The Australian Umty Wellbemg Index. SOCial Indicators Research (in press).

Cummins, R.A., & Nistico, H. (2002). Maintaining life satisfaction: The role of positive cognitive bias Journal of Happmess Studies. 3. 37-69.

Cvetanovski, J. & Jex, S M. (1994). Locus of control of unemployed people and its relationship to psychological and physical well-being Work & Stress, 8, 60-67.

De Neve, K , & Cooper, H. (1998). The happy personality: A meta-analysis of 137 personality traits and SWB. Psychological Bulletm, 124. 197-229.

DeNeve, K.M (1999). Happy as an extraverted clam? The role of personality for subjectIve well-being. Psychological SCience, 8, 141-144.

Depue, R.A., & Collins, P.F. (1999). Neurobiology of the structure of personality: Dopamine facilitation of incentive motivation and extraversion. Behavioral and Bram Sciences. 22. 491-569.

Depue, R A, & Montoe, S. M. (1986). ConceptualizatIOn and measurement of human disorder in life stress research: The problem of chronic disturbance Psychological Bulletm, 99, 36-51.

Diener, E. (1984) Subjective well-being Psychological Bulletm. 95. 542-575. Diener, E., & Larsen, R J. (1984). Temporal stability and cross-situational consistency of affective,

behaVIOral, and cognitive responses. Journal of Personality and SOCial Psychology, 47, 871-883. Diener, E.D., Emmons, R.A., Larsen, RJ., & Griffin, S (1985) The satisfaction with life scale Journal

of Personality Assessment, 49, 71-75


Diener, E., & Diener, M. (1995) Cross-cultural correlates of life satIsfaction and self-esteem. Journal of Personality and SocIal Psychology, 68, 653-663.

Diener, E., & Emmons, R. A (1985) The independence of positive and negative affect. Journal of Personality and SocIal Psychology, 47, 1105-1117.

Diener, E, Emmons, R.A, Larsen, RJ., & Griffin, S. (1985). The satisfaction with life scale. Journal of Personality Assessment, 49, 71-75.

Diener, E. Sandvik, E., Pavot, W ,& Fujita, F. (1992). Extroversion and subjective well-being in a U.S. national probability sample. Journal of Research In Personaltty, 26, 205-215.

Diener, E., Suh, E.M., Lucas, R.E., & Smith, H.L. (1999). Subjective well-being: Three decades of progress. PsychologIcal Bulletin, 125, 276-302

Doyle, K.O., & Youn, S. (2000). Exploring the traits of happy people. SocIal IndIcator Research, 52, 195 DuBois, D.L., Bull, CA, Sherman, M.D., & Roberts, M. (1998). Self-esteem and adjustment in early

adolescents. A SOCIal contextual perspective. Journal of Youth and Adolescence, 27, 557-583. Duckltt, J. H (1983). Psychological factors related to SUbjective health perception among elderly women.

Humamtas, Journal of Research in the Human SCIences, 9, 441-449. Edgerton, R. B. (1990). Quality of life from a longitudinal research perspective. In R. L. Schalock (Ed.),

Quality of life' Perspectives and Issues (pp. 149-160). Washington:American Association on Mental Retardation.

Emmons, R. A., & Diener, E. (1985) Personality correlates of subjective well-bemg. Personaltty and SOCIal Psychology Bulletin, II, 89-97.

Edwards, lR., & Rothbard, N.P. (I999). Work and family stress and well-being: An examination of person-environment fit in their work and famIly domains. OrgamzatlOnal BehavIOr and Human DeCISIon Processes, 77, 85-129.

Elliott, T.R., Marmarosh, C., & Pickelman, H. (1994). Negative affectivity, social support, and the predictIon of depression and distress. Journal of Personaltty, 62, 299-319.

Endo, Y. (1992). Personalized standard of self-esteem. Japanese Journal of EducatIOnal Psychology, 40, 157-163.

Federick,S., & Loewenstein, G. (1999). Hedonic adaptation In D Kahneman, E Diener & N. Schwarz (Eds.), Well-Being: Thefoundatlons ofhedomc psychology (pp. 302-330). New York' Russell Sage FoundatIOn.

Felce, D., & Perry, J (1995) Quality of life: Its defimtion and measurement. Research In Developmental D,sabilIt,es, 16, 51-74.

Filipp, S., & Klauer, T. (1991). Subjective well-being in the face of critical life events. The case of successful copers In F. Strack, M. Argyle & N. Schwarz (Eds.), Subjective well-being: An interdIscIplinary perspectIVe (pp. 213-234). New York: Plenum Press.

Fincham, F.D., & Bradbury, T.N. (1993) MarItal satisfaction, depression and attributions: a longitudinal analysis. Journal of Personality and SOCIal Psychology, 64, 442-453.

FItzgerald, T. E., Tennen, H., Affleck, G., & Pransky, G. S. (1993). The relative importance of disposItional optimism and control appraisals in quality of life after coronary artery bypass surgery. Journal of Behavioral MediCine, 16, 25-43.

Flanagan, J. C. (1978). A research approach to improving our quahty of life. Amertcan PsychologIst, 33, 138-147.

Flemmg, J S, & Courtney, B.E. (1984). The dimensionality of self-esteem: II. Hierarchical facet model for reVIsed measurement scales Journal of Personaltty and SOCIal Psychology, 46, 404-421.

Flynn, RJ., & Cappehez, P. (1993). An integrative cognitive-environmental view of depression. In P. Cappeliez & RJ. Flynn (Eds.), DepreSSIOn and the SOCIal enVIronment (pp. I-II), Montreal: McGiIIQueens UniversIty Press

Fogarty, GJ., Machin, MA, AJbion, MJ., Sutherland, L F ,Lalor, G.I., & Revitt, S. (1999). Predicting Occupational Strain and job satisfaction The role of stress, coping, personality and affectivity variables. Journal of VocatIOnal BehavIOr, 54, 429-452.

Francis, LJ. (1999) Happiness is a thing called stable extroversion: a further examination of the relatIonship between the Oxford Happiness Inventory and Eysenck's dimensional model of personality and gender. Personaltty and IndIVIdual Differences, 26, 5-11.

French, J. R. P., Jr, Rodgers, W., & Cobb, S (1974). Adjustment as person-environment fit. In G. V. Coelho, B. A. Hamburg, & J. E. Adams (Eds.), Coping and AdaptatIOn (pp 316-333). New York: Basic Books


Fyrand, L., Wichstom, L., Moum, T., Glennas, A , & K vien, T .K. (1997) The impact of personality and social support on mental health for female patients with rheumatoid arthritis SocIal IndIcators Research. 40. 285-298.

Gallup (1998). Have and have-nots. Perceptions of fairness and opportunity 1998 Princeton: Gallop Poll SocIal Audll. The Gallop Organization.

Gaulin, SJ.c., & McBurney, 0 H. (2001). Psychology: An evolutionary approach. New Jersey: Prentice Hall.

Glatzer, W. (1991). Quality of life in advanced industrialized countries The case of West Germany In F. Strack, M. Argyle & N.Schwarz (Eds.), Subjective well-being: An interd,sc,plinary perspectIve (pp 261-279) New York. Plenum Press

Goldberg, L.R. (1992). The development of markers of the Big-Five factor structure. PsychologICal Assessment. 4. 26-42.

Gray, J.A. (1970) The psychophysiological basis of introversion-extraversion. BehaVIOr Research and Therapy. 8. 249-266.

Greenberg. J ,Pyszczynski, T ,Solomon, S., Pmel, E., Simon, L., & Jordan, K. (1993) Effects of selfesteem on vulnerability-denying defensive distortions: Further evidence of an anxiety-buffenng function of self-esteemjournal of Expenmental Social Psychology. 29, 229-251.

Groot, W., & VandenBrink, M. (2000). Life satisfaction and preference dnft SocIal Indicators Research. 50, 315-329.

Hagerty M.R., Cummins. R.A., Ferriss AL., Land K., Michalos AC., Peterson M., Sharpe A., Sirgy J. & Vogel J. (2001) Quality oflife indexes for national policy. review and agenda for research SocIal Ind,cators Research. 53. (in press)

Harmon-Jones, E., Simon, L., Greenberg, J., Solomon, S., Pyszcynski, T, & McGregor, H. (1997). Terror management theory and self-esteem: Evidence that increased self-esteem reduces mortality salience effects. Journal of Personality and SocIal Psychology. 72. 24-36.

Hattie, J. (1992) Self-concept. Hillsdale, NJ: Lawrence Erlbaum Associates Publishers Headey, 8., Holmstrom, E., & Wearing, A (1985). Models of well-being and ill-being. SocIal IndIcators

Research. 17, 211-234. Headey, B., & Wearmg, A (1986). The sense of rei alive superlOnty- central to well-being University of

Melbourne (unpublished). Headey, 8., & Wearing, A (1989). Personality, life events, and subjective well-being: Toward a dynamic

equilibrium model. Journal of Personality and SocIal Psychology. 57. 731-739. Headey, 8., & Wearing, A (1992). Understanding happiness: A theory of subjectIVe well-being

Melbourne: Longman Cheshire. Heine, S J., Lehman, D.R., Markus, H.R., & Kitayama, S (1999). Is there a universal need for positive

self-regard? PsychologIcal RevIew. /06. 766-794. Helson, H. (1964). AdaptatIOn-level theory. New York· Harper & Row. Hogg, M A, & Sunderland, J (1991) Self-esteem and intergroup discrimination m the minimal group

paradigm. BrztlshJournal of SocIal Psychology. 30, 51-62. Hoppock, R. (1935) Job satisfactIOn. Harper & Brothers Pubhshers, New York. Huelsman, TJ., Nemanlck, R.C Jr., & Munz D.C.(l998). Scales to measure four dimensions of

dispoSitional mood: Positive energy, tiredness, negative activation, and relaxatIOn. EducatIonal and PsychologIcal Measurement, 58. 804-819.

Hunskaar, S., & Vinsnes, A (1991). The quality of life in women with urinary incontinence as measured by the Sickness Impact Profile. Journal of the AmerIcan Gertatncs SOCIety, 39. 378-382.

Hunt, S.M., Singer, K., & Cobb, S. (1967). Components of depressIOn ArchIVes of General PsychIatry. 16,441-447

Izard, C.E. (1993). Four systems for emotion activation· Cognitive and noncognitive processes. Psychological RevIew. 100, 68-90.

Jacob, J c., & Brinkerhoff, M.B (1999) Mindfulness and subjective well-being in the sustainabihty movement. A further elaboration of multiple discrepancies theory. SocIal IndIcators Research. 46. 341-368.

James, W. (1892). Psychology. New York: Henry Holt Janoff-Bulman, R. (1989) The benefits of illusions, the threat of diSillusionment, and the limitations of

inaccuracy. Journal of SocIal and ClinIcal Psychology. 8. 158-175. Josephs, R A , Markus, H.R., & Tafarodi, R W (1992). Gender and self-esteem. Journal of Personality

and SocIal Psychology, 63. 391-402


Kahneman, D. (1999). Objective happiness. In D Kahneman, E. Diener & N. Schwarz (Eds.), Well-Being: Thefoundations of hedonic psychology (pp. 3-25). New York: Russell Sage Foundation

Kammann, R., & Flett, R (1983). Affectometer 2: A scale to measure current level of general happiness Australtan Journal of Psychology, 35, 259-265.

Kaplan De-Nour (1981). PredictIOn of adjustment to chronic hemodialysIs. In N.B.Levy, (Ed ), Psychonephrology: psychologlcalfactors in hemadlOlys/S and transplantation Volume I (pp 117-132). New York; Plenum.

Kitayama, S., Markus, H.R., Matsumoto, H., & Norasakkunkit, V. (1997). Individual and collective processes in the construction of the self: Self-enhancement in the United States and self-criticism in Japan. Journal of Personality and Social Psychology, 72, 1245-1267.

Klaczynski, P A, & Fauth, J.M. (1996). Intellectual ability, rationality, and intuitiveness as predIctors of warranted and unwarranted optImIsm for future life events. Journal of Youth and Adolescence, 25, 755-770.

Kopp, R.G., & Ruzicka, M.F. (1993). Women's multiple roles and psychological well-being. Psychological Reports, 72, 1351-1354.

Kwan, V.S.Y., Bond, M.H., & SIngelis, T.M. (1997) Pancultural explanatIOns for life satisfaction: Adding relationship harmony to self-esteem. Journal of Personality and Social Psychology, 73. 1038-1051.

Landua, D (1992). An attempt to classiJY satisfaction changes: Methodological and content aspects ofa longitudinal problem. Social Indicators Research, 26. 221-241.

Larsen, RJ, & Diener, E. (1992). Promises and problems WIth the cIrcumplex model of emotion. In M.S. Clarke (Ed.), Structural Bases of EmollOnal BehavIOr (pp. 25-59). London. Sage.

Lau, A., Chi, I., & McKenna, K (1998) Self-perceived quality oflife of Chinese elderly people In Hong Kong. Occupallonal Therapy International, 5, 118-139.

Leary, M.R., & Downs, D L. (1995). Interpersonal functions of the self-esteem motive: The self-esteem system as a soclOmeter. In M.H. Kemis (Ed.), Efficacy, agency, and self-esteem (pp. 123-144). New York: Plenum.

Leary, M.R., Schreindorder, L S., & Haupt, AL. (1995). The role of self-esteem in emotional and behavioural problems: Why is low self-esteem dysfunctional? Journal of Social and ClImcal Psychology, /4,297-314

Lecky, P.( 1945) Self-consistency: A Theory of personality. New York: Island Press. Lefcourt, H.M. (1976); Locus of control: Current trends In theory and research Hillsdale, New Jersey:

Lawrence Erlbaum. LewInsohn, P. M., Redner, J. E., & Seeley, J R. (1991). The relationship between life satisfaction and

psychosocial varIables: New perspectives In F. Strack, M Argyle & N.Schwarz (Eds.), SubjectIVe well-being: An InterdiSciplinary perspectIVe (pp. 141-1 69).Oxford: Pergamon Press.

Llpkus, I. (1991). The construction and preliminary validation ofa global belief in ajust world scale and the exploratory analysis of the multidimensional belief In a just world scale. Personaltty-andIndiVidual-Differences, 12(11), 1171-11 78

Lipkus, I.M., Dalbert, e., & Siegler, I.e. (1996) The importance of distinguishing the belief in ajust world for self versus for others Implications for psychologIcal well-being. Personality and Social Psychology Bulletin, 22, 666-677.

Long, K.M., & Spears, R. (1998). Opposing effects of personal and collective self-esteem on interpersonal and intergroup compansons. European Journal of SOCIal Psychology, 28, 913-930.

Lu, L., & Shih, 1.8. (1997). Sources of hap pines: A qualitative approach. The Journal of Social Psychology, 137, 181-187.

Lucas, R.E., Diener, E., Grob, A, Suh, E.M., & Shao, L. (2000) Cross-cultural evidence for the fundamental features of extraversion. Journal of Personality and Social Psychology, 79, 452.

Luk, C.H ,& Bond, M.H. (1992). Explaining Chinese self-esteem in terms of the self-concept PsychologlO,35, 147-154.

Mallard, AG.C , Lance, C E., & Michalos, AC (1997). Culture as a moderator of overall life satisfaction- Life facet of satisfaction relatIOnships. Social Indicators Research, 40, 259-284.

Marsh, H.W. (1994). The importance of being important· Theoretical models of relatIOns between specific and global components of physical self-concept. Journal of Sport and ExeYC/se Psychology, 16,306-325.

Maslow, A H. (1954). Motivation and Personality. New York: Harpers


Mellor, D., Cummins, R.A., & Loquet, C. (1999). The gold standard for life satisfaction: Confirmation and elaboration using an imaginary scale and qualitative interview. InternatIOnal Journal of SocIal Research Methodology Theory and Practrce, 2, 263-278

Michalos Ae., (1985). Multiple discrepancies theory (MTD) Social Indicators Research 16, 347-413. Mikulincer, M., & Peer-Goldin, I. (1991). Self-congruence and the experience of happiness. Brr/lsh

Journal of SocIal Psychology, 30, 21-35. Miller, P.e., Lefcourt, H.M., Holmes. 1.G., Ware, E.E., & Saleh, W.E. (1986). Marital locus of control

and marital problem solving. Journal of Personalrty and SocIal Psychology, 51, 161-169 Mullis, R.J. (1992). Measures of economic well-being as predictors of psychological well-being. Social

Indicators Research,26, 119-135. Neugarten, B.L., Havighurst, R.1., & Tobin, S.S (1961). The measurement of life satisfaction. Journal of

Gerontology,16. 134-143. Nieboer, AP. (1997). Life Events and well-bemg: A prospectIVe study on changes m well-bemg of elderly

people due to a serious Illness event or death of the spouse. Amsterdam: Thesis Publishers. O'Brien, W.H., VanEgeren, L., & Mumby, P.B. (1995). Predictmg health behaviors usmg measures of

optimism and perceived risk. Health Values, 19,21-28. O'Malley, P.M., & Backman, 1.G. (1979). Self-esteem and educatIon: Sex and cohort comparisons

among high school students. Journal of Personality and SOCIal Psychology. 37, 1153-1159. Omizo, M. M., Omizo, S. A, & D'Andrea, M. 1. (1992) Promoting wellness among elementary school

children. Journal of Counseling and Development, 71. 194-198 Ormel J. (1983). Neuroticism and weIl-being inventories. Measuring traits or states? PsychologIcal

Medlcme. 13, 165-176. Ormel, J., & Schauf eli, W B. (1991). Stability and change in psychological distress and their relationship

with self-esteem and locus of control: A Dynamic Equilibrium model. Journal of PersonalIty and Socral Psychology, 60, 288-299.

Pall ant, 1., & Cummins, R A (200 I). The comparatrve psychomerrlc performance of two scales deSIgned to measure subjectIve qualIty of life (submitted).

Penland, E.A., Masten, W.G , Zelhart, P., Fournet, G.P., & Callahan, T.A. (2000). Possible selves, depression and coping skills in university students. Personalrty and IndiVIdual Differences. 29, 963-969.

Peterson, e. (2000). The future of optimism. American PsychologIst. 55, 44-56. Phares, E.J. (1978). Locus of control. In H. London & J E Exner (Eds.), DImenSIons of Personalrty.(pp.

263-303). New York: John Wiley & Sons. Pugliesi, K. (1988). Employment characteristics, social support and the well-being of women. Women

and Health, 14, 35-58. Pychyl, T.A, & Little, B.R. (1998) Dimensional specificity in the prediction of subjective well-being'

Personal projects in pursuit of the PHD SOCial IndIcators Research, 45, 423-473. Pyszcynski, T., Greenberg, J., & Holt, K. (1987). Depression, self-focused attention, and expectancies for

positive and negative future life events for self and others. Journal of Personality and SOCIal Psychology, 52. 994-1001.

Radloff, L. (1977) The CES-D Scale' A self-report depression scale for research in a general popUlation. Applted Psychologrcal Measurement, 1.385-401

Ralph, A, Merralls, L., Hart, L., Porter, J.S., & Tan Su-Neo, A (1995). Peer interactions, self-concept, locus of control, and avoidance of social situations of early adolescents Australtan Journal of Psychology, 47, 1I0-1I8.

Rawson, H.E. (1992). The interrelationship of measures of manifest anxiety, self~esteem, locus of control and depression in children with behaVIOur problems Journal of PsychoeducatlOnal Assessment, 1 0, 319-329.

Rigby, K.. McCarron, L., & Rigby, l (1990). Qualtty of life m a nursmg home: A report on the perceptIons of resIdents and staff at Julta Farr Centre. Adelaide: South Australian Institute of Technology.

Roberts, lE., & Gotlib, I.H (1997). Temporal variability in global self-esteem and specific selfevaluation as prospective predictors of emotional distress: SpeCifiCity in predictors and outcome. Journal of Abnormal Psychology. 106(4), 521-529

Roberts, S.B., & Kendler, S. (1999). Neuroticism and self-esteem as indices of the vulnerability to major depression in women~Psychological MedICIne, 29, 1101-1109.

R.A. CUMMINS, E. GULLONE AND A.L.D. LAU

Robins, R.W., Hendin, H.M .• & Trzesniewskl, K.H. (2001) Measuring global self-esteem: Construct validation of a single-item measure and the Rosenberg Self-Esteem Scale. Personality and Social Psychology Bulletm. 27, 151-161.

45

Rosenberg, G.S. (l968). Age, poverty, and isolation from friends in the urban workmg class. Journal of Gerontology. 23. 533-538

Rosenberg. M. (1965). Society and the Adolescent Self-Image. New Jersey: Princeton University Press. Rosenberg, M. (1979).ConcelVlng the self. New York: Basic Books Rotter, J.B (I966). Generalized expectancies for internal versus external control of reinforcement.

Psychological Monographs; 80 (Whole No. 609; pp. 1-28). Rotter, J .B. (1990). Internal versus external control of reinforcement: A case history of a variable

American PsychologISt. 45, 489-493. RubIO, Z. & Peplau, LA. (1975). Who believes in ajust world? Journal-of-Soclal-Issues, 31 (3), 65-89 Russell, J .A., & Carroll, J .M. (1999). On the bipolarity of positive and negative affect. PsycholOgical

Bulletm. 125, 3-30. Ryan, R.M , & Deci, E L. (2000). Self-determination theory and the facilitation of intrinsic motivation,

SOCial development, and well-being. American Psychologist. 55. 68-79. Ryff, C D. (I99I). Possible selves in adulthood and old age' A tale of shifting horizons. Psychology and

Agmg, 6. 286-295 Ryff, C.D., & Keyes, C.L.M. (1995). The structure of psychological well-being revisited. Journal of

Personality and Social Psychology, 69(4). 719-727. Sandvik, L, Erikssen, J., Thaulow, E, Erikssen, G., Mundal, R., & Rodahl, K. (1993}.Physical fitness as

a predictor of mortality among healthy, middle-aged Norwegian men. New England Journal of Medlcme, 328, 533-537

Sanna, L.J. (1996) Defensive Pessimism, optimism and simulating alternatIVes: Some ups and downs of prefactual and counterfactual thinking. Journal of Personality and Soc/al Psychology, 71 (5), 1020-1036

Schalock, R L.. Bonham, G.S., & Marchamd, C.B. (2000). Consumer based quality of life assessment: A path model of perceived satisfaction. EvaluatIOn and Program Plannmg, 23. 75-85.

Scheier, M. F., & Carver, C. S. (1985). Optimism, coping, and health: Assessment and implications of generalized outcome expectancies. Health Psychology, 4, 219-247.

Shmotkin, D. (1991). The structure of the Life Satisfaction Index A 10 elderly Israeli adults. International Journal of Agmg and Human Development. 32, 131-150.

Sherwood, J.J. (I965). Self identity and referent others. SocIOmetry. 28, 66-81 Simons, A. D., Angell, K. L., Monroe, S. M., & Thase, M. E (1993). Cognition and life stress in

depression: Cognitive factors and the definition, rating, and generation of negative life events. Journal of Abnormal Psychology, 102, 584-591.

Sloper, P., Knussen, c., Turner, S., & Cunningham, C. (1991). Factors related to stress and satisfaction with life in famihes of children with Down's Syndrome. Journal of Child Psychology and Psychiatry. 32. 655-676.

Staudinger, U.M., Baltes, P.B., & Fleeson, W. (1999) Predictors of Subjective Physical health and global well-being: Similarities and differences between the United States and Germany. Journal of Personality and SOcial Psychology, 76. 305-319.

Suh, E., & Diener, E (1996). Events and subjective well-being: Only recent events matter. Journal of Personality and SOCial Psychology, 70. 1091-1102.

Suh, E, Diener, E., & Fujita, F. (1996). Events and subjective well-being: Only recent events matter. Journal of Personality and SOCial Psychology, 70, 1091-1102

Switzky, H N. (1999). Intnnsic Motivation and Motivational Self-System Processes in Persons with Mental Retardation A Theory of Motivational Orientation. In E Zigler & D.Bennet-Gates (Eds). Personality Development in IndiViduals with Mental RetardatIOn, (pp.70-106). Cambridge. Cambridge University Press.

Tiger, L (1979). Opllmlsm: The bIOlogy of hope. New York: Simon and Schuster Tokar, 0 M , & Mezydlo Subich, L. (1997) Relative contributions to congruence and personality

dimensions tojob satisfaction. Journal of VocatIOnal Behavior. 50. 482-491. Vaillant, C. 0., & Vaillant, G. E. (1993) Is the U-curve of marital satisfaction an illusion? A 40-year

study of marriage. Journal of Mamage and the Family, 55, 230-239. Watson, D., & Clark, L.A. (1984) Negative affectivity. The disposition to experience aversive emotional

states PsycholOgical Bulletm, 96, 465-490.


Watson, D, & Tellegen, A (1985). Toward a consensual structure of mood. PsychologIcal Bulletm, 98, 219-235.

Watson, D., Clark, LA, & Tellegen, A (1988) Development and validation of brief measures of positive and negative affect The PANAS scales. Journal oj Personality and Social Psychology, 54, 1063-1070.

Watson, D., & Pennebaker, J.W. (1989). Health complaints, stress, and distress: Exploring the central role of negative affectivity. PsychologIcal Review, 96, 234-254

Watson, D., & Clark, L.A (1992). On traits and temperament: General and specific factors of emotional experience and their relation to the five-factor model. Journal oj Personality, 60, 441-476.

Watson, D., Clark, L.A., Weber, K., Assenhelmer, 1.S., Strauss, M.E., & McCormick, R.A (1995). Testing a tripartite model. I. Evaluating the convergent and discriminant validity of anxiety and depression symptom scales. Journal oj Abnormal Psychology, 104,3-14

Wilson, K., & Gullone, E. (1999). The relationship between personality and affect over the life span. Personality and IndIVIdual Differences, 27,1141-1156.

Wolinsky, F. D., Coe, R. M., Miller, D. K., & Prendergast, 1. M (1985). Correlates of change in subjective well-being among the elderly. Journal of Community Health, 10,93-\07.

Zautra, A, & Goodhart, D. (1979). Quality of Life indicators: A review of the literature. CommunIty Mental Health ReVIew, 4, 1-10.

Zautra, A, & Reich, 1. (1980). Positive life events and reports of well-being: Some useful distinctions. American Journal ojCommunity Psychology, 8, 657-670.

PAUL BRAMSTON

SUBJECTIVE QUALITY OF LIFE: THE AFFECTIVE DIMENSION

Abstract Subjective quality of life has attracted Increasing research interest over recent decades but more often than not authors complain of a lack of clarity and consensus over definitions, terminology and theoretIcal models. Such disagreement is slowing progress and development in the field Spurred on by the lack of consensus, this chapter firstly reviews the charactenstics of subjective quality of life on which there is some agreement. This is followed by examination of the role that cognitions and emotion are considered to play in subjective quality of life judgements. In this regard, there is eVidence that cognitions playa major role in most Judgements. LikeWise, emotions appear to play an integral part, but this role has yet to be adequately explored in the lIterature. In order to address this deficit, definitions of emotion and mood are reviewed, followed by discussion of the possible role of emotion in subjective quahty of life judgements. It is proposed that emotion and mood, along with cognitIOns and personality, help to shape perceptions of quality oflife.

Positive psychology is enjoying an unprecedented resurgence in both the applied and research domains of psychology (see Lyubomirsky, 2001). Not long ago the number of published articles in psychological journals that focussed on negative aspects of life outnumbered those published on positive aspects by as many as 17: I (Myers & Diener, 1995). More recently, there has been a notable change, with a growing impatience amongst researchers to understand the positive aspects of life and associated positive emotions such as creativity, happiness and fulfilment. Indicative of the heightened interest in positivistic psychological research is a new model of positive emotions proposed by Frederickson (2001) which will be reviewed in more detail in a later section. Clearly we are experiencing a movement toward positive psychology and positive emotions and one aspect of this is quality of life.

QUALITY OF LIFE

Worldwide, researchers and practitioners are increasingly incorporating quality of life into their work as an index of positive affective state. The breadth of interest in quality of life is evidenced by its application in areas as wide-ranging as economics and politics, community development, and health.

Most researchers in the field have come to accept that quality of life is multidimensional, comprising a number of domains that people weight differently according to how important each is in their life. The number of life domains that best comprise quality of life remains a matter of some conjecture although support for a set of seven core domains is increasing (see Best, Cummins, & Lo, 2000; Brarnston & Tomasevic, 2001).

47

E. Gullone and R.A. Cummins (eds.), The Universality oj Subjective Wellbeing Indicators, 47-62. © 2002 Kluwer Academic Publishers.

48 SUBJECTIVE QUALITY OF LIFE

Working definitions of quality of life vary somewhat across researchers and there is no fixed description nor widely accepted theoretical model. However, Cummins and Cahill (2000) reviewed the literature and noted several areas of agreement amongst researchers. These include:

Quality of life is multidimensional in nature and encompasses a number of life domains. Comprehensive definitions of subjective quality oflife incorporate both subjective and objective dimensions. Within the subjective domain, people attribute varying levels of importance to each of the domains comprising quality oflife. Quality of life is culturally defined, particularly in its objective norms. The construct of quality of life must be equally applicable to all people in whatever circumstances.

There is increasing agreement that if quality of life assessment is to embrace the totality of life it needs to incorporate each of the objective and subjective dimensions. Many researchers now believe that to help us develop a greater understanding of this construct we need to investigate the subjective and objective aspects separately. In this regard, it is important to note that the current chapter will focus only on the subjective aspects of quality of life.

CURRENT STATE OF KNOWLEDGE ON SUBJECTIVE QUALITY OF LIFE

Subjective quality of life is widely thought to consist of individuals' evaluations of their life, a process that includes emotional responses, domain satisfactions and global perceptions of satisfaction (Diener, Suh, Lucas & Smith, 1999). According to Cummins and Cahill (2000), it reflects an assessment of affect aggregated across a number of life domains and/or a single gestalt response regarding life as a whole. Individuals evaluate their circumstances differently depending on their expectations, values, and previous experiences, and research into subjective quality of life assigns importance to these differences. The reliance on subjective evaluations rather than objective, external indicators captures the sense of meaning and personal satisfaction perceived by the individual.

Several researchers have noted the tendency of people to rate their subjective quality of life to be relatively high most of the time (Andrews, 1991; Cummins, 2000; Diener & Diener, 1996). These researchers have all reported that, when large Western population samples from all age groups, ethnic groups and socio-economic groups are asked to rate their life satisfaction, their responses consistently fall well above neutral.

In addition to subjective quality of life scores being high, research has shown such scores to be remarkably stable. Magnus, Diener, Fujita, and Pavot (1993) found that self-reported subjective well-being scores correlated .60 over a four-year interval. Costa and McCrae (1988) found high levels of stability in subjective quality of life even when the measures came from different sources (e.g., spouse versus self). Individuals' subjective quality of life reported by family members

P. BRAMSTON 49

correlated .70 with ratings by individuals' friends once the scores were corrected for the unreliability of the measures (Sandvik, Diener & Seidlitz, 1993). Individuals' level of well-being at work was reported to correlate .70 with the mean level of pleasant affect in recreational settings and a similar correlation was found between social and solitary situations (Diener & Larsen, 1984). Clearly, there is considerable stability in the subjective quality of life scores reported in the literature and scores can be expected to remain relatively constant without rapid fluctuations.

Research supporting the concept that people seem to have a set level for their own subjective well-being was published by Headey and Wearing (1989). In a longitudinal study they found that, while negative life events tended to lower subjective well-being, the influence was usually brief and people returned to their resting well-being level over time. The same phenomenon was reported by Bramston and Cummins {I 998) when measuring subjective stress.

In order to understand how subjective quality of life scores stay relatively stable and why they return to some predetermined resting level after they do change, Cummins and colleagues have proposed a process of homeostasis. According to Cummins (2000), the apparent stability most likely involves some form of homeostatic control such that perceptions of life quality may rise or drop according to circumstance but these return to a predetermined level through homeostatic strategies, some conscious and others unconscious. There are several cognitive devices people are thought to use to maintain their equilibrium and these are explained in some detail both in the chapter by Helmes et al. and the chapter by Cummins et al. in this text. The proposed purpose of such cognitive homeostasis is to keep people from feeling overly positive or overly negative for lengthy periods and thus to avoid clinical states such as mania and depression. Many people therefore have the ability to cognitively maintain normal and adaptive levels of subjective well-being in the face ofa wide range of environmental conditions.

Further support for the role of cognition in subjective quality of life judgements can be found in the work of Andrews and Robinson (l991). For example, people use mental strategies to compare themselves to others, to compare their past with their present and to compare their reality with their ideal. It seems likely that future research into such cognitive strategies will continue to grow and to steadily reveal the mechanisms underlying these processes.

In summary, subjective quality of life is becoming an increasingly popular research variable and clear trends regarding this construct are emerging from the literature. Research has consistently documented that people self-report their subjective quality of life at levels well above the neutral scale point and levels of subjective quality of life appear to remain relatively constant over time. Moreover, similar levels of well-being are reported across people and nations. The stability in self-reported quality of life is likely the effect of a homeostatic cognitive mechanism that works to keep it within an adaptive range.

Although the importance of cognition in subjective quality of life is widely acknowledged, there is much less known about the role of emotions or affect in subjective quality of life judgements. Is subjective quality of life an emotion or do emotions shape feelings of life satisfaction? These and other similar questions will be explored in the following section.


SUBJECTIVE QUALITY OF LIFE AND EMOTIONS

Research on emotions and subjective quality of life is surprisingly rare and there are many important questions regarding their interrelationship that remain unexplored. For example, it may be that quality of life is a feeling or emotion, something akin to happiness. If so then could it be a single emotion across a range of dimensions of life or rather a collection of emotions subsumed under a single rubric? The remainder of this chapter explores definitions and models of emotion in an attempt to tease out answers to these puzzles.

Some early evidence of how subjective quality of life relates to emotions was presented by Andrews and Withey (1976) who found that they could be differentiated from each other. Reported life satisfaction emerged from their analyses as a separate factor, the others being positive and negative affect. This finding was later supported by Lucas, Diener and Suh (1996) who demonstrated that pleasant affect, unpleasant affect and life satisfaction were statistically separable constructs. These findings suggest that emotions and subjective quality of life are closely related but not the same. Therefore it seems unlikely that subjective quality of life is simply another emotion to be added to the already substantial list.

A noticeable difference between emotions and subjective quality of life is variability. As noted above, subjective quality of life has demonstrated a remarkable stability over time. In contrast, people's emotions recorded on two mornings two weeks apart correlated only slightly with each other (Green, Goldman, & Salovey, 1993). People's pleasant emotions assessed at random times, in various situations correlated an average of only .10 according to Diener and Larsen (1984). Emotions are typically short-lasting responses (Frijda, 1999). Clearly, emotions can fluctuate a lot, more than is typically seen in subjective quality of life scores, and so it seems fair to conclude that while emotions and subjective quality of life may be closely related, they differ quite markedly in their stability.

EMOTIONS AS BUILDING BLOCKS

According to Diener and colleagues, emotions can be markers or signs of wellbeing. Perhaps then, a high subjective quality of life cannot be experienced without the presence of positive emotions. If so, then are positive emotions a necessary and sufficient foundation for perceptions of high quality of life?

The new broaden-and-build model of positive emotions by Frederickson (2001) may suggest an answer to the above question. In her model, the experience of positive emotions broadens people's thought-action repertoires and expands their physical, social and intellectual resources. That is, feeling positive opens a person up to more thoughts, actions and options that in turn enrich most aspects of their life. Perhaps these expanded repertoires and resources can be thought of as part of good subjective quality of life, thus suggesting that emotions may contribute to and help determine an end state called subjective quality of life. According to Frederickson, the thought-action repertoire associated with negative emotions is said to be

P. BRAMSTON 51

narrower and incompatible with the broadening action associated with the positive emotions. If the, as yet untested, Frederickson model is valid, emotions could then be seen as the building blocks of subjective quality of life.

In support of this view, positive emotions have been shown to do more than signal feeling good. They can also help motivate individuals to achieve things in their life. For example positive emotions can lead to higher creativity (Estrada, Isen, & Young, 1994), improved performance, and better earning capacity (Diener, Nickerson, Lucas, & Sandvick, 2000). According to Lyubomirsky (2001), learning how to cultivate positive emotions can promote health and psychological well-being. Thus positive emotions have been conceptualised as, not only an outcome state, but also as precipitators. They signal flourishing, yet also produce flourishing (Frederickson, 200 I). In addition, positive emotions can do this over an extended period, not just in brief spurts (Lucas, Diener & Larsen, 2001). Clearly positive emotions are to be sought after, not just as end states or objectives but also as a vehicle to achieve personal growth and seemingly high subjective quality of life.

One of the problems of discussing the relationship between subjective quality of life and positive emotions is the inconsistent use of terminology. For example, happiness (thought to be one of the "basic" emotions) is used interchangeably with well-being (widely accepted as synonymous with subjective quality of life) by Lyubomirsky (2001). To explore this further it is helpful to review what emotions are thought to be, how they are defined and how readily that definition fits with subjective quality of life.

WHAT ARE EMOTIONS?

Unfortunately, definitions of emotion vary as widely as definitions of quality of life. It seems that some researchers understand emotions in categories, others in dimensions, some advocate a unipolar concept and others bipolar, some accept a simple structure, some a circumplex model, and some a hierarchy. No one structure or model of emotion has yet done justice to this heterogeneous concept (Lucas, Diener, & Larsen, 2001).

One theory of emotion asserts that emotion is best studied in terms of several basic emotions. For example, the emotions of happiness, surprise, fear, anger, sadness, disgust, and interest are promoted as a set of basic emotions around which others can be clustered (Ekman, Levensen, & Friesen, 1983). Other researchers have considered it important to separate basic or primary emotions from secondary emotions, with the basic emotions postulated to be innate and universal. Among the advocates of the basic emotions approach, there is some disagreement about how to define a basic emotion and which emotions best meet these requirements (Ekman, 1992). Interestingly, feelings of "general pleasantness" are included in the primary list and they seem, at face value at least, to be quite akin to emotions intricately related to subjective quality of life.

Another attempt to categorise emotions has been made on the basis of dimensions or relations between the emotions. According to the dimensional approach, statistical techniques can be used to group variables according to the pattern of their inter-relationships. Data supporting a three dimensional model are


presented by Schimmack and Grob (2000), however, support is much more widespread for the two dimensional model. At the basis of most two factor theoretical models of emotion is the description of two independent dimensions- the degree of pleasantness and the degree of activation. The pleasure dimension is described as pan-cultural and has been documented as important by philosophers for centuries while the activation dimension describes a sense of mobilization or energy, arousal, or tension that can range from sleep to frenetic excitement. It has a diurnal rhythm and is drug reactant. According to this theory, both valence and activation interpretations are required to understand emotion. One of the two factor models in particular has attracted considerable support- the circumplex model. Detailed reviews of this circumplex model can be found in the work by Larsen and Diener (1992) and Watson et aI., (1999) and are therefore not included here.

Models of subjective quality of life generally focus on the pleasantness/unpleasantness dimension but not many quality of life measures tap motivation or activation. One exception is the ComQol measure by Cummins, McCabe, Gullone and Romeo (1994). Cummins has maintained for many years that subjective quality of life scales should not only measure life satisfaction but should also incorporate a loading for importance. It is proposed that activation/deactivation and importance/non-importance might be similar dimensions and future research into an activation dimension of subjective quality of life could prove fruitful.

A recently proposed model of emotion that is consistent with the two factor and circumplex models is that of Russell and Feldman Barrett (1999). These authors tried to find a categorisation structure that could incorporate some of the other theories and their proposed solution described emotion as having two components, core affect and prototypical emotional episodes. They proposed that most events that fall under the term emotion can be categorised as one of these. Each of the categories in this model will be briefly described as the next step in exploring how and where subjective quality of life and emotions interrelate.

PROTOTYPICAL EMOTIONAL EPISODES

Prototypical emotional episodes are thought to be a complex set of events concerned with an object, event, or person that involve intense core affect, overt behaviour, cognitive processes, and bodily events that accompany that emotion. Prototypical emotional episodes involve all of these dimensions and their intensity makes them quite rare according to Russell and Feldman Barrett (1999).

Emotional episodes have been categorised or grouped according to similarity by these authors who suggest there are 500-2000 categories of prototypical emotional episodes in the English language. The idea of grouping emotions into categories has long held some appeal to researchers but little consensus has emerged over the years.

P. BRAMSTON 53

SUBJECTIVE QUALITY OF LIFE AND INTENSE EMOTIONAL EPISODES

Subjective quality of life does not appear to share much in common with intense emotional episodes associated with a recent event. The reported stability of subjective quality of life suggests that it does not rapidly fluctuate in association with life events, if anything, ratings of subjective quality of life, for the most part, seem to be immune to emotional episodes. This is evidenced by distressed people rating their quality of life reasonably high and suggests that people do not necessarily weight their current emotional state very strongly in their quality of life ratings. As discussed by Cummins et aI., (current volume) in their chapter, buffering mechanisms may act to keep subjective quality of life within a pre-set homeostatic range. It is only when negative events exceed buffering resources that subjective quality of life may become compromised.

The second component of the emotions model proposed by Russell and Feldman Barrett (1999) is core affect. This too will be briefly described with a view to adding to our understanding of subjective quality of life.

CORE AFFECT

Core affect is thought to include most elementary, consciously accessible, affective feelings. It is ever present and not necessarily directed at anything specific (Russell & Feldman Barrett, 1999). Considerable research has reported on core affect, often under terms such as tension, pleasure, depression or elation. Core affect ebbs and flows over time, can be free-floating and non directed (as in moods) or directed when part of an episode. It is subject to many causes (e.g. weather) and varies in intensity over time. A person is thought to always be in some state of core affect, even if neutral. Core affect can be mild, in the background, or can be very intense and at the forefront. Examples include waking up in a particular mood, feeling contented after listening to music, exhausted and stressed after a hard day, or calm when lying on the beach.

SUBJECTIVE QUALITY OF LIFE AND CORE AFFECT

Core affect, as described above, appears quite consistent with definitions of subjective life quality. The latter is widely regarded as part of a relatively stable affective state, non-directed, something that does not fluctuate greatly over time, much as core affect is thought to do. It therefore seems reasonable to conclude that this core aspect of emotions has considerable overlap with subjective quality of life.

Having distinguished some similarities and some differences between subjective quality of life and components of the theoretical model of emotion described by Russell and Feldman Barrett (1999), it is useful to search for similar areas of overlap and difference between subjective quality of life and mood, given that mood is not included in the Russell et al. model. Not all writers distinguish between emotion and mood but many do, generally on the basis of intensity and duration (see Terry, 2001 for a review).


MOOD

While emotions are frequently regarded as part of a reaction that is usually brief, spasmodic and sometimes intense, with an identifiable cause, descriptions of mood generally refer to more subtle, longer lasting, less intense, more in the background feelings that cast a positive or negative light over things (Bower, 1981). While emotions are assessed by asking how one feels right now, when that is extended over time one is assessing a mood (Russell & Feldman Barrett, 1999). According to Frijda, Kuipers and Schure (1989), moods, unlike emotions, refer to one's relationship to life or to the world in general and not to a special object. In other words, moods are appraisals of the person's world as a whole. On a similar theme, Lazarus (1991) also proposed that moods refer to the larger, pervasive, existential issues in life. He suggested that moods are a diffuse reaction that refers to abstract and long-lasting life contexts. Mood biases the individual for readiness to react in certain ways, it provides a guiding framework in which some emotions become more likely to occur. It is an emotional tone (Dreikurs Ferguson, 2000).

In summary, moods clearly differ in time and intensity from emotional episodes. Emotional fluctuations, averaged over several weeks or months seem to be labelled moods and represent the person's mean level of emotion. Individuals are thought to differ widely from one another in these levels. As suggested by Diener and Lucas (2000), there is stability in the average emotional life of people that transcends the momentary fluctuations of emotions and this is known as mood. Core affect as described by Russell and Feldman Barrett (1999), on the other hand, seems to have more in common with mood than it does with intense emotional episodes. These authors propose that core affect might be present in mood and present in prototypical emotional episodes as a sort of underlying personality dimension, essentially an ever-present individual difference factor. These distinctions are by no means clear and await further exploration and research.

SUBJECTIVE QUALITY OF LIFE, MOOD AND EMOTIONS

Clearly mood has much in common with subjective quality of life. Both are reasonably long lasting, relatively stable constructs that relate to life or the world in general rather than to specific events. Recent research by Eid and Diener (200 I) examined the distinction between well-being and mood. They measured both variables in a sample of people across multiple occasions, and found that while the two constructs were correlated, mood did not greatly influence ratings of well-being. Interestingly, the stability of subjective quality of life appeared to transcend even the relatively constant background nature of mood.

In summary, this chapter has compared subjective quality of life to selected models of emotion and mood and found that it shares many of the characteristics of both core affect and mood. It appears that subjective quality of life does not tap the fluctuating, intense emotional episodes described by Russell and Feldman Barrett (1999) and there is a suggestion that people may ignore their current episodic high or low feelings when rating subjective quality of life. The remaining sections of this chapter discuss these issues.

P. BRAMSTON 55

DISCUSSION

The broaden-and-build model of emotions proposed by Fredrickson (2001) suggests an upward spiral where the experience of positive emotions leads to the broadening of cognitive and behavioural repertoires and the building of personal resources, which in tum predisposes people to experience more positive emotions, leading to further broadening. As the upward spiral builds, it seems likely that more positive perceptions of life are an inevitable outcome. "As this cycle continues, people build their psychological resilience and enhance their emotional well-being." (Fredrickson, 200 I; p. 223). Thus the experience of positive emotions over time is predicted to lead to appreciable increases in subjective quality of life.

The "broaden and build" theory of positive emotions awaits empirical scrutiny. For example, there currently is little published evidence that positive emotions lead to a measurable broadening of cognitions or behavioural repertoires. Nevertheless, publication of the theory is an exciting landmark that has raised the profile of positive psychology and may inspire renewed research into positive emotions and subjective quality of life. Perhaps research into this model over the next few years may empirically demonstrate "upward spirals" that lead to broadening of resources. Given the close relationship between SUbjective quality of life and aspects of emotions and mood, such research may help us better understand why people, on average, rate their quality of life so high and why levels of subjective quality of life remain so stable.

CHARACTERISTICS OF SUBJECTIVE QUALITY OF LIFE

Several problems in the subjective quality of life literature have been traced to the fact that empirical work has seldom been theory-driven or guided (Feist, Bodner, Jacobs, Miles & Tan, 1995). In addition, theories have frequently remained separate and distinct rather than being integrated into more synthetic conceptualisations. While some of the categories of well-being theories have been summarised in a review by Diener (1984), there is little consensus in the literature regarding the structure and characteristics of well-being and subjective quality oflife.

This discussion sets out some of the characteristics of SUbjective quality of life in an attempt to link it to the theories of emotion. The literature on emotions is used to inform the discussion as subjective quality of life is compared to emotions and mood. The table that follows is presented as an approximate conceptual guide to help the reader appreciate some of the features of subjective quality of life that relate to current theories of emotion. It must be noted that Table I is not a theoretical model, it is simply a representation of some possible dimensions or characteristics of subjective quality of life. The left-hand column lists a collection of features that is often found in discussions of emotion. These features are used here to help compare and contrast emotional episodes, mood and SUbjective quality of life.


Table 1. Comparison of Mood. Emotion & Subjective Quality of Life on Characteristics Commonly Mentioned in the Literature.

Characteristics Emotional Episodes Mood & Core Affect Subjective quality of life

Duration Short Longer Lengthy (seconds/minutes) (hours/days) (years)

Speed of onset Very Slow Probably very fast/immediate slow

Intensity Great Lesser, background Very low Visibility Distinctive facial Inferred from general Difficult to

signals signals perceive Physiology Distinctive patterns No distinctive patterns No distinctive

ofCNS activity patterns Function Distinctive General adaptive role General

functions e.g. adaptive role fight/flight

Antecedents Often identifiable Lesser specificity Little known Consequences Specific behaviours General patterns of Possible

e.g. fight/flight behaviour/cognition homeostatic process

Variability Fluctuates sharply Less variable Stable over time

Note: ThIs table IS an adaptatIon and extensIOn of charactenstlcs of mood and emotIon pubhshed by Terry (2001)

Duration refers to the length of time the characteristic is exhibited with emotions thought to be short lasting, brief and spasmodic (Frijda, 1999) and normally considered a state rather than a trait (Fiedler & Forgas, 1987). Mood is often thought of in terms of hours or days (Morris, 1999) or sometimes as always present, if only in a neutral state (Diener & Lucas, 2000).

Speed of onset of an emotion can be very quick after a stimulus event (Fiedler & Forgas, 1987) but, in the absence of research on onset times of moods and subjective quality of life, it is assumed to be slower.

Intensity refers to the degree of arousal with emotions sometimes intense (Diener, Larsen, Levine, & Emmons, 1985; Russell & Feldman Barrett, 1999). Mood can be mild or intense if it is part of an episode but is usually accepted as being less intense (Fiedler & Forgas, 1987). Subjective quality of life is not normally discussed as being intense despite reports of high satisfaction ratings in published research (Bramston, Pretty, & Chipuer, in press).

Visibility refers to obvious signs of the characteristic. For example, Ekman (1992) has reported on facial expressions and movements associated with emotions. Moods and subjective quality of life may be ever present and generally have not been linked to visible signs, although clinically there could be obvious indicators of mood disturbances as in the depressed state.

P. BRAMSTON 57

Physiology deals with physiological and autonomic nervous system action in the body. Adrenalin induced arousal as a cue for emotion has been reported on by Schachter (1967) and Levenson (1992) has investigated emotions and arousal of the autonomic nervous system. Research into physiological aspects of moods can be found throughout the clinical literature but seems not to have extended to subjective quality of life.

Functions refers to a purpose and an obvious example is the longstanding link between emotions and motivation. Positive emotions are thought to motivate people and lead to higher creativity (Fredrickson, 200 I). One function of mood is thought to be preparing the scene for emotions (Fiedler & Forgas, 1987) while a function of subjective quality of life might be as an indicator of balance lim balance in some sort of homeostatic mechanism (Cummins, 2000).

Consequences of emotional episodes, moods and perceptions of well-being have not been directly contrasted in the research literature but it seems likely that the three constructs differ on this dimension. For example, positive emotions and moods are thought to advance physical and psychological health (Lyubomirsky, 2001) while the consequences of high or low subjective quality of life (if not extreme) are thought to probably result in adjustments to restore balance (Cummins, 2000).

Variability has been shown in previous sections to be high in emotions (Diener & Larsen (1984), less so in moods (Terry, 2001) and very little in subjective quality oflife (Eid & Diener, 2001).

Clearly, some of the above characteristics are generalisations based on trends in the literature that have not yet been substantiated by empirical research. Further research and discussion of the characteristics shown in Table I is encouraged to add increasing clarity to our understanding of subjective quality of life.

FACTORS IMPACTING ON SUBJECTIVE QUALITY OF LIFE

Many researchers have recommended investigations into the factors impacting on subjective quality of life as fundamental to the future development of knowledge in the field. We will never sharpen our definitions and clarify our models until we identify and understand the variables that significantly determine subjective quality of life judgements. One response to this call for investigation of determinants is a theoretical model proposed by Cummins and Cahill (2001) who presented a model consisting of three levels of determinants of subjective quality of life.

The fIrst order determinants are thought by Cummins and colleagues to be traitlike, individual, personality factors; the second order determinants are described as conscious cognitive schemata involving such variables as control, self-esteem and optimism and the third order determinants are suggested to be experiences of the external world like social support that influence the second order determinants. Cummins has encouraged researchers to test the validity of the proposed model and to investigate the inter-relationships between the three levels of determinants.

By using the existing research on positive emotions, the three stages proposed by Cummins and Cahill (2001) will be explored and discussed.


1. Personality factors

These are proposed as the first level of quality of life determinants in this model. Support for the suggestion that personality or inherited, genetic temperament is one of the strongest influences on emotions and emotional well-being can be found in DeNeve and Cooper (1998). Diener, Suh, Lucas and Smith (1999) provide an extensive review of this topic and conclude that the happy individual is extraverted, optimistic and worry-free. However, identifying causal mechanisms that determine how personality impacts on emotions remains unclear. Diener et al. (1999) propose three processes to explain this relationship- affective predispositions, rewarding behaviours and person-environment fit, and they present evidence that partly supports all three approaches.

2. Cognitive factors

These are second level determinants of subjective quality of life in the Cummins model where feelings are thought to be cognitively evaluated across aspects of life to form an affective response that we label quality of life. Some support for this can be found in the emotion literature. For example, self-reflection and evaluation (cognitive processes) have been shown to influence subjective quality of life through positive thinking (McCrae & Costa, 1986), anticipating positive futures (Scheier & Carver, 1993) and feeling in control (Bandura, 1997). Simply put, happy people perceive, evaluate and think about things within a more positive framework than do unhappy ones (Lyubomirsky, 200 I).

The cognitive-judgemental aspect of subjective quality of life evaluations is an example of the top-down model of well-being discussed by Diener (1984). Prior to this, bottom-up situational influences were popular, an approach that is built on the meeting of basic universal human needs to achieve life satisfaction. Yet the relatively small effect sizes of external, objective influences such as wealth and health on well-being led researchers to look to a more top-down approach that included variables such as personality, discrepancy theory, goals, adaptation and coping. Most contemporary researchers now emphasise a top-down approach which posits that thought and motivation direct or at least moderate emotional responses made to external stimuli (Dreikurs Ferguson, 2000). While objective factors exert considerable influence on subjective quality of life, it is suggested that they do so through cognitive and motivational processes. Thus both top-down and bottom-up models of emotion can be seen interactively to work in the prediction of subjective well-being (Feist, Bodner, Jacobs, Miles & Tan, 1995).

A related theory is presented by Lyubomirsky (2001) in an attempt to understand why some people are happier than others. She suggested that researchers should look at cognitions and motivations in what she terms a construal theory. The cognitive processes that she identifies as important include social comparison, dissonance reduction, self-reflection, and coping. Several of these appear to overlap with the third level determinants of the Cummins and Cahill (2001) model.

P. BRAMSTON 59

This review of the literature on emotions also raises questions regarding the role of cognitive goal setting, motivation and coping skills. Theories on the determinants of sUbjective quality of life have yet to comprehensively investigate these three variables. Perceptions of goal-related progress in life and motivation has been shown to impact on ratings of life quality by Lyubomirsky (200 I) and coping skills have been shown to be related to levels of well-being (Diener, Suh, Lucas, & Smith, 1999). Coping strategies that seem to be positively related to subjective quality of life include the cognitive restructuring strategies of positive illusions (Scheier & Carver, 1993) and those that focus upon deriving positive aspects from negative events (Folkman, 1997).

3. Socio-environmental Factors

In the Cummins and Cahill model, these factors are considered as third level operants that moderate the cognitions on the second level. Support for this group of factors can also be found in the study of emotions. Social comparison or discrepancy theories suggest that people make a number of comparisons with others and the results of these comparisons influence their feelings and judgements. Judgement of how satisfied people are with their present state of affairs is based on a comparison with a standard which each individual sets, which balances on the person's own judgement and not on any externally imposed criterion. Research suggests that comparisons are typically with peers (Diener & Fujita, 1997), previous experiences (Tversky & Griffin, 1991) and with desires (Carver & Scheier, 1998). Lyubomirsky (200 I) notes that her findings show that happy people are less sensitive to social comparison and use it more sparingly and selectively than unhappy people.

CONCLUSIONS

This selective review ofthe research on emotions has demonstrated how the study of subjective quality of life can be enriched by acknowledging its links with emotions and mood. Clearly quality of life is a complex construct with mUltiple levels of psychological determinants influenced by factors such as personality, cognitive factors and socio-environmental moderators. This review has offered support for the three-level determinants model proposed by Cummins and Cahill (2000).

This review has also highlighted the failure of subjective quality of life ratings to reflect intense emotional episodes. Diener, Sandvik and Pavot (1991) similarly noted that judgements of well-being are not strongly determined by intense affect. These authors argued that intense positive emotions are less important to long term wellbeing because they are rare and often associated with or followed by intense negative episodes (Larsen & Diener, 1987). However, they offered no supportive evidence for this rationale and it is suggested that future researchers investigate this issue which is worthy of closer scrutiny. It is important that researchers be aware of this limitation in subjective quality of life measures and avoid using it to detect fluctuating emotional change.

In conclusion, there has been increasing research interest in the role of cognitions in subjective quality of life judgements (see the chapter by Cummins et al. in this


text) but the role of emotions in such life judgements has only very recently attracted the interest of researchers (Diener & Lucas, 2000). From the preceding discussion, it is clear that subjective quality of life differs substantially from both emotional episodes and moods on a number of characteristics. Evidence has also been cited that subjective quality of life differs from negative and positive affect. However, subjective quality of life still has so much in common with emotions that Diener and Lucas (2000) concluded that well-being (often equated with subjective quality of life) probably did fit within the circumplex model of emotions, somewhere towards the end of the pleasantness dimension. It is also possible that emotions could be the building blocks of subjective quality of life and that they may actually be a necessary but not sufficient component of subjective quality of life. Clearly there are many questions that remain unanswered and it is necessary to acknowledge that the answers will require sophisticated examination (both empirical and conceptual) of the many models and definitions of emotions and moods that characterise publications in this field.

Address for Correspondence: Dr Paul Bramston, Dept. of Psychology, University of Southern Queensland, Toowoomba, Australia, 4350. Email: [email protected]

REFERENCES

Andrews, F.M. (1991). Stability and change in levels and structure of subjective well-being. USA 1972 and 1988. SOCial Indicators Research, 25, 1-30

Andrews, F.M .• & Robinson, J.P. (1991). Measures of subjective well-being. In J.P. Robinson & L S. Wrightsman (Eds.), Measures of personality and SOCial psychological attitudes (pp 61-114). New York: Academic Press

Andrews, F M , & Withey, S.B. (1976) SOCial mdlcators of well-be mg' America's perceptIOn of life quality. New York: Plenum Press

Bandura, A. (1997). Self efficacy: The exerCise of control. New York: Freeman. Best, C.J., Cummins, R.A., & Lo, SK (2000). The quality of rural and metropolitan life. Australian

Journal of Psychology, 52,69-74. Bower, G.H. (1981). Mood and memory American PsychologISt, 36, 129-148. Bramston, P., & Cummins, R.A. (1998). Stress and the move Into community accommodation. Journal of

Intellectual and Developmental Disability, 23, 295-308. Bramston, P., & Tomasevic, V (2001). Health locus of control, depreSSIOn and quality oflife in people

who are elderly. Australian Journal on Agemg, 20, 192-195. Bramston, P., Pretty, G , & Chipuer, H. Unravelling subjective quality oflife: An Investigation of

individual and community determinants. SOCial Indicators Research (accepted for publication). Carver, C.S, & Scheier, M.F. (1998) On the self-regulation of behavIOr New York: Cambridge

University Press. Costa, PT., & McCrae, R.R. (1988). Personality in adulthood: A six-year longItudinal study of self

reports and spouse ratings on the NEO Personality Inventory. Journal of Personality and SOCial Psychology, 54, 853-863.

Cummins, R A (1999). Normative life satisfaction: Measurement issues and a homeostatic model. In B.lumbo (Ed.), Methodological developments and Issues m quality of life research. Amsterdam Kluwar (in press).

Cummins, RA (2000). Objective and subjective quality of life" An mteractive model SOCial Indicators Research, 52, 55-72.

Cummins, R.A , & Cahill, J. (2000). Progress in understandmg subjectIve quality of life. IntervenclOn PSlsoclal: Revlsta Sobre Igualdad y Calidad De Vida, 9, 185-198

P. BRAMSTON

Cummins, R.A., McCabe. M.P., Gullone, E., & Romeo, Y. (1994) The Comprehensive Quality afLife Scale: Instrument development and psychometric evaluation on tertiary staff and students. EducatIOnal and PsychologIcal Measurement, 54, 372-382.

DeNeve, K.M., & Cooper, H. (1998). The happy personality: A meta-analysis of 137 personalIty traits and subjective well-being. PsychologIcal Bul/etln, 124, 197-229.

Diener, E. (1984). Subjective well-being. PsychologIcal Bul/etm, 95, 542-575. Diener, E., & Diener, C. (1996). Most people are happy. Psychological SCIence, 7, 181-185.

61

Diener, E., Emmons. R.A., Larsen. R.J., & Griffin, S. (1985). The Satisfaction with Life Scale. Journal oJ Personality Assessment, 49, 71-75.

Diener, E., & FujIta, F. (1997). Social comparisons and sUbjective well-bemg. In B.Buunk & R Gibbons (Eds.), Health, coping and socwl comparIson (pp. 329-357). Mahwah, Nj:Erlbaum.

Diener, E .• & Larsen, R J. (1984). Temporal stability and cross-situational consistency of affective, behavioural, and cognitive responses. Journal oj Personality and Social Psychology, 47, 871-883.

DIener, E., & Lucas, R E (2000) Subjective emotional well-being In M.Lewis. & J.M. Haviland-Jones (Eds.), Handbook oj emotIOns (2nd ed., pp 325-337). New York: Guilford Press.

Diener. E , Nickerson, c., Lucas, R.E., & Sandvik, E. (2000). Income and happiness Does subjective well-being predict future income? Paper submitted for publication. University of Illinois.

Diener, E., Sandvik, E, & Pavot, W. (1991). Happiness is the frequency, notthe intensity, of positive versus negative affect. In F.Strack, M.Argyle. N.Schwarz (Eds ), Subjecllve well-being: An interdIscIplinary perspectIve. International series in expeY/mental socwl psychology, Vo121. (pp. 119-139). Oxford: Pergamon Press.

Diener, E., Suh, E.M • Lucas, R.E., & Smith, H. (1999). Subjective well-being: Three decades of progress- 1967-1997. PsychologIcal Bulletin, 125, 276-302.

Dreikurs Ferguson, E. (2000). MotivatIOn: A blOsoclal and cognlllve integratIon oj mol/vatlOn and emotion. New York: Oxford University Press.

Eid. M, & Diener. E (in press). Global judgements of subjective well-being: Situational variability and long-term stability. Annual ReVIew oJQuaizty oj Life Methods.

Ekman. P. (1992). An argument for basic emotions. CogmtlOn and EmotIOns, 6, 169-200. Ekman, P., Levensen, R W , & Fnesen, W.v. (1983). Autonomic nervous system activity distinguishes

between emotions. SCIence, 221, 1208-1220. Estrada, C.A..lsen, A.M., & Young, M.J. (1994). Positive affect improves creative problem solving and

influences reported source of practice satisfactIOn in physicians. Mol/vatlOn and Emollon, 18, 285-299

Feist, G.J., Bodner, T.E., Jacobs, J.F , Miles, M., & Tan, V (1995). Integrating top-down and bottom-up structural models of subjective well-being: A longitudinal investigation. Journal oj Personality and SOCIal Psychology, 68, 138-150.

Fiedler, K. & Forgas, J. (1987). Affect, cogmtlOn and SOCIal behaVIOr. Toronto' Hogrefs. Folkman, S. (1997). PositIve psychological states and coping with severe stress. Socwl SCIence Medicme,

45,1207-1221. Frederickson. B.L. (2001). The role of positive emotions in positive psychology' The broaden-and-build

theory of positive emotions AmeY/can Psychologist, 56, 218-226 Frijda, N.H. (1999). Emotions and hedonic experience. In D.Kahneman, E. Diener, & N.Schwarz (Eds.),

Well-being: The Joundallons oj hedOniC psychology (pp. 190-210). New York' Russell Sage Foundation

Frijda, N.H .• KUipers, p. & Schure, E. (1989) Relations among emotion, appraisal, and emotional action readiness Journal oj Personality and SOCIal Psychology, 57, 212-228.

Green. D.P., Goldman, S.L.. & Salovey, P. (1993). Measurement error masks bipolarity in affect ratings. Journal oj Personality and Socwl Psychology, 64, 1029-1041

Headey, B., & Weanng, A (1989). PersonalIty. lIfe events and subjective well-being. Toward a dynamic equilibrIUm model. Journal oj Personaizty and Socwl Psychology, 57, 731-739.

Larsen, R J., & Diener. E. (1987). Emotional response intenSity as an individual difference characteristic. Journal oJResearch m Personality. 21, 1-39.

Larsen, R.J., Diener, E. (1992). Promises and problems with the circumplex model of emotion. In M S. Clark (Ed.), ReVIew oJpersonallty and socwl psychology: EmotIOn, Vol 13, (pp. 25-59). Newburry Park, CA Sage.

Lazarus, R S. (1991). EmotIOn and adaptatIOn. New York. Oxford University Press. Levenson. R. W. (1992) Autonomic nervous system differences among emotions. PsychologIcal SCIence,

3,23-27


Lucas, R.E., Diener, E., & Larsen, R.J. (2002). Handbook of positIVe psychological assessment. (in press) American Psychological Association.

Lucas, R.E., Diener, E., & Suh, E.M. (1996). Discriminant validity of subjective well-being measures. Journal of Personality and Social Psychology, 71, 616-628.

Lyubomirsky, S. (200 I). Why are some people happier than others? The role of cognitive and motivational processes in well-being. American Psychologist, 56, 239-249.

Lyubomirsky, S., Lepper, H.S , & Giordam, P. (200 I). What are the differences between happmess and self-esteem? Manuscript submitted for publication.

Magnus, K.B, Diener, E., Fujita, F., & Pavot, W. (1993) Extraversion and neuroticism as predictors of objective life events: A longitudinal analYSIS. Journal of Personality and Social Psychology, 65, 316-330.

McCrae, R.R., & Costa, P.T. (1986). Personality, copmg, and copmgeffectiveness in an adult sample. JournalofPersonaltty, 54, 385-405.

Morris, W.N. (1999). The mood system. In D.Kahneman, E. Diener, & N.Schwarz (Eds), Well-being: Thefoundations of hedonic psychology (pp. 169-189). New York: Russell sage Foundation.

Myers, D.G., & Diener, E. (l995). Who is happy? Psychological SCience, 6, 10-19. Redelmeier, D., & Kahneman, D. (1996). Patients' memories of painful medical treatments: Real-time

and retrospective evaluations of two minimally invasive procedures, Pain, 116, 3-8. Russell, JA, & Feldman Barrett, L.F. (1999). Core affect, prototypical emotional episodes, and other

things called emotion: Dissecting the elephant. Journal of Personaltty and Social Psychology, 76, 805-819.

Sandvik, E., Diener, E., & Seidlitz, L. (1993) Subjective well-being: The convergence and stability of self-report and non-self-report measures. Journal of Personaltty, 61, 317-342.

Schachter, S (1967). Cognitive effects on bodily functIOning' Studies of obesity and eating In D.C. Glass (Ed.), Neurophysiology and emotIOn (pp. 117-144) New York: Rockefeller UmveTSIty Press.

Scheier, M.F., & Carver, C.S. (1993). On the power of positive thmking: The benefits of being optimistic. Current DirectIOns m Psychological Science, 2, 26-30.

Schimmack, U , & Grob, A. (2000) Dimensional models of core affect: A quantitative comparison by means of structural equatIOn modeling. European Journal of Personaltty, 14, 325-345.

Terry, P. (2001). Mood and emotions in sport In T.Morris & J.Summers (Eds.), Sport psychology: Theory, appltcatlOns and Issues. Brisbane: Wiley.

Tversky, A., & Griffin, D. (1991) Endowment and contrast in judgments ofwell-bemg In F.Strack & M Argyle (Eds.), Subjective well-bemg' An interdisciplinary perspective (pp. 101-118). Oxford, England: Pergamon Press

Watson, D., & Clark, L.A. (1984). Negative affectivity. The dispOSition to experience negative affective states. Psychological Bulletin, 96, 465-490

Watson, D, Wiese, D., Vaidya, J., & Tellegen, A. (1999). The two general activation systems of affect: Structural findings, evolutionary conSiderations, and psychobiological evidence. Journal of Personaltty and Social Psychology, 76, 820-838

JOUKO KAJANOJA

THEORETICAL BASES FOR THE MEASUREMENT OF QUALITY OF LIFE

Abstract. The ideal of liberahsm defines quality of life as bemg up to each individual In the present chapter, it is argued that this view is not satisfactory. The attributes of a good life need to be defined and it is best to do this conSCiously. A well justified process with which to achieve this is proposed to be on the basis of dialogue and communalism. On the other hand the liberal Ideal is needed to avoid intolerant controls on defining ways of life. Arguments are presented to justifY that universality is an unattainable but worthy objective. We should strive towards measurement of functionings although we cannot avoid the measurement of resources, capabilities and rights as well. Indicators of living conditions are ultimately needed in democratic decision making but surveys descnbing preferences are also needed for increasing our self understanding. Both social theory and welfare theory are required as a basis for the measurement of the quality of hfe Finally, a list is presented of the most urgent improvements in depicting the quality of life. The greatest shortcomings are due to the way that welfare indicators are traditionally based on atomistic liberalism. The review concludes by calhng for more communalism in the mdicators of quality of life (i.e., descriptions of the distribution, discrimination, partiCipatIOn and social capital).

1. ON THE HISTORY OF MEASURING THE WELFARE OF THE SOCIETY

After the Second World War there was a growing interest in measuring welfare of societies. Gross domestic product (GOP) per capita soon became the key measure in this respect. The idea of social indicators gained official status in United Nations (UN) in the 1950s. Their use was a critique of the GOP measures. GOP per capita was criticised since it did not measure the distribution of income nor did it depict health, nourishment, housing, clothing, employment and working conditions, education, social security, leisure time, recreation nor human rights. Social indicators were developed primarily under the auspices of international organisations. The Nordic countries were forerunners in terms of national development. I This trend was probably linked to the accelerated development of the welfare state in the Nordic countries.

However, the development of indicators got side-tracked in the 1980's. Ooyal and Gough (1991, pp. 153-155) maintained that this was because of the rise of the new right in politics and the rise of relativism in the social sciences. An additional reason may have been the fear of technical ratio (i.e. against scientific politics). It was thought that indicators of quality of life (QOL) would become tools of

The history of social mdicators has roots back to the writings of economic and social scientists and ILO reports in the 1920s and 1930s. It should also be remembered that Aristotle's thoughts on good life greatly resemble the debate on social indicators (Nussbaum 1993) The history of social indicators is depicted by, among others, Johansson (1970, 19-21 and 1979), Roos (1973, pp 168-188), Galtung (1980), UNDP (1990, p 9), Doyal & Gough (1991, pp. 151-152) and Erikson (1993, p. 67). In addition to social indicators, there was talk about indicators using the terms the quality of life, welfare and the standard of living Welfare and the standard of living were often linked to GDP, while indicators of the quality of life and social indicators were understood as broader based measures of welfare.

63

E. Gullone and R.A. Cummins (eds.), The Universality af Subjective Wellbeing Indicators, 63-80. © 2002 Kluwer Academic Publishers.

64 THEORETICAL BASES

bureaucratic control and that those defining the indicators would assume a position of dogmatic authority. Indicators of QOL were seen as a part of the repressive " brave new world"?

One factor behind the diminished emphasis on measuring QOL may be the "bland empirism" prevailing in circles developing indictors. There was concentration on development of empirical measures without very deep consideration of the theoretical and philosophical foundations. The development of these indicators may divert attention away from the circumstances generating the processes. According to this way of thinking, the development and adoption of measures supports conservative stagnation.3

There are signs of a revival in interest about indicators of QOL. One example of this is Len Doyal's and Ian Gough's award winning book "A Theory of Human Need." (1991), which includes a wide range of suggested indicators.4 Another example is the excellent collection of articles "The Quality of Life" (Nussbaum & Sen 1993) presented at a conference arranged by the WIDER Institute. Both represent a leap forward in the theoretical and philosophical aspects of measuring QOL. Since 1990 the United Nations Development Programme (UNDP) has compiled an extensive annual report by renowned experts on the subject, in the form of the "Human Development Report", UNDP (1990-2001). The report includes a large number of indicators and comprehensive studies on the development of measures. The development programme reports have spawned critical debate, which has been taken into consideration when developing the reports.5

The most important sign of the revival is the rapid spread of measurement and assessment at all levels of society. Many development indicators ranging from QOL of special groups and local living and working conditions to global measures are once again "in fashion".

There is perhaps a strong connection between the revived interest in measures of QOL and the ongoing changes in production and society. Such indicators are well suited to societies based on networks and autonomous production units where matters are handled via negotiation and dialogue more than hierarchical decisions.

2 J.P. Roos addressed the problem of mdlcators related to bureaucratic controls and noted that mdicators were developed precisely for this purpose. (Roos 1973, 217-222). Johan Galtung regarded universal indicators as a dangerous illusion. The fear was that deep cultural and politIcal problems are solved as scientific technical problems. (Galtung 1980, 72-.)

3 This was noted by, among others, J P Roos (1973,168-193). • Their indicators measure phenomena like adequate nutrItional food and water and protective housing,

a non-hazardous work and physical environment, appropriate health care, security in childhood, significant primary relationships, physical and economic security, safe birth control and child-bearing, basic education, production, reproduction, cultural transmiSSIOn, political authority, civil and political rights, rights to access to need satisfiers, political participation. They succeed in finding statistical measures (at least proxies) to them. (Doyal & Gough 1991, p. 170.)

5 A lot of work on social mdlcators has been carried out in connection with development research on developing countries. Development research is an Important basis for the UNDP reports. An example of relatively comprehensive development research, covering industrialIzed countries as well, is the indicator project by Richard J. Estes (Estes 1988) consisting of analysis of social, economic, popUlation and cultural development. Estes' SOCIal development index based on 36 indicators gives its highest points to Denmark in all measurement years, I.e. 1970, 1980 and 1983.)

1. KAJANOJA 65

The revival of indicators is not a question only of measuring QOL, but rather a broader range of change indicators. The selection of the indicators is nevertheless motivated by the idea of promoting QOL. In this respect we are close to the social indicators of the 1960s and 1970s. Nevertheless, the differences are great, too. The previous phase was initiated at the top with primary emphasis on measurement of the state of the entire nation. Now the impetus has come from the bottom focusing on various aspects of society as all sorts of activities have been the objects of measurement and comparison.

In the second wave of the measurement of QOL, there appears to be even less consideration of theoretical and philosophical issues than in the 1960s and 1970s.6 The above mentioned books (A Theory of Human Need and The Quality of Life) have turned out to be exceptions.

2. FREEDOM OR A DEFINITION OF GOOD LIFE?

The basis for measurement of QOL was liberalistic individualism. This approach focused on a view that emphasised freedom, resources, autonomy, self-fulfilment and possibilities of choice. The main criteria were deemed to be the freedom of individuals to realise themselves.

The development of indicators was affected strongly by the psychologists Abraham Maslow and Clayton Alderfer, who studied universal needs. Their needs hierarchy extended to the needs for interaction and love. But even with regard to these issues they focused on the satisfaction of needs from an individualistic standpoint. Both ended up emphasising human "growth" and "self-fulfilment" as higher needs corresponding to a person's uniqueness (Maslow 1970; Alderfer 1972).

The orientation was so clear that it was a rare exception when Sten Johansson, often cited as the developer of indicators of QOL in the literature, considered measures of the good life as a possible point of departure. He ended up focusing on the idea of avoiding bad conditions, thus bypassing the question of a definition of the good life as the basis for measures. Ultimately he concluded that the indicators are a matter of resource measurement. Everyone thus uses resources to achieve their own concept of a good life (Johansson 1970; 1979).

The development of new indicators is strongly tied to individualistic liberalism. The indicators used by the UNDP are not based on a theoretical framework, but the main criteria for indicators often mentioned in the text is "freedom of choice". In line with individualistic liberalism, Doyal and Gough emphasised autonomy and health as basic needs from which individual indicators are derived. The main criteria in their line of thinking is expressed as follows: "The search must be for universalisable preconditions which enable minimally impaired participation in the forms of life both in which individuals find themselves and also which they might subsequently choose if they believe their existing form of life to be wrong." They stress dialogue and interaction. (Doyal & Gough 1991, pp. 50-55). Thus, they are already taking steps toward explicit definitions of communalism and the good life.

" As far as we know there are no systematic evaluation of the current measurement of the quality of life. So the view is based on random observations


"The Quality of Life" (Nussbaum & Sen 1993) is one of the first books to have articles featuring definitions of a good life and communalism. A common theme in the articles is the communitarian concept.7 Aristotle's concept of virtue is emphasised especially in the article by one of the two editors, Martha Nussbaum (1993).

The critique of individual liberalism can be linked to two hypotheses.s Individualistic liberalism cannot avoid defining the attributes of the good life even if one thinks so. The consequence is concealment of the choices regarding the attributes of a good life. The other hypothesis is that individualistic liberalism distorts the depiction of QOL since it attempts to avoid questions of content of the good life. Communal viewpoints in particular are avoided.

Mere freedom, resources, autonomy, self fulfilment or possibilities for choice are impossible, absurd, empty. Taylor describes this colourfully: "Full freedom would be situationless. And by the same token, empty. Complete freedom would be a void in which nothing would be worth doing, nothing would deserve to count for anything. The self which has arrived at freedom by setting aside all external obstacles and impingements is characterless, and hence without defined purpose, however much this is hidden by such seemingly positive terms as "rationality" or "creativity". These are ultimately quite indeterminate as criteria for human action or mode of life. They cannot specify any content to our action outside of a situation which sets goals for us, which thus imparts a shape to rationality and provides an inspiration for creativity." (Taylor 1975, p. 561). Taylor admired the ideal of freedom, but "the kind of critique we need is one that can free it of its illusory

7 When conSidering fair distribution of assorted benefits, the communitarian Michael Walzer concludes that all benefits (education, health, wealth etc.) have their own criteria for the fairness of distribution that depend on how the matter is seen in the culture m question. The overall result can nevertheless be assessed in light of the notion "complex equality" Walzer ends up focusing on citizenship that the complex equality bnngs with it Citizenship means full membership in the political community and does not allow for discrimmation in any sense. David Millar emphasises the meaning of fully participating citizens when interpreting Walzer's framework. (Walzer 1994,26; Miller 1995,12-15.)

R Taylor focuses his critic on Marxism when describing the emptiness of striving for freedom (Taylor 1975,560-562) Marxism finds strong emancipation elements, i.e. active striving toward a person's complete freedom. On the other hand, linking Marxism to the Hegelian-Marxist tradition there IS a strong common desire related to communalism (Pulkkmen 1996). The idea is that society becomes perfectly aware of itself and controls its life and thus frees itself from slavery of necessities. Thus Marxism becomes free of the critique of individualism, but it cannot avoid the critique regarding the emptiness of freedom. The next Citation in the text on Taylor's freedom may apply equally well to complete freedom of an individual as well as a completely free community There are many types ofliberalism. One of the strong common denominators behind various forms of liberalism IS the capability of individuals to be free and make choices. According to liberalism, freedom and the ability to choose are in-born traits of people. They are the pomt of departure and an important task is to safeguard them against threats coming from society. The idea of one-eyed freedom related to the critique of emptiness is focused on liberalism in a more straightforward manner than on Marxism

J. KAJANOJA 67

pretensions to define the totality of our lives as agents." And - he adds - the ideal of freedom should be purged of its "atomistic distortions". (Taylor 1985, pp. 7_9)9

Another way to express the emptiness of freedom is to note that when we speak of this we actually have in our mind an answer to the question of what do we need freedom for (cf. Erikson 1993, pp. 73-74; Scanlon 1993, p. 198). We cannot think of emptiness. Thus we cannot think of freedom without content.

Jussi Koskivirta perhaps represents the current mainstream - influenced by the critique of liberalism - thinking in social philosophy, when he writes as follows:

" . It has been claimed that when liberalistic thinking has nothing to say about the nature of good life, it actually commits to a certain view or promotes the realisation of a view in practice. In my opmion there is some truth to these claims." (Kotkavirta 1997, p. 172). A similar view IS expressed by another Finnish social philosopher Juha Sihvola in a study that won an award as the best scientific book of the year. After emphasising and praising the liberal concepts of the meaning of individual freedom, he maintains that "social ethics and policy needs the model of good life ... we need a description of the content of what we think a successful human life is like." (Sihvola 1998, p. 162).

The discussion about values as determinants of behaviour has increased even in behavioural sciences and even in economics (see e.g. Sen 2000). This discussion backs up the conclusion that freedom and resources are not enough as descriptions of QOL. We cannot avoid the value judgements concerning the content of life.

Since the content of a good life has to be defined in any case, it is better to hold open discourse about it and to make choices consciously. In the following sections several views that have identified as possible grounds for selecting indicators will be discussed.

The critique presented above does not negate the value of individual liberalism as a tremendous achievement of modernism. Great weight must be given to its principles. This topic will be dealt with in section four.

The choice between, on the one hand, individualistic liberalism and, on the other hand, communalism and the definition of a good life is not mere theoretical and philosophical hair-splitting. The choices affect concrete choices about the objects to be measured, as demonstrated in the previous section.

3. COMMUNICATIVENESS AND DIALOGUE

JUrgen Habermas regards communicativeness as the most comprehensive criterion when considering the communal solutions related to a good life (Habermas 1989 and 1991, 219). By communicativeness Habermas means striving toward a common understanding when solving matters requiring co-ordination among people. The alternative to communicativeness is an instrumental (i.e. based on game or on force)

9 Taylor's depiction can be augmented with the expressions that communitarlans used when describing the methodological approaches of liberalism. Communitarians criticized people-related expressions associated with liberalism such as 'independent', 'disengaged', 'unencumbered', and 'free to choose' (see e.g. Sandel 1992, 83-85; Walzer 1990). The ideas of Taylor have been integrated mto the communitarian critique


relationship with another person. The outcome is communicativeness if all parties can accept the manner in which individuals' conceptions about a good life are consolidated in the solution. This kind of outcome fills the requirements of "discourse ethics" as Habermas calls it.

Melissa S. Williams addresses the same question from the viewpoint of marginalised groups. She emphasises that we do not have the prerequisites to represent these groups: "... we need to confront the actual particularity of others' experience to reflect what our just obligations towards them may be ... Justice is not defined analytically, but discursively." (Williams 1995, pp. 80-81). In order to evaluate what fairness and equality mean, we have to have a good idea of the special benefits of marginal groups (Le. the demands related to their special situation). This requires a direct dialogue between the distributors of benefits and marginal persons.

This is a matter that concerns not only displaced persons but rather all of us. Is everyone special, (Le. "real"). Ultimately we are all the distributors of benefits if the society is supposed to be democratic. The most important aspect of well functioning communicativeness, (Le. discourse) in Williams' terms, is the requirement that benefits are distributed in a fair and equal manner, because this is by definition a requirement for fairness and equality.

To Habermas, communicativeness (discourse ethics) is an ideal measuring stick by which communal decisions can be assessed. As such it is an unattainable goal. In ancient times it was perhaps a realistic goal for free men, but in complicated modem society decision making, even in the most democratic of nations, is delegated to representatives and experts. Furthermore, the delegation of a certain degree of decision making to markets appears to be overwhelmingly efficient. The key question is how well does communicativeness steer the delegating process (i.e. what kind of conditions and constraints does it pose).

Communicativeness requires the removal of unnecessary physical, economic, scientific and administrative barriers from the path of communicative discussion and participation. Habermas nevertheless regards communicativeness as a mere method and denies that it entails choices for a good life. 10

Michael Walzer nevertheless notes how communicativeness's preconditions "constitute in fact a way of life" (Walzer 1994, p. 12). This is not a question merely of the procedure, but rather that the choice of the procedure determines the concept ofa good life. Agnes Heller (1982, pp. 35-42), Charles Taylor (1991, pp. 29-32) and Gianni Vattimo (1995) have also praised Habermas communicativeness for providing a means of classifying the main ethical and political choices of the modern world, but like Walzer they have disputed that Habermas' concept of communicativeness is just a procedural approach.

Habermas' critics have also emphasised that communicativeness is not a comprehensive notion even though it is a key criterion when considering communal solutions regarding good life. For example, future generations cannot participate in the discourse on common understanding. We nevertheless wish to take their interests into account when making our decisions. Thus it is not enough for the participants in

10 Habermas presents his theory In the extensive study Theorie des Kommunikativen HandeIs (198\). He has subsequently refined his theory, but the main notions have remained unchanged

J. KAJANOJA 69

the discussion to reach a common understanding of their own interpretation of what constitutes a good life. There is also a need for some sort of universal notion of human value and its prerequisites.

Emmanuel Levinas's dialogic philosophy can be regarded as a theory that raises the idea of communicativeness to a new level that Heller, Taylor, Waltzer and Vattimo have already mentioned in their critiques. In Levinas's philosophy, an interaction between two persons and the related ethical issues thereof have a philosophical basis. One person cannot control the other so the relationship is "infinite". There are always third persons involved in a relationship between two persons. Levinas's philosophy can well be called a social philosophy, as he himself does. What is significant is the emphasis on dialogue and communalism as the counterweight to the western individualism (Levinas 1996).11

It is well justified that the definition of the good life be made on the basis of communicativeness and dialogue. Thus the references to Habermas and his critics as well as to Levinas are examples of the directions from which a definition of QOL can be sought. These references illustrate themes of the debate that are the most relevant, interesting and the most neglected in the measurement ofQOL.12 We need a discussion of the attributes of the "good life".

4. LIBERAL FREEDOM AS SECOND POINT OF DEPARTURE

Can communicativeness and dialogue be used as direct indicators of QOL? How can they be measured? Closer evaluation of this matter unveils the power of liberalistic thinking on freedom and resources. The direct depiction of communicativeness and dialogue easily lead to a detailed description of behaviour. This may lead the analysis to get bogged down in detailed definitions of ways of life, about which the liberalists have justifiably warned the communitarians. When depicting good life directly there is a temptation to resort to use of opinion and preference surveys. The problem with this approach is discussed in section 7.

The direct depiction of communicativeness and dialogue and the related opinion surveys are useful when learning to understand ourselves and others. But they are questionable when used as indicators of QOL with respect to evaluating social policy solutions. We must therefore satisfy ourselves with measuring the prerequisites for communicativeness and dialogue. Is this tantamount to giving into the liberalistic resource viewpoint? Partially, but not completely since we are now asking what resources are being used for. When we emphasise communicativeness and dialogue, the focal point is on their prerequisites. Then we notice, for example, the significance of gender, race and marginal factors, the unequal distribution of prerequisites, and displacement (i.e. the prerequisites for equality and participation).

Perhaps this is one of the most important and interesting aspects of the philosophy of the good life and measuring QOL: how do we fit together the

II Levinas is a French philosopher who died m 1995. His ideas have recently become the subject of lively debate both in Europe and the United States.

12 By neglect it is referred to researchers. The citizenshIp debate and art deal with the questions about the content of good life. In research circles such a discussion is often avoided as bemg unscientific


inevitable description of the content of a good life with the idea that we should define it as little as possible, so that we do not end up promoting intolerant regulation of ways of life and stagnation.

5. UNIVERSALITY

Richard Rorty (1980) maintains that the objective should be to seek a common understanding as long as the means used entail argumentation and not force. But he is opposed to the calls for universality as smothering, as perpetual normalisation. According to Rorty it is sensible to start from the premise that "abnormal discourse" replaces the prevailing normal discourse and becomes the new normal discourse which eventually experiences the same fate of being replaced (Rorty 1980; 1991, pp. 211-220).

Charles Taylor has similar thoughts. He speaks of transitional universality (Taylor 1993). Taylor believes that universal indicators of QOL are not permanent criteria to be evaluated outside the context of the culture.

Taylor's thinking on transitional universality is illustrated by the following example. In cultures discriminating against women, the idea of the equality of women is often difficult to fathom. But one may get a representative of a culture discriminating against women to agree that all people are of the same human worth. Thereafter he may admit that in principal women could be given the same status as men. Next he will probably claim that equality could be achieved if only women were not so decisively suitable for housework or physically weaker and mentally less competent to hold jobs outside the home. The expression "if only" opens the door for rational argumentation. Is there any evidence to support the claims about the suitability for housework or weakness or incompetence? The argumentation may lead the representative of the culture where women are discriminated against to change his mind and the equality of women may be accepted as a universal criterion for QOL (Taylor 1993, p.228).

Taylor's transitional universalism thus requires common basic values (i.e. in Taylor's language the strongest of the strong values) (Taylor 1985, pp. 40-42). For example, one of the basic values was the combined human worth of individuals. But basic values change or at least may change. Even more susceptible to change are the rational arguments based on these basic values. This does not, however, mean it is right to give into agnostic relativism, as Taylor says (Taylor 1993, p. 230). \3

11 The similarity of Rorty's and Taylor's ideas becomes clear when Taylor's transitional universalism is compared to Rorty's answer to Jean·Francois Lyotard. Lyotard emphasizes disagreement and regards the tendency to strive toward a common understanding as dangerous (Lyotard 1988) Rorty stresses "the basis of invidious comparisons with suggested concrete alternatives, not .. that these Institutions are .. in better accordance with the universal moral law .. " He maintains that "The only common ground on whIch we can get together is that defined by the overlap between their communal beliefs and desires and our own" He emphasizes that 'The only 'we' we need is a local and temporary one .. " Furthermore, he notes that "there is no a prion philosophical reason why this attempt [to replace differends with Iiglte i.e. difference WIth consensus] must fail [as Lyotard argues], just as there is no a priori reason why it must succeed." He "only inSISts that, if these new terms [to respect the form oflife] have been adopted as a result of persuasion rather than force, they will be better than

1. KAJANOJA 71

Michael Walzer uses a similar idea called thin universalism, for which there is no ultimate foundation (i.e. the peoples do not find a set of ultimate values, to which they all could commit). (Walzer 1994, p. 18). Here he suggests an idea based on the writings of Rorty, Taylor and Walzer, which could perhaps be called weak universalism. We admit that we are never sure of whether the values we regard as universal will really turn out to be universal and admit that the thought of universal values is dangerous. On the other hand, we are seeking values that are as universal as possible.

Striving toward universality - or at any rate striving to sustain the universality already achieved - entails the danger of ethnocentric totalitarianism (i.e. denial of different ways of life). But a striving toward universality, that is conscious about its danger described above, may include a tendency which justifies striving toward universality. When the criteria used in describing QOL are adapted to be used for different cultures, they become increasingly general in nature and they allow an increasingly wider range of the forms of life. The assessment of different cultures obliges us to depict even more basic functions (i.e. to seek common denominators with respect to different ways of life thus leaving room for the internal pluralism of a culture). 14

As an example we can take divorces included in QOL depictions used by the UNDP. A low number is regarded as favourable. This question is strongly tied to culture. We can well think that a low number of divorces is a sign of weakness, the continuation of family turmoil because of the financial and cultural dependence of women. The UNDP explains the divorce indicator as follows: "the replacement systems - nurseries, health and unemployment insurance, and other social services -have not yet emerged. The uneasy transition [divorce] has often been marked by considerable hardship, especially for the children ... " (UNDP 1990, 39). This explanation in turn begs the following question: Shouldn't the indicator measure the situation of children after divorce? But nothing is permanent as Walzer notes:

. our common humanity will never make us members of a single universal tribe The crucial commonality of the human race is particularism: we participate, all of us, In thick cultures that are our own. With the end of imperial and totalitarian rule, we can at last recogmze this commonality and begIn the dIfficult negotiatIons it requires." (Walzer 1994, p. 83)

6. FUNCTIONINGS OR CAPABILITIES, RESOURCES AND RIGHTS?

Amartya Sen (1992) suggested the concepts of functionings and the closely related concept of capabilities. QOL could be depicted by functionings and capabilities.

the ones we are presently using - for that IS analytic of the meaning of 'better'" (Rorty 1991,211-220)

14 Similar ideas to this argument defending the aspirations toward universality have been presented in Habermas's defence of formalism as prInciples allowing the divergence of ways of life (Habermas 1982) Similar thoughts have been presented by Hawthorn and Sen as they seek to make even more thorough descriptions of activities as opposed to descriptions of goods. They also note that this facilitates the greater coverage of indicators In a universal sense (Hawthorn 1987, X-XI, Sen 1993, 47)


Sen's concepts are an important analytic step forward. According to Sen, QOL should not be assessed in terms of goods, money or other resources or as freedom or rights. 15 Life consists of functionings, which means beings and doings. According to Sen, assessments of QOL should be based on these factors. The greatest problem in depicting goods, resources, freedoms and rights is their different significance to different people. For example, a rather large portion of the population cannot drive a car. Nevertheless car ownership is often used as an indicator of welfare. Instead of car ownership, we should describe functionings (i.e. for example, how people reach destinations that are near and far).

Capabilities for Sen is a concept that derives from functionings. It depicts the functionings at the disposal of an individual. Sen uses the concept, because he justifiably thinks that the possibility to choose is an asset in itself. Capabilities cannot be observed directly and their existence is always subject to interpretation. Continuing with the above-mentioned example, even if a person has a driver's license and the financial means or physical proficiency necessary for driving a car, there is no guarantee that they will actually drive a car. They may be prevented from using it by old age or health reasons compounded by a lack of medication. The main phenomena that can be observed are functionings (Sen 1992; 1993).16

Free elections are an example of rights. We get closer to capabilities when we describe cognitive skills, such as literacy and education. They increase the possibilities offered by free elections to influence political activities. But the political system may be so authoritarian that it alienates the potential voters. The problem can be overcome by depicting the authoritarianism and corruptness of the political system. But there can always be other barriers to influencing political activities. In addition to capabilities and resources, we have to describe functionings, for example, voter tum-out. This provides a way of checking how influence is actually wielded via the political system.

Another example is the social right to receive sickness compensation. It depends on the definition of the sickness, which is always to some extent arbitrary. The depiction of QOL should include not just a description of social rights, but also a description of the realisation of social rights (e.g. sickness compensation actually received and not merely promised).

Various types of freedoms as well as financial, social and cultural rights are capabilities. Their mere existence is considered a form of welfare. On the other hand, it is impossible to verifY the existence and scope of rights until they are put to the test.

15 Sen illustrates his observations by rejecting John Rawls' notIon of primary goods in his description of welfare. Rawls' primary goods include rights and freedoms, capabIlitIes and resources, income and wealth. Capabilities and resources include goods, i.e. merchandise and services. (Rawls 1971,92-93.)

16 Stem Ringen interprets Sen's concept of capabilities by dividing it into resources and arenas. Arenas are alternatives offered by the socio-economic environment. Resources provide the capabilities to use these alternatives. Ringen's distmction is a clever contribution With respect to descnptions of welfare. It reduces the problems related to descnbing capabilities, but does not eliminate them (Ringen 1995, 7-8).

J. KAJANOJA 73

Functionings are events, the exercising of capabilities. Depictions of QOL should go beyond surveying capabilities to describing their use if we wish to verify the realness of the capabilities. On the other hand, unused capabilities that are omitted from the description of used ones go unnoticed. The depiction of functionings entails a dictatorial element. Functionings do not entail alternatives (Le. the freedom of choice is not included in a depiction of functionings).

The conclusion that can be drawn is that many indicators are needed to describe QOL. We have to cover functionings. We also have to depict rights, possibilities and resources. The reason for the latter is not just the description of possibilities of choice. There are problems in describing functionings. As Sen notes, it is often necessary to measure resources because the measuring of functionings is impossible in practice. For example, it is necessary to measure the disposable income of people because it is not possible to measure all the functionings that can be realised with the money.

The theoretical problem of where to draw the line entails the future. When social solutions are evaluated from the viewpoint of QOL, it is necessary to take into account not only the direct welfare impacts, but also sustainable development (i.e. the future welfare capabilities). Thus we cannot avoid talking about capabilities and resources.

7. LIVING CONDITIONS OR PREFERENCES?

Should QOL be depicted as a state of pleasure and satisfaction of preferences? Or should the object of the description be living conditions (i.e. functionings and capabilities as indicated in the preceding section). In the case of the former we concentrate on the measurement of subjective feelings of preferences. In the latter we focus on the measurement of objective living conditions. Regarding alternative forms of housing, for example, in the case of former we measure enjoyment and in the case of the latter we measure population density and other housing conditions.

The problems related to this question were illustrated in the presentations that Helsinki University Sociology Professor Erik Allardt and Stockholm University Sociology Professor Robert Erikson made at the 1998 Quality of Life conference arranged by the WIDER institute in Helsinki. Both are internationally renowned researchers in this field. "To base the choice of welfare criteria entirely on the subjective views of the people themselves is ... likely to lead to an unfruitful conservatism. On the other hand, a complete disregard of what people themselves say in its tum allows for a dogmatism of experts." (Allardt 1989, pp. 8-10).

"People's opinions and preferences should influence SOCietal planning through their activities as citizens in the democratic political process, not through survey questions and opinion polls. That IS,

goals for planning should be set up in terms of factual conditions, not in terms of people's satisfaction With these conditions It is the assumption that the planning and executive organs of the state act directly to influence people's satisfaction and happiness that is the basis for many of the futuristic hells suggested to us in literary works." (Erikson 1993, p. 78)

Allardt resorts to measurement of preferences in order to limit the power of experts. Erikson resorts to measuring living conditions for the same reason. The


approaches are very divergent. The explanation is that Allardt's alternatives are either solutions based on atomistic preferences or the experts' power while Erikson's alternatives are expert's power or democracy. Erikson fears top-down Gallup-based governance. Allardt fears top-down governance without opinion polls.

A common solution cannot be directly built based on the preferences of individuals. The consolidation of preferences requires either democratic discussion or less democratic intervention. 17 Opinion polls or some other way of assessing individual preferences are not in accordance with communicative ideals. Every interested party should participate in the debate seeking joint solutions. Atomism and sampling are not valid approaches. The evaluation of preferences can be an important matter when trying to understand the ways of the world (Le. when developing theories to increase our self understanding). Recently, for example, one phenomenon that has gained attention, in relation to QOL, is the growing problem of lonely elderly persons. The concern is based on survey studies. There are grounds to predict that the significance of research into subjective experiences will grow in studies of QOL (Beck et al. 200 I). The interaction between subjective feelings and objective conditions is an increasingly relevant focus in QOL studies. But it is a different matter how the measurements of subjective feelings of preferences will be used. They are needed in connection with democratic discussion, but they cannot replace it. In this respect the view of Erikson is well justified.

On the other hand, Allardt is a realist. Social solutions are not based on communicative discussion. It is pure utopia to think as Erikson does that social solutions are the result of discussion between all interested parties. Opinion polls are undoubtedly necessary as counterweights to concentration of power and the power of experts. Studies on opinion polls nevertheless show how important it is to be sceptical of opinion polls and to engage in critical discussion. The studies show how the choice of questions and their explanations as well as the interests of the survey maker may have a great impact on the results (Kangas 1995).

8. SOCIAL THEORY

Indicators of QOL have been criticised for diverting attention from processes generating living conditions. We need information about the processes (i.e. theories on the functionings of society and nature). The critique can be interpreted so that it does not downplay the necessity of measuring welfare, but it suggests that it would be more useful to divert some of the measurement resources to describing functional mechanisms. There are good grounds for taking the opposite view. Throughout the enlightenment that began in the 1600's and 1700's there has been criticism of squandered values and blind admiration of efficiency. Max Weber was the mantle bearer in the critique of the technical ratio of industrial society. He describes the strengthening grip of cold equipment, technological dominance and bureaucracy.

17 There have been proved In economIcs that It is Impossible to find an analytIcal solution to consolIdation of the pleasures and preferences of indIVIduals. One has to take Into consideration the views to be Integrated (Hausman 1992,58-65)

J. KAJANOJA 75

According to Weber, they threaten the freedom and significance of western peoples' lives as genuine choices vanish. Special technical ratio will continue its victory march until, as he wrote in the first years of the century, all fossil fuel is burned (Weber 1987). Zygmunt Bauman thinks that concentration camps are the logical outcome of a society based on technical ratio (i.e. a technologically advanced and bureaucratically functioning society based on rational thinking) (Bauman 1989). There are thus good grounds for demanding a shift in emphasis to the measurement ofQOL and as a result - to the evaluation of value choices.

At the same time it must be noted that measurement of QOL has an inseparable connection with social theory and natural sciences. Only those things that we think we can influence should be depicted. The weather is part of the description of welfare to the extent that it can be influenced.

9. WELFARE THEORY

What are the attributes of welfare that we seek to measure? The description is based on the theory of human welfare and its social conditions. According to Doyal and Gough, the ultimate criterion for welfare is participation in the form of life of the community. Preconditions for participation are physical health and the autonomy of the agent. They build around them a welfare theory (Doyal & Gough 1991). These authors present their suggestions for indicators in three categories namely basic needs, intermediate needs and social prerequisites. They also present concrete statistics comparing regional groups of the world's countries as well as separate figures for England, the United States and Sweden as well as regional differences between women and men.

Jussi Kotkavirta approaches the attributes of good life by compiling "an array of various needs into a framework of general concepts and assessing them as preconditions for a satisfying life". Kotkavirta continues: "I also want to claim that if the next three preconditions are met, it is at least not impossible that the life of an individual would be satisfying and in this respect happy. There are no guarantees for realisation of a good life." Those three preconditions are security, health and freedom (Kotkavirta 1997, p. 179). Kotkavirta's notion of general preconditions is thus rather close to that of Doyal and Gough. This is interesting given that the book by Doyal and Gough is not included in Kotkavirta's bibliography.

The welfare theory presented by Doyal and Gough offers a good base for developing and classifying indicators. There is room for a livelier discussion on welfare theory as a basis for indicators. It would be important that the discussion extend to the critical evaluation of the indicators used.

10. DIRECTION OF FUTURE DEVELOPMENT

At the end of this chapter it is in order to present some views on the direction in which indicators of QOL should be developed. Suggestions are based on the abovedescribed critique, according to which indicators have until now be developed on the basis of atomistic individualism. The development has centred around freedoms and


rights at the expense of good life although communalism is gaining more attention nowadays.

The preconditions for communicativeness and dialogue should reflect attention on, for example, gender, race and marginal position, unequal distribution of prerequisites, displacement as well as participation and its prerequisites. The indicators should provide greater insight into distributions than is currently the case. Distributions, groups and participation are rather slowly integrated in development of indicators. There has been a shift from averages to distributions, but they have often remained on a dichotomous level: men and women, whites and blacks, natives and immigrants, employed and unemployed as so forth. This is good, but distributions should be evaluated in an even more disaggregated fashion.

How do we develop welfare indicators that get behind the rather formal notion of political participation? How do we get an indicator of discriminatory use of power? How do we measure the power wielded by supranational and national market forces and political systems? How do we measure the employer's repressive power at work and use of repressive power in families and small communities? How do we measure more closely the realisation of economic, social and cultural rights? All in all the issue of the criticism is close to the critical concept of traditional structural violence and which has obtained new dimensions, for example, along with Foucault's power analysis.

How do we measure the other side of the coin (i.e. participation and trust in the community as a overseer of joint problems)? Aku Alanen and Lea Pelkonen describe this problem in their report on difficulties in compiling measures of social capital (Alanen & Pelkonen, 2000). Ready-made indicators simply do not exist.

In section 5, with the help of an example using divorce rates as a welfare indicator and how this should be replaced with indicators depicting the situation of children, the danger of universalism and ethnocentrism was presented. Almost all segments of analysis on QOL could benefit from this type of development work, where we seek to move away from culturally problematic indictors toward measurement of underlying phenomena.

In section 6 the problems that are associated with use of freedoms, rights and resources when depicting welfare were presented. The problems are with both the remoteness of the depiction from activities as well as the blindness of the depiction with respect to the capabilities to realise the freedoms, rights and resources as well as their distribution in the popUlation. Overcoming the problems related to freedom, resources and rights requires an array of indicators.

Sustainable development is a recent newcomer among the indicators of QOL. The first theme of sustainable development (i.e. ecology), was relatively remote still in the 1980s, as Allardt notes (1989, pp. 5-6). Global development gaps have appeared in the indicators even more slowly. The discussion on sustainable development has proceeded quickly to human and social capital and cultural questions. The broad-based notion of sustainable development is very new. 18 The

18 The broad-based concept of sustainable development was mentioned in the UNOP report from the year 1994 (UNOP 1994, 4-2 I). That notion was probably one ofthe first.

J. KAJANOJA 77

discussion will surely serve to change and widen the concept of sustainable development and it will affect the depiction of QOL.

The problem is neither merely the developers' selection of indicators nor difficulties in gathering data. The problems are much deeper. Societies gather data on things that the authorities regard as important. The discussion on indicators is a part of the debate on social policy choices.

SUMMARY

1. The 1960's and 1970's were the golden age for welfare measurement. There was talk of social indicators and QOL. The measures and their development came under criticism and the work on this subject waned. The criticism came from both the left and right. The criticism was part of that directed toward social planning and the rise of relativism. On the other hand, the measures were criticised as tools of bureaucratic control. It was feared that cultural and political problems would be solved "scientifically" as technical problems. It was feared that focusing attention on QOL would divert attention from processes generating living conditions. The development of indicators was downplayed as bland empirical research: attention should rather have been focused on the theoretical and philosophical bases of measurement. Now the development of indicators has experienced a revival, but perhaps less attention is being placed on its theoretical and philosophical bases than in the previous golden age.

2. Liberalism, Marxism and mainstream economic theory see welfare as both a form of freedom itself and as a prerequisite for freedom. Of pivotal importance is widening the scope of choices. There is talk of utility (i.e. maximising satisfaction given certain preferences). This view leaves QOL and good life as being up to each individual. This view is not satisfactory. It is not enough. The attributes of a good life have to be defined regardless of whether we want to do so or not. We have to answer the question of what freedom will be used for. It is best to do this consciously, and not try to conceal the notion of good life.

3. A well justified possibility is to base the definition ofQOL on communicativeness (JUrgen Habermas) and dialogue and communalism (Emmanuel Levinas).

4. In slight contradiction with what was just said above, it is also suggested that another main pillar of our work should be the liberal approach of leaving the attributes of the good life as undefined as possible. This is perhaps one of the main bones of contention in welfare philosophy: on the one hand, we cannot avoid defining good life and this debate is very challenging, but on the other hand good life should be defined as little as possible so that we do not end up promoting intolerant controls on ways of life.

5. Universality is an unattainable but worthy objective ifit can be achieved by means of the useful debate (i.e. dialogue between different schools of thought). But if striving toward universality leads one to apply the notion of QOL to different cultures, it also leads one to depict even more basic activities (i.e. attempts to consolidate assorted ways oflife while at the same time giving room to divergent forms of culture).


6. The focus of attention regarding depictions of welfare should be on activities, not resources, not capabilities, nor rights. Resources, capabilities and rights are nevertheless needed in describing welfare.

7. Depiction of democracy and living conditions go hand in hand. Democratic decisions are not based on survey studies of citizens' preferences, but rather they set the background for discussion on indicators of living conditions. Survey studies describing preferences are nevertheless needed for increasing selfunderstanding and addressing democratic deficiencies.

8. Measurement ofQOL must be increased as a counterweight to technical ratio. The indicators must be based on social theory. Theory is needed because the attention should be focused only on things that can be influenced. At the same time we notice that the objects of attention are those processes that augment QOL.

9. Defining indicators ofQOL requires welfare theory. Welfare conditions have to be described using, among other things, the notions autonomy, freedom, health and security. Welfare theory has been used to assess the relationships between indicators ofQOL and social structure, ecological factors, people's needs, desires and capabilities. There is room for a livelier discussion on welfare theory as a basis for indicators.

10. Finally, a list of the most urgent improvements in depicting QOL has been presented. The greatest shortcomings are due to the way that welfare indicators are traditionally based on atomistic liberalism. We have to add the communalism of the indicators (i.e. descriptions of the distribution, discrimination, participation and social capital). We also have to include descriptions of common activities associated with various ways of life and sustainable development. The reason for the dearth of indicators of communal phenomena, basic activities and sustainable development is not just a problem of gathering information. Societies gather information about things that decision-makers regard as important. This is a matter of political choice.

Address for Correspondence: Jouko Kajanoja, Docent, PhD (Pol.Se.), Director of Administration (ad int.), & Senior Researcher, P.o.Box 269, FIN-0053!, Helsinki, Finland. E-mail: [email protected] Tel: +358-9-7032985, gsm: +358-50-5810070 Fax: +358-9-7032969

REFERENCES

Alanen, A. & Pel konen, L (2000)' Can Regional Economic Growth be explamed by Social Capital. In J. Kajanoja & J. Simpura (eds.) Socwl Capital Global and Local PerspectIves. VATT-Publications Series A, no. 29. Helsinki: Government Institute for Economic Research.

Alderfer, C. P. (1972): EXistence, Relatedness, and Growth. New York: Free Press. Allardt, E. (1989)' An Updated Indicator System' Havmg, Lovmg, Bemg. Helsmki Department of

Sociology, University of HelsinkI, Workmg Papers N.o 48 Bauman, Z. (1989): Modernity and the Holocaust. Ithaca, N.Y.' Cornell University Press. Beck, W., van der Maesen, L., Thomese, F., & Walker, A. (eds.) (2001): Socwl Qualrty - A VISIOn for

Europe. The Hague. Kluwer Law International. Doyal, L. & Gough, l. (1991): A Theory of Human Need London: Macmillan

J. KArANOJA

Enkson, R. (1993). Descriptions ofInequality: the Swedish Approach to Welfare Research. In M. Nussbaum, & A. Sen (eds.): The Quality 0/ Life. Oxford: Clarendon Press.

Estes, R. J. (1988): Trends m World SocIal Development The SocIal Progress o/NatlOns, 1970-1987 New York: Praeger Publishers

Galtung, J (1980): The Basic Needs Approach. In K. Lederer (ed.): Human Needs. Cambridge, Massachuttes: OeIgeschlager, Gunn and Hain

79

Habermas, J. (1981). Theone des KommumkatlVen Handels. Band I, Handlungsrationalitat und gesellschafiliche Rationalisierung. Band 2, Zur Kritik der funktionalisllschen Vernunfi Frankfurt am Main: Suhrkamp

Habermas, J. (1982): A Reply to my Critics. In J.B. Thompson and D. Held (eds.). Habermas Cm,cal Debates. Cambndge, Massachuttes: The MIT Press.

Habermas, J. (1989). Ykstfollmen tahdonmuodostus: M,kii on tarko; tuksenmukalsta, mlkii hyviiiija mlka oikeudenmukalsta Tiede & edistys, vol. 14,86-95.

Habermas, J. (1991)' A Reply. In H Axel & J. Hans (eds): CommunicatIve Action. Essays on Jurgen Habermas's The Theory o/CommunlcatlVe Press Cambridge, UK: Pohty Press.

Hausman, D M. (1992)' The Inexact and Separate SCIence 0/ EconomIcs. Cambridge UK' Cambridge University Press.

Hawthorn, G. (1987) Introduction In A. Sen (ed. Geoffrey Hawthorn) The Standard of Llvmg Cambrtdge' University Press.

Heller, A (1982): Habermas and Marxism. In J.B. Thompson, & D Held (eds) Habermas Crttlcal Debates. Cambndge, Massachusetts: The MIT Press.

Johansson, S. (1970)' Om levnadsundersokmngen. Stockholm' Almanna F()rlaget. Johansson, S. (1979): Mot en teortfor social rapportermg. Stockholm: Institutet fur SOCial forskning. Kangas, O. (1995) You Get What You Ask/or Impact o/QuestlOn Wordmg on Responses. Paper

presented at the meeting of the Nordic Sociological Association, Helsinki 9. -11.6.1995. Kotkavlfta, J. (1997): Hyvan elilman yksil()lhsia Ja yhteis()lIisia ehtoja In J Keskitalo & J Kotkavirta

(toim.): Jarki, usko ja eettisyys. Jyvaskyla: SoPHi, Jyvaskylan yhopisto, Yhteiskuntatieteiden, valtioopin ja filosofian julkalsuja 17.

Levinas, E (1996)' BaSIC Phtlosophical Wmmgs. A T. Peperzak, S Cntchley, & R. Bernasconi, (Eds), Bloomington [Ind 1: Indiana University Press.

Lyotard, J (1988): The DifJerend: Phrases m DIspute. Manchester Manchester University Press. Maslow, A H. (1970). MotivatIOn and Personality New York. Harper. Miller, D. (1995)' IntroductIOn. In D. Miller, & M. Walzer (eds)' Pluralism, JustIce and Equality

Oxford: Oxford University Press. Nussbaum, M. (1993). Non-relative Virtues. In M. Nussbaum, A Sen (eds.) The Quality 0/ Life Oxford:

Clarendon Press. Nussbaum, M. & Sen, A. (eds): The Quality o/Life (/993). Articles are based on papers presented at the

WIDER conference on QualIty 0/ Life, Helsinki, 25-29.7 1988 Oxford: Clarendon Press. Pulkkinen, T (1996) The Postmodern and Political Agency. HelSinki: Department of Philosophy,

University of HelSinki. Rawls, J. (1971): A Theory 0/ JustIce. Cambridge (Massachusetts). The Belknap Press of Harvard

University Press Ringen, S (\995): Well-bemg, Measurement, and Preferences Acta Sociologica., vol 38,3-15. Roos, J.P (1973) Welfare Theory and SOCIal Policy. A Study m Policy SCIence. Commentationes

Scientariurn Sociahum 4/1973. Helsinki: Societas Scientarium Fennica. Rorty, R. (1980): Phtfosophy and the M,rror of Nature. Oxford: Basil Blackwell Rorty, R. (1991): ObjectIvity, RelatiVIsm, and Truth. Phtfosophlcal Papers Volume I Cambridge'

Cambridge Umversity Press. Sandel, M.J (1992): The Procedural Republic and the Unencumbered Self In r.B Strong (ed ): The Self

and the PolitIcal Order Oxford Blackwell. Scanlon, T. (1993): Value, DeSIre and Quality of Life. In M. Nussbaum, & A. Sen (eds.): The Quality of

Life. Oxford: Clarendon Press 1993. Sen, A. (1992): Inequality Re-exammed. Oxford: Clarendon Press Sen, A. (1993)' Capability and Wellbeing In M Nussbaum, & A Sen (eds ). The Quality of Life Oxford.

Clarendon Press. Sen, A. (2000): Development as freedom. New York. Knopf Sihvola., J. (1998): TO/von vuosltuhat. Eurooppalamen Ihmlskuvaja suomalmsen yhtelskunnan

tulevaisuus. Jyvaskyla: Alena Kustannus


Taylor, C. (1915): Hegel. Cambridge, UK: Cambridge University Press Taylor, C. (1985): Human Agency and Language. Philosophical Papers I. Cambridge, UK: Cambridge

University Press. Taylor, C. (1991): Language and Society. In H. Axel & J. Hans (eds.): Communicative ActIon. Essays on

Jargen Habermas 's The Theory of Communicative Action. Cambridge, UK: Polity Press. Taylor, C. (1993): Explanation and Practical Reason. In M. Nussbaum, & A. Sen (eds.): The Quality of

Life. Oxford: Clarendon Press. UNDP (United Nations Development Programme) (1990-2001): Human Development Report 1990-2001

(yearly). http://www.undp.orglhdro/ Vattimo, G. (1995): Kommunikaatioetlikka val tulkmnan elllkka? niin & nain. 2. 38-41. Walzer, M. (1990): The Communltarian Crlltque of Liberalism. Polttical Theory, 18, 6-23. Walzer, M. (1994): Thick and Thin. Moral Argument at Home and Abroad. Notre Dame, Indiana:

University of Notre Dame Press. Weber, M. (1981): D,e protestantische Ethlk. GOtersloh: Mohn. Williams, M.S. (1995): JustIce towards Groups: Political Not JuridIcal. Polmcal Theory. 23, 61-91.

JOAR VITTERS0, ESPEN R0YSAMB AND ED DIENER

THE CONCEPT OF LIFE SATISFACTION ACROSS CULTURES: EXPLORING ITS DIVERSE MEANING

AND RELATION TO ECONOMIC WEALTH

Abstract. The structural valIdity of the Satisfaction With Life Scale (SWLS) was tested with confirmatory factor analyses in 41 nations. In addition, life satisfaction was correlated with national wealth (GNP) in these societies, after correcting for the reliability of the life satisfaction measure. The homogeneity of the SWLS was found to be acceptable across all nations, but differed between levels of analyses. Aggregated to the national mean level, a nested factor was found to cause a rather strong covariance between the first and fifth SWLS items. Initially, the zero-order correlation between the SWLS and GNP was .42 en < .05). However, after controlling for the psychometric properties of the SWLS scale, such as reliability and model fit, the relation between national wealth and life satisfaction was reduced to .25 (n.s.). Regressing GNP on the factor loadings for all five SWLS items revealed that, relatively, the first item (i.e. In most ways my life is close to my ideal) is a more significant satisfaction with life construct in rich nations compared with poor countries. This and other findings relating to how the meaning of life satisfaction varies with culture are discussed.

Coming to grips with a scientific concept of quality of life is a challenging task. Gradually, however, an agreement seems to be developing that quality of life refers to both objective and subjective components. Lane (\996) for example, considers quality of life to be properly defined by the relation between two subjective-based elements and a set of objective circumstances, and related views are presented by (Cowen, 1991; Cummins, 2000; Diener & Suh, 1997; 1999; Fernandez-Ballesteros, 1998; Haybron, 2000; Kammann, et aI., 1984; Lerner, 1997; Mourn, 1992; Olson & Schober, 1993; Raphael, 1996; Ringen, 1995; Schwarz & Strack, 1999; Veenhoven, 2000; Watten, et aI., 1995; WHOQOL, 1998).

One aspect of quality of life, people's satisfaction with life, has attracted the attention of a substantial number of researchers in recent years (Argyle, 1996; Averill & More, 2000; Cummins, 1998; Diener, et aI., 1999; Eid, 1997; Hazelrigg & Hardy, 2000; Wright & Larsen, 1993). Defined as the degree to which an individual evaluates the overall quality of his or her life as favorable, several measurement scales have been constructed to tap into the construct (Cummins, 1998; Fordyce, 1988; McIntosh, 2001; Pavot & Diener, 1993; Pavot, et aI., 1998; Pelletier, et aI., 1996; Watson, 2000). Even if these scales have strong psychometric properties in industrialized western societies (e.g., Larsen & Fredrickson, 1999; Lyubomirsky & Lepper, 1999; Sandvik, et aI., 1993; Veenhoven, 1996), validity studies are still needed for the other nations and cultures of the world before considering them established (Segall, et aI., 1990). Researchers must test the cross-cultural generality of their findings before they can consider them to be established. Consequently, in a recent review of subjective well being across cultures, Diener and Suh (2000) argue that there are some promising, although limited, data on the structural properties of

81

E. GulloneandRA. Cummins (eds.), The Universality of Subjective Wellbeing Indicators, 81-103. © 2002 Kluwer Academic Publishers.

82 QUALITY OF LIFE IN RESIDENTIAL CARE

life satisfaction measures across nations, but more research must be conducted in this area. An important aim of this study is thus to provide a structural validation of the life satisfaction concept across nations.

Everywhere in the world the quality of people's lives depends on their ability to think well of their life. Even if what is needed for a person to be able to experience satisfaction and self-worth varies profoundly across cultures, the literature on crossnational differences in satisfaction with life has consistently revealed a high correlation between economic wealth and subjective well-being. For example, Cantril (1967) found the correlation between life satisfaction and his index of economic development in 14 countries to be .67. Ouweneel and Veenhoven (1991) reported that the correlation of mean income in 28 nations and the average SWB in them was .62, and Diener, et al. (1995a) reported the correlation to be .59 in another sample, covering 55 nations. Schyns (1998) discovered a .64 zero-order correlation between happiness and GDP in a study of 36 nations. Finally, Inglehart and Klingemann (2000) found the correlation between well-being and GNP to be .70 in their data.

However, these associations are only meaningful if the concept and measurement of life satisfaction are comparable across nations. In cross-cultural research there are many substantive things that might be going on, but one overlooked reason is that the reliability of the measure might differ across cultures. Hence, a second aim of this study is to explore possible co-variation between economic wealth and the psychometric qualities of satisfaction with life measurements. Given that such an association does exist, the correlation between happiness and wealth might be due, at least in part, to methodological rather than economical reasons. For example, if the measures of life satisfaction were less reliable in poor nations, the strong correlation between the wealth of nations and life satisfaction might arise because in the less wealthy nations the measurement of life satisfaction reflects more constructirrelevant variance.

THE NEED FOR CONCEPTUAL CLARITY

A serious obstacle toward scientific understanding is the casual nature in which scientific concepts are often defined and used. Holland (I995) refers to this problem as the "eye of the beholder" error, referring to how messy concepts seem to thrive within academic circles as long as the phenomenon being investigated is insufficiently constrained by empirical testing. Without such constraints researchers are left with too much freedom in assigning arbitrary labels to the ideas they are describing. To illustrate the importance of clear and unifYing concepts, consider one of Lakatos' favorite examples of the opposite, namely the way in which philosopher Georg W. F. Hegel defined electricity: "Electricity ... is the purpose of the form from which it emancipates itself, it is the form that is just about to overcome its own indifference; for electricity is the immediate emergence, or the actuality just emerging, from the proximity of the form ... " (Lakatos, 1999, p. 23). The sloppiness of this and similar definitions, based entirely upon mere reflection, suffer from too much freedom in assigning labels to what is not yet properly understood.

J. VITTERS0, E. R0YSAMB AND E. DIENER 83

In contrast, standard models from well-established disciplines constrain this freedom by putting the assumed relations among elements of a concept to empirical tests. The amazing success achieved in the natural sciences owes much to the development of fertile concepts. In biology for example, progress seems to have developed, "not so much by individual discoveries, no matter how important, or by the proposal of new theories, but rather by the gradual but decisive development of new concepts and the abandonment of those that had previously been dominant" (Mayr, 1982, p. 856, italics added).

In confirmatory factor analysis, an a priori conception of what is believed to exist in Nature is tested mathematically against observations. If the measurement does not fit the data, our understanding of what we are modeling is not supported, and the conceptualization should be reconsidered. By testing a priori specified hypotheses, a confirmatory factor analysis is helpful in investigating whether a measure designed to assess a construct actually contains a meaningful dimensional structure. As Descartes postulated long ago, establishing the meaning of a concept starts by partitioning it into its fundamental components so that one is able to see clearly and distinctly what these components are (e.g., analyzing the concept). Next, working forward, one must join the components together so that they reproduce the original phenomena (i.e., synthesizing the concepts). Such an analytical dialectic is precisely what a confirmatory factor analysis is about (Mulaik, 1987). In factor analysis we seek to break. down observed variables and their interrelations into the effects of linearly independent factors. We first analyze the observed relations to find the distinct component of the concept, and by means of mathematical algorithms, reproduce the observed relations from these components. Construct validity within this framework is supported if the factor structure of the scale is consistent with the constructs the instrument purports to measure (Floyd and Widaman, \995).

The above analysis means that if the underlying conceptual structure of satisfaction with life differs systematically across countries, the scale is not invariantly consistent with the underlying construct, and cross-cultural comparisons makes less sense. What is being compared, if the factor structure varies across cultures, is not the same phenomenon. Hence, a major goal for our article is to contribute to conceptual clarification by comparing the factor structure of the Satisfaction With Life Scale (Pavot and Diener, \993) across nations. Without structural in variance across nations, the scale might capture different meanings in different countries, and cross-cultural comparisons, for instance investigating the relation between wealth and well-being, become difficult to interpret.

LIFE SATISFACTION AND NATIONAL WEALTH

Notions of progress are closely linked to material growth. Although progress can take many forms, in the modem era it is principally defined as standard of living, and measured as per capita Gross National Product or Gross Domestic Product. And as noted, people in wealthy nations tend to report greater levels of satisfaction with life than people in poor nations, and a series of studies have revealed correlation coefficients in the area between .50 and .70 for economic wealth correlated with


subjective well-being (Cantril, 1967; Diener, et aI., 1995a; Diener & Suh, 1999; Inglehart & Klingemann, 2000; Ouweneel & Veenhoven, 1991; Veenhoven, 1989). Whether this is because satisfaction produces economic growth or whether economic growth produces life satisfaction is still debated (Mazumdar, 2000), although more evidence seems to suggest that the causal path goes from income to well-being rather than the other way (for a review see Diener & Oishi, 2000).

Although an association between national wealth and satisfaction with life seems well established, different theories still compete in terms of how to explain the observed relationship. First of all Easterlin (1974) seems to argue that, due to principles of social comparison, national economic prosperity is of no consequence for people's appreciation of life. As documented by the literature reviewed above, however, the non-relationship hypothesis has received little empirical support. Cantril (1967) on the other hand, suggests a linear relationship between economic growth and life satisfaction, and the above mentioned correlations provide some support for his explanation. Others, such as Zolatas (1981) point to the increasing costs of collective welfare and argue that it is more likely that in rich nations the costs of economic growth outweigh its benefits. Thus, Zolatas suggests a curvilinear relationship between welfare and quality of life, depicting a declining yield of economic growth that may even tum negative. After re-analyzing previous data, Veenhoven (1989) found some support for the curvilinear model. In a similar vein, Max-Neef (1995) has proposed a "threshold hypothesis" for the relation between economic growth and quality of life. Only up to a threshold point will material growth affect people's happiness. If there is more economic growth, quality of life may begin to deteriorate.

Recently Lane (2000b) argued that further economic development in modem democracies will reduce the level of well-being, basically because market and democratic institutions are unable to improve the hedonic level of their populations. According to Lane, the common belief that, even above the poverty line, subjective well-being is a direct function of income level, is a so-called economic fallacy. Lane's conclusion is straightforward: " ... to increase the quality of life, the US and probably all Western advanced societies should move from an emphasis on money and economic growth toward an emphasis on companionship" (Lane, 2000a, p. \04). Further evidence of a weakening nexus between growth and well-being comes from the development of new measures of progress, such as the Index of Sustainable Economic Welfare (ISEW) and Genuine Progress Indicator (GPI) (Eckersley, 2000). According to Eckersley, the tendencies in western countries have been that, whereas GNP has continued to climb in recent years, the ISEW and opr has leveled off or fallen.

Partly in contrast to the above hypothesis and results, Schyns (1998) in her analysis of cross-national differences in subjective well-being, observed that rich and poor countries grouped into two distinct clusters, each of which revealed a different relation between happiness and wealth. Actually, when separate analyses were conducted for poor and rich countries independently, the correlation between wealth and happiness became negative (r = -.31) for the poorest countries, and remained positive (r = .40) for the rich countries (Schyns, 1998). These results strongly contradict the threshold hypothesis or need theories that argue that for

1. VITTERS0, E. R0YSAMB AND E. DIENER 85

people to thrive, basic needs must first be met. Note however, that Schyn's results do not take differences in the distribution of wealth into account.

The reason why money does not predict wen-being in poor countries, or that it predicts it negatively (i.e. in the data reported by Schyns), is not wen understood. It might be related to confounding variables that covary with both GNP and subjective well-being. For instance, the relation between money and satisfaction might be biased by the fact that economic wealth correlates with a variety of other developmental characteristics. In a study of 101 nations, Diener and Diener (1995) found that wealth was significantly correlated with 26 of 32 indices of quality of life. Similarly, others have shown how typical features of modernization, such as the fulfillment of basic human needs, greater human rights, individualism and social equality, correlate highly with economic prosperity, and might thus confound the reported relation between wealth and happiness (Schyns, 1998). Because the sample of nations is limited in size, however, it is difficult to isolate the influence from the possible covariates. In other words, we still do not know whether it is the economy in itself, or other phenomena that cause the higher life satisfaction reported in richer countries.

For similar reasons, one can inquire about the validity of life satisfaction measurements when used in different cultures. Although issues such as translation problems (Ouweneel and Veenhoven, 1991) and response artifacts (Diener, et al., 1995b) seem to be of lesser concern, the method of measurement has some influence on life satisfaction estimates for nations. For example, a positivity bias might influence global measures of subjective well-being more in some countries than in others (Diener et al., 2000).

With respect to structural consistency and factorial invariance across nations, we also need more knowledge to provide better interpretations of cross-cultural data. After an, the important message from current cultural psychology is that the character and meaning of everyday life are systematically related to cultural communities (Shweder & Haidt, 2000), which implies that similar questionnaire items might be given different interpretations depending on cultural group membership. Because implicit meanings determine the observed response pattern in questionnaire data, it is important to understand whether these interpretations differ systematically from one cultural group to another. Revealing structural similarities and differences in cross-cultural data is an important step toward this goal.

AIMS OF THE STUDY

The aims of this study were twofold. First, to give an account of the conceptual structure of the Satisfaction with Life Scale across 41 nations. Second, to provide a critical exploration with regard to how methodological effects might preclude the relationship between life satisfaction and economic wealth across these countries. No specific hypotheses were framed for the study, but the following research questions were central in guiding the analyses.

Research question 1. To what extent does the single factor model of the Satisfaction With Life Scale fit with data from different cultures?


Research question 2. To what extent does Gross National Product per capita predict the national mean level of life satisfaction?

Research question 3. Is national wealth a better predictor of satisfaction in rich countries compared with poor, as suggested in other studies?

Research question 4. How do methodological effects such as reliability and model fit influence the relation between satisfaction and GNP?

Research question 5. Does the relative meaning of each SWLS item differ across cultures?

METHOD

Participants

The data were collected during 1995-96 by Ed Diener and his colleagues from around the world, compiled by Diener and made available to the collaborators on the project. The study is referred to as the International College Student Data (ICSD), and further details about it are provided in Suh, et al. (1998). The current analyses are based on the 6,949 participants in the study (60% females) representing college students from 41 countries. Their mean age was 22.4 years, ranging from 16 to 69 years. The 41 nations and their respective sample sizes are summarized in the two first columns in Table 2.

Measures

The Satisfaction With Life Scale (SWLS) is a short, 5-item instrument designed to measure global cognitive judgments of one's live (Diener et aI., 1985; Pavot and Diener, 1993). The items are: In most ways my life is close to my ideal (SLWS-I), The conditions of my life are excellent (SWLS-2), I am satisfied with my life (SWLS-3), So far I have gotten the important things I want in life (SWLS-4), and If I could live my life over, I would change almost nothing (SWLS-5). The SWLS items are responded to on a scale ranging from I ("strongly disagree") to 7 ("strongly agree") and are assumed to reflect one single latent variable of life satisfaction.

Gross National Product. Per Capita Gross National Product (GNP) is a standard measure of economic growth that represents the total amount of money exchanged in a country and year, divided by the number of residents. It is an adequate indicator for measuring a specific country's economic performance. The source of the GNP was the World Bank..

ANALYSES

Initially the data were analyzed by means of Confirmatory Factor Analyses (CF A) using the EQS 5.7 for windows (Bentler, 1995). CFA represents a sophisticated extension to the traditional exploratory factor analyses. Basic to CFI is a desire to explain the interrelations among a set of variables with a model, and then test this


model statistically. In contrast to the essence of exploratory factor analysis, which, according to a series of papers written by Louis Guttman in the 1940s and 1950s, is to generalize from measures created by the researcher to more measures of the same kind (cited in McDonald, 1999, p. 7), a common way to interpret a CF A is that it enables the researcher to establish the objective validity of a substantive model (Mulaik & James, 1995). In our case, the model to be tested is the assumption that only one hypothetical construct (a factor) causes the observed intercorrelations among the five items in the Satisfaction With Life Scale. Herein lies the real power of CF A, namely that we can reject models that are inconsistent with the data. If we detect a serious gap between the model and the data, the notion of a common satisfaction with life construct should be rejected.

When judging the adequacy of a model, more than one fit index should be consulted (Bollen & Long, 1993), and quite often both so-called incremental and absolute fit indexes are presented together with the Chi-square statistics (Hu & Bentler, 1995). We use the Comparative Fit Index (CFt) as our incremental fit measure, and the Root Mean Square Error of Approximation (RMSEA) as our absolute fit index. Whereas the Chi-square test assesses the magnitude of the discrepancy between the sample and the fitted covariance matrices, the CFI assesses the adequacy of a target model in relation to a baseline model. It represents the proportion of total covariance among observed variables explained by a target model, when using a non-correlated model as the baseline model (Hu & Bentler, 1995). For a one-factor model, a low CFI -value indicates that much covariance is left unexplained (Le., one or more other common causes operating on the observed correlations). A high CFI-value indicates that we cannot reject the hypothesis of one common cause behind the observed correlations. The RMSEA estimates the error of approximation, which refers to the error due to model misspecification (Steiger & Lind, 1980). Recent analyses have documented that it is somewhat sensitive to sample size (Jackson, 200 I).

One obvious limitation of the Chi-square test, is its sensitivity to sample size and thus turns the power of a test against a careful researcher. Analyses involving large samples can lead to high Chi-square values and to rejection of good models. In a similar way, low statistical power can mislead us into retaining poor models. CFI, on the other hand, seems to be relatively unaffected by sample size (Marsh, et aI., 1996).

Traditional convention suggests a CFI cutoff point of .90 for model acceptance (Bentler & Bonett, 1980), but recent investigations have shown that a CFI value close to 0.95 (or a RMSEA value lower than 0.06) are needed before it can be concluded that there is a good fit between the hypothesized model and the observed data (Hu & Bentler, 1999). Moreover, a small confidence interval for the RMSEA estimate indicates that the factor structure is highly stable. A confidence interval located entirely below a limit of 0.08, is acceptable for a model of mediocre fit (MacCallum, et aI., 1996).

According to classical test theory, reliability is the consistency of a measurement. For confirmatory factor models the reliability of a variable can be defined as the magnitude of the direct relation that all variables except the error term has with it (Bollen, 1989, p. 221). This indicates that a variable's squared factor


loadings measure the proportion of variance that is explained by the variables that directly affect it, which is to say that the sum of the squared factor loadings (except that from its error term) reflects the reliability of the variable. Hence, in the current article, the reliability of the SWLS is measured as the mean of the squared factor loadings for the five SWLS items.

RESULTS

The SWLS items

Our first objective was to describe each item in the SWLS with regard to its raw score means, skewness, and loadings on a single, common factor (see Table 1). Separate analyses were undertaken for each nation, and the model converged without further complications for all nations but one. For the Nepalese data, the error variance related to item-3 was negative (a Heywood case - cf. Dillon, et aI., 1987). To remedy this situation, the error variance for item-3 was restricted to be 0.001. Except for this problem, the five items behaved properly across nations. For instance, none of the items were severely skewed, and all factor loadings were in the range between .40 and .94 for the first three items, and between .23 and .89 for items four and five (Nepal was excluded from these descriptives due to the lack of convergence ).

Table 1. Item descriptivesfor the five Satisfaction With Life Scale items across 41 nations

Nation Item 1 Item 2 Item 3 Item 4 ItemS

Mean Skew F.L. Mean Skew F. L. Mean Skew F.L. Mean Skew F. L. Mean Skew

Argcntma 4.7 -0.6 0.53 3.9 0.2 0.62 5.0 -0.7 0.67 4.6 -0.4 0.55 4.2 -0.1

Australia 4.5 -0.4 0.86 4.9 ·0.7 0.78 4.9 -0.7 0.81 4.7 -0.5 0.71 4.1 -0.1

Austna 5.0 -0.9 0.84 5.1 -0.5 0.66 5.2 -0.9 0.77 4.8 -0.5 0.72 4.4 -0.4

Bahrain 4.4 -0.8 0.72 4.0 -0.3 0.56 4.7 -0.6 0.62 4.0 -0.1 0.61 2.7 1.0

Brazil 4.5 -0.7 0.55 4.0 -0.1 0.58 4.7 -0.3 0.64 4.6 -0.5 0.59 3.8 0.2

China 3.6 0.1 0.62 3.7 0.1 0.66 3.8 0.1 0.72 2.4 1.3 0.90 2.9 0.8

Colomblil 5.0 .0.6 0.86 5.3 -1.1 0.62 5.7 -1.4 0.87 5.4 -0.9 0.69 5.0 -0.6

Derunark 5.1 -0.6 0.82 4.9 -0.7 0.69 5.6 -1.2 0.90 5.0 -0.7 0.63 4.6 -0.5

Egypt 4.4 -0.4 0.85 4.5 -0.4 0.94 4.6 -0.4 0.91 4.4 ·0.1 0.89 4.2 -0.3

Estorua 3.7 -0.3 0.70 3.8 0.2 0.48 4.8 -0.7 0.65 4.5 -0.6 0.55 4.0 0.2

Finland 4.6 -0.7 0.77 4.8 -0.8 0.70 5.0 -0.8 0.89 4.7 -0.7 0.63 4.3 -0.1

Gennany 5.2 -1.0 0.81 4.4 -0.2 0.55 40 -0.8 0.88 4.3 -0.2 0.60 4.3 -0.2

Ghana 4.6 -0.7 0.70 3.9 -0.1 0.77 4.7 -0.5 0.75 3.3 0.4 0.62 3.5 0.4

Greece 4.1 -0.2 0.87 4.2 -0.3 0.59 49 -0.8 0.76 4.0 ·0.1 0.65 3.6 0.1

Guam 4.5 -0.5 070 44 -0.5 0.71 4.5 -0.6 0.67 4.0 -0.1 0.58 4.0 -0.1

Hong 3.8 -0.1 0.63 3.8 01 0.78 4.3 -0.3 0.87 3.9 ·0.1 0.63 3.5 0.2 Kong

Jo VITTERS0, Eo R0YSAMB AND Eo DIENER 89

T bl I ed a e contmu

Nabon Item 1 Item 2 Item 3 Item 4 Item 5

Mean Skew F.L. Mean Skew F. L. Mean Skew F. L. Mean Skew F. L. Mean Skew

Hungary 4.! -0.4 0.72 4.7 -0.4 040 6.9 -O.! O.!I 4.2 -0.1 0.74 3.9 0.1

Italv 4.3 -0.3 0.79 4.0 0.2 0.54 4.7 -0.7 0.79 4.5 -0.4 0.66 4.2 -0.1

India 5.1 -0.9 0.42 4.6 -0.4 0.52 4.6 -0.2 0_!3 4.4 -0.2 0.75 4.5 0.5

Indonesia 4.9 -1.0 0.79 4.9 -0.8 0.73 4.4 -0.4 0063 4.! -1.0 0.62 2.9 1.I

Japan 4.3 -O.! 0.74 4.3 -0.1 0.77 4.4 -0.3 0.83 4.2 -0.1 0.64 3.1 0.6

Korea 3.7 0.1 0.68 4.1 -0.1 0.77 3.9 0.1 0.79 4.0 -0.1 0.69 3.0 0.7

Lithuania 3.7 0.1 0.69 3.3 0.4 0.64 4.3 -0.4 0.70 3.9 0.2 0.51 3.5 0.4

Nepal· 4.! -0.6 0.13 4.5 -0.6 0.32 4.7 00.4 0.99 3.8 -0.1 0.61 3.2 0.4

Netherland 5.0 -1.4 0.80 5.4 -1.0 0.73 5.B -1.0 0.66 5.B -1.4 0.61 5.0 -0.9

Nigert,a 5.3 -1.0 0.52 4.5 -0.5 0.73 5.0 -0.9 0.70 3.4 0.5 0.33 3.3 0.5

Norway 4.B -0.9 0.B5 5.1 -1.0 0.69 5.4 -1.2 0.80 5.2 -0.9 0.76 4.8 -0.7

Polus"" 4.6 -0.4 0.67 4.6 -0.6 0.68 4.9 -0.1 0.76 4.8 -0.6 0.65 3.9 0.1

Peru 4.7 -0.7 0.71 4.5 -0.2 0.40 5.0 -0.7 0.78 5.0 -0.7 0.57 4.2 -0.1

Portugal 5.0 -1.0 0.74 4.4 -0.4 0.49 4.3 -0.7 0.83 4.9 -O.! 0.68 3.8 0.1

Puerto 5.0 -0.6 0.80 5.1 -0.4 0.71 5.3 -0.7 00!5 5.2 -.08 0.84 4.7 -0.4 Rico

Singapore 4.6 -0.7 0.78 4.8 -0.6 0.54 4.9 -0.9 0.!2 4.4 -0.3 0.64 3.8 0.3

Slovenia 4.8 -0.4 0.90 4.9 -0.6 0.65 5.5 -1.I 0.77 5.1 -0.8 0.67 4.2 -0.1

Sth Africa 4.4 00.6 0.71 4.2 -0.2 0.77 4.5 -0.5 0.81 4.3 -0.4 0.61 3.5 0.4

SpaID . 4.3 -0.4 0.73 4.4 -0.1 0.61 4.9 -0.6 0.!2 4.5 -0.4 0.73 4.2 -0.1

Taiwan 4.1 -0.1 0.77 4.5 -0.3 0.67 4.5 -0.4 0.86 3.7 0.2 0.67 3.4 0.4

Thailand 4.9 -0.7 0.74 4.2 -0.1 0.73 5.5 -1.2 0.63 5.0 -0.8 0.59 4.1 -0.1

Turkey 4.0 -0.7 0.50 3.3 0.2 0.49 4.0 -0.6 0.71 4.5 -0.9 0.40 3.2 0.4

USA 4.8 -0.9 0.84 4.8 -0.7 0.80 5.1 -1.\ 0.85 4.8 -0.7 0.67 4.2 -0.2

Zunbabwe 4.3 -0.5 0.62 3.4 0.2 0.83 4.1 -0.4 0.80 3.5 0.2 0.60 2.8 0.9

MEAN 4.5 -006 .72 4.4 -0.3 .66 4.8 -0.6 .77 4.4 -0.3 .64 3.8 0.1

SD .44 -.42 .Il .51 -.22 .12 .46 -.02 .09 .63 -.67 .10 .60 -.08 ~ Skew· Skewness, F. L.:< Factor loadmg; SO '" Standard Devlltiott~ $) m the Nepalese sample, the vanance oCE3 was restricted tobe .00'1 due to a Heywood case. The petalese data is excluded from the mean and standard dcviahon for factor loadings.


The first item had a factor loading ranging from .42 (India) to .90 (Slovenia) with M = .72 and SD = .1 I. The second item ranged from .40 (H un gary ) to .94 (Egypt) with M = .66 and SD = .12. For item 3, the mean was .77 ranging from .43 (Tanzania) to .91 (Egypt) with a standard deviation of .09. Descriptives for item 4 were: M = .63, minimum score = .33 (Nigeria), and maximum score = .89 (Egypt) and SD = .10. Item 5 ranged from .23 (Indonesia) to .79 (Puerto Rico) with a mean of .57 and a standard deviation of .12. The third item in the SWLS yielded a higher raw score mean, lower standard deviation and a higher factor loading compared with the other items. In several respects, the results for item 3 are in accordance with previous analyses of the scale (see for example Pavot & Diener, 1993), indicating that question 3 ("/ am satisfied with my life") is the most central item in the scale.

CONCEPTUAL STRUCTURE OF THE SWLS SCALE

The SWLS inventory is constructed to reflect a single dimension of life satisfaction (cf. Figure 1), and this assumption was investigated by a series of confirmatory factor analyses, including analyzing the total sample (n = 6806), an aggregated sample (n = 41), and each sub-sample from the 41 nations separately. The one-factor model achieved acceptable goodness-of-fit estimates for the total sample. Although the Chi-square was significant and rather high: Ii (6806, 5) = 160.1, P <.001, the CFl of .99 suggested excellent fit, and the RMSEA (0.07, CI = 0.06 - 0.08) suggested a mediocre fit. For the aggregated nation data, the model fit was less promising, as indicated by Ii (41,5) = 17.7, P = .003, CFl = .91, and RMSEA = 0.25 (0.13 - 0.38), which means that satisfaction with life scale is worse off at the cultural level compared to the individual level. A multi-group analysis revealed a significant difference between the factor structure for the two levels (Ii (5) = 51.8, P < .00 I). A post hoc modification check on the aggregated data (using the Lagrange Multiplier test) showed that items 1 and 5 were influenced oppositely from a second (nested) factor. A re-specified model including a nested factor improved the fit for the aggregated data to Ii (41, 4') = 6.0, P = .194, CFl = .99, and RMSEA = 0.12 (0.00 -0.28). In other words, at the cultural level the first and fifth SWLS items share some covariance over and above what is provided by the satisfaction with life construct. The factor loadings for both models are presented in Figure I, with the results from the aggregated data in parentheses.


In most ways my hfe IS close to my Ideal

72( 7

The condltloos of my hfe are excellent (-45)

)E------'-'.J....:.:"-+ I am sattsfied WIth my hfe i '-_------'I ~ ~

61 (87)

So far I have gotten the Important

tllIngs I want m hfe

If I could hve my hfe over, I would

change almost notIlIng

( 49)

Figure I. One factor modelfor the Satisfaction With Life Scale withfactor loadings (aggregated model in parentheses)

When analyzing the single-factor model across all of the 41 nations, a total of 26 nations revealed non-significant Chi-square values (atp < .05), 10 nations had Chisquares probabilities in the range of .05 to .001, and 5 nations had Chi-squares significant at p < .00 I. However, since the test statistics for Chi-squares are sensitive to sample sizes, these results depend on the number of participants from each nation in addition to mere model fit. Actually, we discovered a significant and negative correlation between sample size and p value (r = -.34, P < .05) across the 41 subsamples. All of the 41 nations reached CFI estimates above 0.90, and 29 nations reached a level of 0.95 or above. With regard to the RMSEA, twelve nations had RMSEA levels below 0.05, and seven nations had their scores between 0.05 and 0.08. Fifteen nations had RMSEA values above 0.10. Table 2 provides details from these analyses.

Table 2. Descriptives of the one factor Satisfaction With Life Model across 41 nations

Nation Sum- Standard Skewness Muitiv. R' 'It P CFI RMSEA Clof score deviabOD kurtoIIS RMSEA

AlgeobDa 89 22.4 4.6 -0.15 -0.9 33 22 0.826 1.000 0.000 0.000.0.088

Australia 288 23.0 6.6 -0.42 2.0 56 83 0.142 0.999 0.048 0.000- 0.103

Austna 159 24.3 57 -067 5.6 52 \02 0.680 0.982 0.082 0.000 - 0.153

Bahram 122 19.9 5.5 -0.70 0.4 35 128 0.250 0.927 0.114 0.037 -0.192

Brazil III 21.6 5 I -0.25 -0.5 33 89 0.111 0.952 0.085 0.000-0.172

Cbma 543 164 5.6 0.26 0.8 34 282 0.000 0.950 0092 0.061-0.127

Colombia 99 264 55 -1.29 4.0 54 64 0.273 0.993 0.054 0000 - 0.156

Denmark 90 25.0 5.5 -0.59 3.3 51 159 0.7 0.939 0.157 0.074-0245

Egypt 119 22.0 6.7 -0.19 13.6 77 98 OBO 0.991 0.091 0.000- 0.173


Table2c_

NatiOll N Sum- StaDdardD Skewness Multv. R' t P CFI RMSEA Cl of score eviatiOD kurtoSlS R.MSEA

E,toma 116 20.9 5.2 -0.25 -1.7 31 131 0.22 0.904 0.119 0.041- 0.199

Fluland 90 23.4 6.2 -O.n 1.0 56 82 0.148 0.983 0.085 0.000-0.183

Germany 105 23.3 6.4 -0.39 3.8 51 70 0.222 0.990 0.063 0.000 - 0.159

Gbaaa 113 20.1 7.0 -0.06 -1.6 47 225 0.0 0.903 0.177 0.106-0.253

Greece 127 207 6.2 -0.09 -0.9 SO 83 0140 0.985 0.073 0.000-0.156

Guam 170 21.4 62 -0.25 1.7 41 224 0.0 0.914 0.144 0.086-0.206

Hoog Koog 137 19.3 5.4 -0.18 3.1 47 68 0.232 0.991 0053 0.000 - 0 138

Hungary 72 22.5 5.7 -0.01 -0.5 45 95 0.90 0.956 0 114 0.000 - 0.220

Italy 282 21.5 6.2 -0.18 -0.4 47 88 0112 0.991 0.052 0.000 _ 0.107

Judi. 92 22.1 5.1 0.16 3.2 38 44 0.4B7 1.000 0.000 0.000-0.137

Indonesia 90 21.9 5.1 -0.48 4.6 40 152 0.9 0905 0.152 0.068-0.240

Japan 195 20.2 5.3 0.05 1.3 SO 80 0.157 0.991 0056 0000-0.124

Korea 273 18.7 5.9 013 20 52 78 0.168 0.994 0.045 0.000 _ 0.103

Lithuania 99 18.7 5.7 0.22 -1.1 41 22 0.814 1.000 0.000 0.00 - 0.086

Nepal" 92 20.9 4.8 -0.16 -2.0 24 55 0.485 1.000 0.000 0.000 _ 0.129

Nelherland 37 26.8 4.7 -1.34 2.7 47 26 0.762 1.000 0.000 0.000 _ 0.157

NIgeria 233 21.4 5.6 -OOB 0.2 33 In 0.3 0.933 0.105 0.054-0.159

Norway 98 25.2 6.1 -0.99 4.0 60 22 0.824 1.000 0.000 0.000-0.084

Pakistan 151 22.8 6.0 -0.32 1.8 47 255 0.100 0.912 0.165 0.105 -0.231

Peru 128 23.3 4.9 -0.23 6.7 36 84 0.136 0.971 0.073 0.000-0.156

Ponugal 139 23.0 57 -0.53 2.9 44 60 0308 0.995 0.038 0.000 - 0.128

Puer. RIco 86 25.3 5.7 -0.54 1.5 64 66 0.606 1.000 0.000 0.000 - 0.126

Singapore 130 22.4 5.5 -035 1.0 45 117 0.38 0.964 0.102 0.022 -0.179

Slovema SO 24.4 6.1 -0.64 -0.3 56 88 0.118 0.964 0.126 0.000-0.254

S. Africa 366 21.0 6.7 -0 17 1.7 49 94 0.93 0.993 0.049 0.000 - 0.097

Spain 321 22.4 6.0 -0.23 17 49 n 0.173 0.995 0.041 0.000-0.095

TaIwan 530 20.1 6.4 0.05 1.7 5S 566 00 0954 0.140 0.108-0.173

TIIIZOlIia 129 21.0 5.8 -O.3S 0.1 37 206 0.0 0.918 0 157 0.090 - 0.228

Thailand 91 23.6 5.2 -0.18 4.3 44 132 0.20 0.933 0.136 0.048 -0.225

Turkey 98 28.9 5.0 -0.08 -1.3 27 94 0.94 0.911 0.096 0.000 - 0.187

USA 438 23.6 6.7 -0.70 9.1 60 22.1 0.0 0.985 0.088 0.053- 0.127

Zimbabwe 108 18.0 6.5 -0.07 0.9 48 153 0.9 0.942 0.139 0.063- 0.220

Mean 166') 11.Z 5.7 -0.32 19.3 46 III 0.31 0.966 O.oso 0.015 - 0.163 ~ of_ a S. "Voriance ofEl ... 1rlcted to be .001 due to a Heywood ..... (alao implies !hot dfz 4). ij a Total n a 6806. Multi Kurtosis - nmItivarialo _ as measured by Maldiaos nolllllllized coefficient, CF\ a Conpuative Fit Index, RMSEA - Root Mean Squaue Error of Approximatioo, CI of RMSEA - Confidence Inlerval of Root Mean Square Error of Approximatioo.


Note that the SWLS was negatively skewed in all except six countries, five of which are Asian cultures (Le., China, India, Japan, Korea, and Taiwan). The data from Lithuania showed a slightly positive skew (0.22). Although none of the nations revealed severe deviations from normality, the tendency for negative skewness is well-known among well-being researchers (e.g., Andrews & Robinson, 1991; Jacob & Brinkerhoff, 1997; Kammann & Flett, 1983), a fact that might suggest some contamination of social desirability in scales measuring global satisfaction with life. But the SWLS tends to be no more skewed than other commonly used measures in social psychology, and the skew could also represent the nature of subjective wellbeing (Cummins, 1995; Pavot & Diener, 1993; Vitters0, 1998).

Another feature of the SWLS is investigated by adding the squared factor loading for each item and dividing the sum by five. This gives an estimate of the total variance accounted for by the initial one-factor model. For 13 nations, the single factor accounted for more than 50 percent of the variance, for 26 nations between 30 and 50 percent of the variance was accounted for by one factor. In the remaining two nations, less than 30 percent of the variance was accounted for. These nations were Turkey with 27 percent and Nepal with 24 percent of the variance accounted for by a single common factor.

Overall, the results from the confirmatory factor analyses reveal that the onefactor model can account reasonably well for the relation among the SWLS items, even across a broad variety of nations. The scale seems to measure a quite homogeneous concept with moderate to high factor loadings for almost every country in the sample.

LIFE SATISFACTION AND NATIONAL WEALTH

As a basis for investigating relations between life satisfaction and wealth at the national level we first performed an ANOV A in order to estimate the proportion of the total SWLS variance due to between-nation variance. It was found that 13% of the total variance was accounted for by between-nation differences, and 87% was due to within-nation variance. Thus, variability in SWLS within nations is much higher than variability between nations. Yet, in comparison with gender, which accounted for 1.5% of the total variance, and age, which explained only 0.7% of the variance, the effect of national differences is relatively high. Moreover, it should be noted that random measurement error will contribute to variance at the within-nation level, implying that the true level of between-nation variance is slightly higher than 13%.

Based on sum-scores for the five SWLS items, a correlation analysis between national wealth and satisfaction with life was undertaken on data aggregated to the level of nations. The Pearson correlation coefficient was .42 m < .01). As was the case for the World Value Survey reported in Schyns (1998), the scatter-plot in our data also indicated two clusters of observations, one for the poor nations and one for the rich. The pattern is interpreted from Figure 2. We conducted separate analyses for poor and rich countries respectively, revealing a correlation between wealth and life satisfaction of .1 0 among the poor countries and .23 for the rich countries. None of these relations reached significance.


28

25 c

c 26

31 c

33 c

24

CI) g 22 CI) c

21 20 c

16 2~ c

41~ 18 c

6

16 c

Rsq = 0 1740

-10000 0 10000 20000 30000 40000

GNP

Figure 2. Sums cores of the Satisfaction With Life Scale plottet against Gross National Product per capita.

An important goal of this article was to explore how method effects might influence the relation between SWLS and GNP. Method effects were operationalised as reliability (considered as the mean of the five squared factor loadings within each country), and as model fit assessed by the Comparative Fit Index (which accounts for the explained covariance relative to the total covariance in the one-factor model). For example, reliability correlated with GNP (r = .39, p < .01) and with SWLS (r = .41, P > .01). Model fit, on the other hand, showed only tendencies of association with GNP and SWLS: r = .18 (n.s.) and r = .10 (n.s.) respectively.

To account for possible method effects, both the reliability measure and the model fit measure were included as independent variables along with GNP in a regression analysis with SWLS as the dependent variable. The result of this analysis showed that the association between SWLS and GNP was reduced from a significant .42 to a non-significant .25 after controlling for reliability and model fit. In particular, reliability seems to play an important role as it was the only variable with a significant regression path to the SWLS variable (Table 3").

The assumption underlying the use of aggregates to represent a national characteristic is that the aggregated variable represents another form of the construct at a higher level of analysis. Thus, to get a better understanding of cultural variations in the relationship between national wealth, scale reliability and satisfaction with life, we examined how each of the aggregated SWLS items relate to GNP. The item


raw score means as well as the factor loadings were analyzed. The latter variables reflect an item's amount of reliable satisfaction with life variance within each nation. When regressing GNP against these variables, we find the degree to which economic wealth is associated with each item's relative significance in relation to the cultural construction of life satisfaction. In comparison, the raw score mean for a single item indicates how much the total variance in that variable relates to GNP. Any difference between a raw score variable and its corresponding factor loading variable could thus be attributed to a difference in the relative significance of that particular item as a cultural construct of satisfaction with life.

Table 3. Satisfaction with life regressed upon GNP, reliability and model fit among 41 nations.

GNP Reliability Model fit

B 0.0005 7.76 10.56

S.D. 0.00 3.59 10.46

Beta .25 .36 .16

t-value 1.52 2.17* 1.0

Note * = p < .OS. GNP = Gross National Product; Reliability = Mean of the five squared factor loadmgs for the SWLS items, Model fit = Comparative Fit Index for the one-factor SWLS model.

Inspection of Table 4 reveals that all but the first raw score items were significantly related to GNP at the level of bivariate regressionlll. For the factor loading variables, mUltiple regression weights are more informative, because they illustrate more elegantly the relative difference between each item. For the factor loading variables, only the first and third items were significantly related to GNP. The second item seems to be completely unrelated to GNP, whereas the two last items have negative, but non-significant relationships.

Table 4. Bivariate and multiple regression weights for GNP as regressed upon each

Item 1 Item 2 Item 3 Item 4 Item 5

Bivariate R.S. Multiple F.L. .21 .59** .36* .05 .40** .36* .43**

.35* -.16 -.12

Note. R.S. - Raw score, F L - Factor Loading;· - p < 05; **- P < .01.

DISCUSSION

This article presented a series of confirmatory factor analyses of the Satisfaction With Life Scale based on cross-national data. At the individual level, the model fit measures for the one-factor satisfaction model were good, with goodness-of-fit indexes above .90 for all nations. On the other hand, the factor structure differs significantly across level of measurement. For the national data, the factor loadings were somewhat higher than for the individual country data, and a nested factor influencing only items I and 5 was detected. Continuing with aggregated data, we


analyzed the associatIOn between national wealth and satisfaction with life. A significant 17% of the SWLS variance was explained by the GNP, but when the cultural differences in reliability and model fit were taken into account, only 6% (n.s.) of the variance was accounted for by GNP. In other words, our study shows that people in poorer nations have less reliable responses to satisfaction with life measures, and this is an additional explanation of why, in general, wealthy nations score higher on subjective well-being scales. In general, low reliability in poor nations could be due to unstable life conditions rather than the psychometric properties of the SWLS. However, since we are analyzing concurrent data, this argument is less relevant for the present results. Moreover, lack of reliability is not the whole answer to the correlation between wealth and well-being, but rather a part of the answer that so far seems to have been overlooked in the literature.

Although people in rich nations tend to report a higher level of satisfaction with life compared with people in poor nations, there was no correlation between GNP and life satisfaction in a sub-sample of poor countries, and only a slightly higher, but still non-significant, tendency in the rich countries. Finding no significant relationship between national economy and satisfaction with life in poor countries parallels the result reported in Schyns (1998). It partly contradicts the common notions of economic threshold effects (Max-Neef, 1995), curvilinear relationships (Zolatas, 1981) and the proposal that further economic development in modem democracies will reduce the level of well-being (Lane, 2000b). Actually, what these results indicate is that overall, people in poor countries are less satisfied with their lives compared with people living in rich countries, but that factors other than economy determine the well-being of people in these countries. In rich countries, on the other hand, there is a small tendency for national wealth to be a better predictor of self reported satisfaction with life. The regression slope was found to be steeper for rich compared with poor countries although differences were not significant, and the conservative conclusion would be that they do not differ. There are some exceptions to the pattern, most notably for Japan and Hong Kong, which are rich nations with relatively low scores on well-being measures (and without which the correlation between GNP and life satisfaction in rich countries becomes larger and significant: !: = .48, Q < .05).

An important goal for the current article was to reveal whether method effects could explain some of the covariance between GNP and satisfaction with life. In particular, reliability and model fit were scrutinized, and both had a positive relationship with GNP, but only the former at a significant level. Because more stable measurement in general gives higher correlation coefficients, part of the explanation for why rich countries score higher on the Satisfaction With Life Scale is due to the more stable measurements in these countries. One reason for this finding could be that participants in richer countries are more familiar with the context/method of psychological tests. Another is that the meaning of these originally North-American items probably translates easier to other rich nations, which also are culturally closer to the USA. According to this argument, Japan is closer to the west economically, but not culturally, which might explain this country's large aberration from the regression line (cf. Figure 2). Moreover, the relationship found between the first and fifth SWLS item at the aggregated level

1. VITTERSO, E. ROYSAMB AND E. DIENER 97

could be due to different cultural practices about defining standards. Could it be that nations differ on how much the ideal standard is defined as satisfaction? On this point we are still guessing, and more research is needed to illuminate this result.

Model fit was not significantly related to national wealth, which indirectly means that the notion of homogeneity in the Satisfaction With Life Scale holds for crosscultural testing. This is good news, because the idea of cross-cultural indicators of subjective well-being has been under attack, not the least from social anthropologists. Barth (1999) for example, has argued that "anything as spare and simple as "indicators" of the quality of life cannot but do violence to that which it is designed to indicate" (p. 96). Contrary to Barth's claim, our data show that indicators of subjective quality of life, do seem to hold somewhat similar meanings across cultures.

Of course, we do not know the exact meaning of the latent construct across all the 4 I nations, but we do know that, at the individual level, a single underlying construct is able to causally explain the relation between the satisfaction with life indicators used in this study. This is not to say that cross-national differences in the interpretation of indicators do not exist. Rather to the contrary, analyzed at the cultural level our results indicate that the five SWLS items are differently related to economic wealth. Regressing GNP on the five factor loading variables revealed that the first and third items represent more significant indicators of satisfaction with life in rich nations. Relative to items I, 2 and 3, the last two items tend to be more salient for satisfaction with life items in poor countries. Given the statistical nonsignificance of the latter result, these arguments are somewhat speculative. However, due to the small sample size in the aggregated data file, it is also problematic to completely ignore these results, because we then make ourselves vulnerable to committing Type II errors. Regarding the last item of the scale, "If I could life my life over, I would change almost nothing", both the wording and the result showing a relatively low factor loading calls for a comment or two. First, the face validity for this item seems somewhat questionable. If given a choice, why would it be more positive to wish for living the same life twice rather than to live a different life. There are several reasons why one would prefer to choose a different life, even if one was satisfied with the life one has led so far. However, in a recent German study (Westerhof, et aI., 2001), a good match was found between lay theories of life satisfaction (as measured by a sentence completion task) and four of the five SWLS items, the "live my life over" item being among the confirmed items. (The only item without a good match was the first item.) The face validity for the fifth item is in other words confirmed to some extent.

Second, the complexity of the fifth item is at a different level compared to the other four. The respondent is first asked to consider an impossible option, namely to live one's life over, and then to decide whether one would like to make changes or not. In factor analysis, such differences in item complexity are unfortunate (e.g., Gorsuch, 1983). Finally, and probably related to the two former remarks, the lower factor loading for this item suggests that an improvement of the scale could be achieved by replacing item 5 with an item more coherent with the first four items of the SWLS. Nevertheless, the item's mean factor loading among 41 nations was .57,


which is not a dramatically low loading. Recall also that only three countries in the current study showed factor loadings below .40 for this item.

With respect to the zero-order correlations between national wealth and each item of the Satisfaction With Life Scale, our results show that several of the single items are just as highly correlated with GNP as the sum-scores for all five items. Comparing these results with the lower correlations found between GNP and the items as represented by their reliability scores (i.e., their factor loadings) leads us to believe that there exists some systematic variance in these items that is related to national wealth, but not to the satisfaction with life concept. For example, item 2 correlates a significant .36 with GNP at the raw score level, but only a nonsignificant .05 on the factor loading level. This result implies that although the item "The conditions of my life are excellent" relates to economic wealth in itself, the salience of the item, as an indicator of life satisfaction, is unrelated to GNP. For the first item, the situation seems to be reversed. The pure satisfaction with life part of the item is larger in rich countries than in poor, that is, the importance of the "ideal"aspect seems to increase with GNP. For the third item, both raw score and factor loading are significantly correlated with GNP, which indicates that the variance in the former is approximately the same as its relative satisfaction with life variance, and that this variance correlates with GNP.

In several studies reported in the cross-cultural literature, one-item measures are used to assess a subjective well-being concept (like happiness or satisfaction with life). Often the correlation between happiness/satisfaction and economic wealth has been higher than the one we found in the current data. Based on the previous discussion, this might indicate that these one-item measures contain systematic variance from more than one concept, and that these concepts covary with national economy. It is well-known that a measure can be conceptually over-represented (Messick, 1995). In the interest of conceptual clarity, it must be a goal of crosscultural researchers to further identify the conceptual structure of broad-spanning self-report items such as "are you happy" or "are you satisfied with your life". The present article is only a first step along this path. Another reason why the correlation between life satisfaction and GNP is somewhat lower in our data compared with other studies, might relate to the current sample. Students sometimes differ from the general population, and it might be that college students in poor countries are representatives of higher socio-economic classes compared with students in rich nations.

A critic might argue that our study is of limited generalisability because our participants have been college students. Sears (1986) argues that the overwhelming use of college students as participants in social psychological research may have contributed to a serious problem in our understanding of central elements of human nature. College students are unrepresentative of the general population in a number of important ways. They are for instance both younger and better educated, and in most countries the students come from higher social classes. Another problem relates to the reliability of the measures, which is likely to be lower in more heterogeneous study populations. Thus, student data should be cautiously interpreted. On the other hand, a series of studies within the field of subjective wellbeing have revealed that student samples give proper estimates of the general


population. For example, MacConville and Cooper (1992) found that a student sample gave a good approximation to the general population in terms of the structure of mood. Suh et aI., (1996) found that college students adapt relatively quickly to life events, a result akin to data based on a representative sample (Heady & Wearing, 1989). Moreover, the relationship between personality traits and subjective well-being reported from a student sample in Norway (Vitters0, 200 I) was basically replicated in a representative sample of the population in Northern Norway (Vitters0 and Nilsen, in press). Finally, Waller and Lea (1998) conclude that a student sample and a general population sample were generally similar with respect to their leisure motivation. After reviewing the potential pitfalls of generalizing from student populations, Visser et al. (2000) conclude that

"Social psychology can happily proceed doing most research with college sophomores, assuming that our findings generalize. And we can live with the skepticism of scholars from other disciplines who questIOn the generalisability, having documented the profound impact that context and history have on social processes" (pp. 224-225).

In other words, until our assumptions of generalisability are challenged by new data casting doubts on the arguments we have presented above, we will assume that they are replicable in representative samples as well. Nevertheless, we acknowledge that the argument may be considered somewhat weak by some (but that does not imply that it is wrong).

There are some other limitations to the results from the present study. One is the intriguing, but complex discussion of the "level of analysis" problem. On the statistical side, correlations among values using the cultural mean as the unit of analyses could change from those based on the individual as the unit. At the macro level correlations can be stronger than, weaker than, or equal to the individual level, although several researchers have recently proposed the general hypothesis that the correlations between two variables will be stronger at the aggregated level than at the individual level (Ostroff, 1993). Hence it is vital not to assume that the relationship reported on the cultural level represent the same relationship at the individual level. It is important for future research to analyze data using several different strategies. For example, one could standardize the data within each culture before cross-cultural comparison, or one could partial out each individual's scale mean scores from the items to control for response bias.

On the theoretical or meta-theoretical level one can always question whether it is appropriate to consider culture as an aggregate of individual responses. Following common ground in social anthropology, culture is something like a shared world view, and not the mean of a group of individuals. Needless to say, better theories are needed to advance on this issue.

What other implications do these results suggest for future research? We obviously need to know more about the culture-specific meaning of items intending to tap into happiness and satisfaction with life. One way to learn more is to delve deeper into the varied meaning each of the five SWLS items holds in different countries. Qualitative approaches might prove helpful in this respect. In our data we found some indication that ideal selves are more salient to the rich countries' conception of satisfaction with life, whereas past achievements seem to be more significant in poorer countries. These results however, need quantitative replications.


Finally, it would be worth giving the different regression slopes for satisfaction with life regressed on GNP further thOUght. Why does the slope seem to be steeper in rich nations compared with the poor? There are probably no simple answers, but as long as modem ideas about progress still center around economic growth, and governments worldwide seem to be captive to an ideology of growth, we should be eager to raise our knowledge to a higher standard with respect to both historical and cross-cultural variations between wealth and subjective well-being.

Endnotes

Due to a Heywood case we had to constrain the error variance of item 5, leaving us with 4 degrees of freedom

2. As both reliability and model fit are somewhat sensitive to number of observations (Jackson, 2001) we also dId a regression whIch included sample SIze, but thIS inclusion did not alter the pattern revealed in Table 3.

3 Because of high multIcollinearity, multiple regression made no sense for the raw score varIables

Joar Vittersf1. University ofTromsf1. Norway; Espen Rf1ysamb. National Institute of Public Health, Norway; Ed Diener, University of Illinois, USA

Address for Correspondence: Joar Vittersf1, Department of Psychology, University ofTromso. N-9037 Tromso. Norway. Tel: + 47 77646341 (Voice), + 47 7764 52 91 (Fax), E-mail:[email protected].

REFERENCES

Andrews, F.M and J.P. Robinson: 1991, Measures of subjective well-being, in J.P. Robinson, P.R. Shaver, and L.S. Wrightsman (eds.), Measures of Personality and SocU1l Psychological Attitudes (Academic Press, New York), pp. 61-114

Argyle, M . 1996, Subjective well-being, in A. Offer (ed), In PursUIt of the Qualrty of Life (Oxford University Press, Oxford), pp. 18-45.

Averill, J.R. and T.A. More: 2000, Happiness, In M Lewis and J HaVIland-Jones (eds ), Handbook of EmotIOns (Guilford, New York), pp. 663-676

Barth, F.: 1999, Comparing lives, FeminIst Economics 5, pp. 95-98. Bentler, P.M. and D.G. Bonett: 1980, Significance tests and goodness-of-fit In the analysis of covarIance

structures, Psychological Bulletm 88, pp. 588-606. Bentler, P.· 1995, EQS Structural Equations Program Manual (Multivariate Software, Inc, Encino, CAl. Bollen, K.A.: 1989, Structural EquatIOns With Latent VarIables (John Wiley & Sons, New York). Bollen, K A. and J.S. Long: 1993, TestIng Structural Equation Models. (Sage, London). Cantril, H.' 1967, The Pattern of Human Concern (Rutgers Umversity Press, New Jersey) Cowen, E L.: 1991, In pursuit of well ness, AmerIcan Psychologist 46, pp 404-408. Cummins, R.A. (1995). On the trail ofthe gold standard for life satisfaction. Social Indicators Research,

35,179-200 Cummins, R.A.: 1998, The second approximation to an international standard for life satisfaction, SOCial

Indicators Research 43, pp 307-334. Cummins, R.A.: 2000, Objective and subjective quality of life: An interactive model, Social Indicators

Research 52, pp. 55-72 Diener, E. & Diener, C.: 1995, The wealth of nations reVIsited. Income and quality of life, SOCial

IndIcators Research 36, pp. 275-286. Diener, E., M. Diener and C. Diener: 1995a. Factors predicting the subjectIve well-being of nations,

Journal of Personalrty and SOCIal Psychology 69, pp. 851-864


Diener, E., R.A. Emmons, R.J. Larsen and S. Gnffin' 1985, The Satisfaction With Life Scale, Journal of Personality Assessment 49. pp. 71-75

Diener, E., C.K. Napa, S. Oishi, V. Dzokoto and E M. Suh: 2000, Positivity and the construction oflife satisfaction judgements: Global happiness is not the sum of its parts. Journal of Happiness Stud,es 1, pp. 159-176.

Diener, E. and S. Oishi: 2000, Money and happiness: Income and subjective well-being aecross nations, in E. Diener and E.M Suh (eds.), Culture and SubjectIve Well-bemg (The MIT Press, London), pp 185-218.

Diener, E. and E Suh: 1997, Measuring quality of life: Economic, social, and subjective indicators, SocIal IndIcators Research 40, pp. 189-216.

Diener, E., E. Suh, H. Smith and L Shao 1995b, National differences In reported subjective well-being Why do they occur?, SocIal IndIcators Research 34. pp 7-32.

Diener, E. and E.M Suh 1999, National differences in subjective well-being, in D. Kahneman, E Diener, and N. Schwarz (eds.), Well-Bemg' The FoundatIOns of Hedonic Psychology (Russell Sage Foundation, New York), pp. 436-452.

Diener, E. and E.M. Suh' 2000, Measuring subjective well-being to compare the quahty oflife of cultures, in E Diener and E.M. Suh (eds.), Culture and Subjectrve Well-bemg (The MIT Press, London), pp. 3-12.

Diener, E., E.M. Suh, R.E. Lucas and H L. Smith: 1999, Subjective well-being: Three decades of progress, Psychological Bulletm 125, pp 276-302.

Dillon, W.R., A Kumar and N. Mulani: 1987, Offending estimates in covariance structure analysis: Comments on the causes of and solutions to Heywood cases, PsychologIcal Bulletm 101, pp. 126-135

Easterlin, RA.: 1974, Does economic growth improve the human lot?, in P A. David, and M W. Reder (eds ), NatIOns and Households m EconomIc Growth (AcademiC Press, New York), pp 89-125.

Eckersley, R.: 2000, The mixed blessing of material progress: Diminishing return In the pursuit of happiness, Journal of Happmess Stud,es I, pp. 267-292.

Eld, M.: 1997, Happiness and satisfaction: An application ofa latent state-trate model for ordinal variables, in 1. Rost and R. Langeheine (eds.), Application of Latent TraIt and Latent Class Models m the SOCIal SCIences (Waxman, MUnster), pp. 148-154.

Fernandez-Ballesteros, R 1998, Quality ofhfe: Concept and assessment, in J G. Adair, D. Belanger, and K.L. Dion (eds.), Advances m Psychologrcal Science - Congress Proceedmgs. XXVI InternatIOnal Congress of Psychology. Montreal. 1996. Vol. I (Psychological Press, Hove, UK), pp. 387-406

Floyd, F.J. and K.F. Widaman: 1995, Factor analysis in the development and refinement of clinical assessment instruments, PsychologIcal Assessment 7. pp. 286-299

Fordyce, M.W: 1988, A review of research on the happiness measures: A sixty second Index of happiness and mental health mapping, SocIal Ind,cators Research 20, pp 355-381.

Gorsuch, R L. 1983.: Factor AnalySIS. (Lawrence Erlbaum, Hillsdale, NJ). Haybron, D M : 2000, Two philosophical problems In the study of happiness, Journal of Happmess

StudIes I. pp. 207-225. Hazelrigg, L E. and M.A Hardy: 2000, Scaling the semantics of satisfaction, SOCIal Ind,cators Research

49. pp. 147-180. Headey, B and A. Wearing 1989, Personality, life events, and subjective well-being Toward a dynamic

equilibrium model, Journal of Personality and SOCIal Psychology 57, pp. 731-739. Holland, J.H: 1995, HIdden Order. How AdaptatIon BUilds CompleXIty (Perseus Books, Cambndge,

MA). Hu, L.-T and P Bentler: 1995, Evaluating model fit, in R.H Hoyle (ed.), Structural EquatIOn Modelmg.

Concepts. Issues. and ApplicatIOns (Sage, London), pp. 76-99. Hu, L.-T. and P.M. Bentler 1999, Cutoff criteria for fit indexes in covariance structure analysis.

Conventional cnteria versus new alternatives, Structural EquatIOn Modelmg 6, pp. I-55. Inglehart, R and H -D. Klingemann: 2000, Genes, culture, democracy, and happiness, in E. Diener and

E.M. Suh (eds.), Culture and SubjectIve Well-bemg (The MIT Press, London), pp. 165-183. Jackson, D.: 2001, Sample size and number of parameter estimates in Maximum Likelihood confirmatory

factor analysis: A Monte Carlo investigation, Structural EquatIOn Modelmg 8, pp. 205-223. Jacob, J.e. and M.B. Bnnkerhoff: 1997, Values, performance and subjective well-being in the

sustainability movement: An elaboration of multiple discrepancies theory, SOCIal Indicators Research 42, pp. 171-204


Kammann, R., M. Farry and P. Herbison' 1984, The analalysis and measurement of happiness as a sense of well-being, Social Indicators Research 15, pp. 91-115.

Kammann, R and R. Flett: 1983, Sourcebook for Measurmg Well-Bemg with Affectometer 2 (Why not? Foundation, Dunedin, New Zealand).

Lakatos, I.: 1999, Lectures on scientific method, In M. Motterlini (ed.), For and Agamst Method. Includmg Lacatos's Lectures on SCientific Method and the Lakatos-Feyerabend Correspondence (The University of Chicago Press, Chicago), pp 19-109.

Lane, R.E.: 1996, Quality of life and quality of persons' A new role for government?, in A. Offer (ed), In Pursuit of the Quality of Life (Oxford University Press, Oxford), pp. 256-293.

Lane, R E.: 2000a, Diminishing returns to mcome, companIOnship and happmess, Journal ofHappmess Studies I, pp. 103-119.

Lane, R.E.: 2000b, The Loss of Happmess m Market DemocraCies (Yale University Press, London). Larsen, l.R. and B L. Fredrickson: 1999, Measurement Issues in emotion research, in D. Kahneman, E.

Diener, and N. Schwarz (eds.), Well-Bemg' The FoundatIOns of Hedonic Psychology (Russell Sage Foundation, New York), pp. 40-60.

Lerner, S.: 1997, Indicators of human welI-being: Fine-tuning vs takmg action?, Social Indicators Research 40, pp 217-220.

Lyubomirsky, S. and H.S. Lepper: 1999, A measure ofsublective happiness. Preliminary reliability and construct validation, Social Indicators Research 46, pp. 137-155

MacCallum, R.e., M.W. Browne and H.M Sugawara: 1996, Power analysIs and determination of sample sIze for covariance structure modehng, Psychological Methods I, pp. 130-149.

Marsh, H W, lR Balla and K.-T. Hau: 1996, An evaluation of incremental fit indices: A clarification of mathematical and empincal properties, m G.A Marcouhdes and R.E. Schumacker (eds.), Advanced Structural Equation Modelmg. Issues and Techmques (Lawrence Erlbaum, Mahwah, NJ), pp. 315-353.

Max-Neef, M : 1995, Economic growth and quality ofhfe A threshold hypothesis, Ecological EconomiCs 15, pp 115-118.

Mayr, E.: 1982, The Growth of BIOlogical Thought Diversity, Evolutton, and Inheritance (Harvard University Press, Cambridge).

Mazumdar, K.: 2000, Causal flow between human well-being and per capita real gross domestic product, Social Indicators Research 50, pp 297-313.

McConville, C. and C. Cooper: 1992, The structure of mood, Personality and IndiVidual Difforences 8, pp 909-919.

McDonald, R.P.: 1999, Test Theory A Unified Treatment (Lawrence Erlbaum, London). Mcintosh, e.N : 2001, Report on the construct vahdity of the temporal satisfaction with life scale, SOCial

Indicators Research 54, pp. 37-56. Messick, S.: 1995, Validity of psychologIcal assessment. Validation of inferences from persons' responses

and performances as scientific inquiry mto score meaning, American Psychologist 50, pp. 741-749. Mourn, T.: 1992, Needs, rights and indicators on quahty of life research, in T Mourn (ed.), Quality of Life

and Health m the General PopulatIOn - Phtlosophy, Methods and Empmcallssues (Unpublished doctoral dissertation, Oslo).

Mulaik, S A 1987, A brief history of the philosophical foundations of exploratory factor analysis, Multivariate BehaVioral Research 22, pp 267-305.

Mulaik, S A and L.R James: 1995, Objectivity and reasoning m sCience and structural equation modehng, in R.H. Hoyle (ed.), Structural EquatIOn Modelmg. Concepts, Issues, and AppilcatlOns (Sage, London), pp 118-137.

Olson, G.1. and B.1. Schober: 1993, The satisfied poor. Development of an interventlOn-onented theoretical framework to explain satisfaction with a life in poverty, Social Indicators Research 28, pp. 173-193.

Ostroff. C.: 1993, Comparmg correlatIOns based on individual-level and aggregated data, Journal of Applied Psychology 78, pp. 569-582.

Ouweneel, P. and R. Veenhoven' 1991, Cross-national differences in happmess Cultural bias or social quahty, in N. Bleichrod and P.J.D. Drenth (eds ), Contemporary Issues m Cross-Cultural Psychology (Swets and Zeitlinger, Amsterdam), pp 168-184

Pavor, W. and E. Diener' 1993, Review ofthe SatisfactIOn With Life Scale, PsycholOgical Assessment 5, pp.42-55.

Pavot, W , E. Diener and E. Suh' 1998, The Temporal Satisfaction With Life Scale, Journal of Personality Assessment 70, pp 340-354

J. VITTERSO, E. ROYSAMB AND E. DIENER 103

Pelletier, L.G., L.R. Legault and K.M. Tuson. 1996, The envIronmental satisfactIOn scale A measure of satisfaction with local envIronmental conditions and government environmetal policIes, Environment and BehavIOr 28, pp. 5-26

Raphael, D.: 1996, Defining quality of life. Eleven debates concerning Its measurement, in R. Renwick, I Brown, and M Nagler (eds.), Quality of Life m Health PromotIOn and RehabdilatlOn Conceptual Approaches, Issues, and ApplicatIOns (Sage, London), pp. 146-165.

Rmgen, S .. 1995, Well-being, measurement and preferences, Acta SoclOlogica 38, pp 3- I 5 Sandvik, E., E. Diener and L. Seldlitz: 1993, SubJective well-being: The convergence and stability of

self-report and non-self-report measures, Journal of Personality 61, pp. 317-342. Schwarz, N and F. Strack: 1999, Reports of subjectIve well-being: Judgemental processes and their

methodological implicatIOns, m D. Kahneman, E. Diener, and N Schwarz (eds.), Well-Bemg' The FoundatIOns of Hedonic Psychology (Russell Sage Foundation, New York), pp. 61-84

Schyns, P . 1998, CrossnatlOnal dIfferences m happiness' EconomIc and cultural factors explored, Socwl Indicators Research 43, pp 3-26.

Sears, DO.: 1986, College sophomores in the laboratory: Influences ofa narrow data base on social psychology's view of human nature, Journal of Personality and Social Psychology 51, pp. 515-530.

Segall, M H., P R Dasen, J W. Berry and Y H. Poortinga: 1990, Human BehaVIOr m Global Perspective: An IntroductIOn to Cross-cultural Psychology (Pergamon, New York)

Shweder, R.A. and 1. Haldt: 2000, The cultural psychology of the emotIOns: Ancient and new, in M Lewis and J.M. Haviland-Jones (eds), Handbook of EmotIOns (The GUIlford Press, London), pp. 397-414

Steiger, 1.H. and J.M. Lind. 1980, Statistically based tests for the number of common factors. Paper presented at the annual meeting of the Psychometric SocIety, Iowa City, June 1980

Suh, E , E. DIener and E. Fujita: 1996, Events and subjective well-being' Only recent events matter. Journal of Personality and Social Psychology 70, pp 1091- I 102

Suh, E., E. Diener, S. Oishi and H.C. Triandis: 1998, The shifting basis oflife satisfaction judgements across cultures: EmotIOns versus norms, Journal of Personality and Socwl Psychology 74, pp.482-493.

Veenhoven, R . 1989, National wealth and individual happiness, in K G. Grunert and F Olander (cds.), Understandmg Economic Behavror (Kluwer, Dorsrecht, NL) pp 9-32

Veenhoven, R 1996, Developments in satisfaction-research, SocwlIndlcators Research 37, pp. 1-46 Veenhoven, R.: 2000, The four qualities of life Ordering concepts and measures of the good life, Journal

of Happmess Studies I. pp 1-39. Visser, P.S., JA Krosnick and PJ. Lavrakas. 2000, Survey research, m H T. Reis and CJ. Judd (eds.),

Handbook of Research Methods m Social and Personality Psychology (Cambndge Umversity Press, Cambndge, UK). pp. 223-252

Vitters0, J . 200 I. Personality traits and subjecllve well-being' Emotional stability, not extraverSIOn, IS probably the important predictor, Personality and IndiVidual Differences 31, pp. 903-914

Vitters0. J. and F. NIlsen. in press. The conceptual and relational structure of subjective well-being, neurollcism. and extraversion: Neuroticism IS again found to be the Important predictor of happiness, Social Indicators Research.

Waller, J. and S E.G Lea' 1999, Seeking the real Spam? Autenticlty In motIvation. Annals of TOUrISm Research 26, pp 110-129

Watson, D 2000, Mood and Temperament (The Guilford Press, London). Watten, R G .. 0 Vassend, J.-L Syversen and T. Myhrer 1995, Personality and quality of life. Socwl

Indicators Research 35, pp. 289-302. Westerhof, GJ .. F. Dittmann-Kohli and T. Thissen: 2001, Beyond life satisfaction: Lay conceptIOns of

well-being among middle-aged and elderly adults, Socwl Indicators Research 56. pp 179-203. WHOQOL: 1998. Development ofthe World Health Organization WHOQOL-BREF Quality of Life

Assessment. Psychological Medlcme 28. pp 551-558. Wright, N.D. and V. Larsen' 1993, Materialism and the life satisfaction. A meta analysis, Journal of

Consumer SatisfactIOn, DlssalisfactlOn and C omp/amIng BehaVIOr 6, pp. 158-165. Zolatas, X.: 1981, Economic Growth and DeclInmg Socw/ Welfare (New York UniverSIty Press, New

York).

EDWARD HELMES AND LYNNE AUSTIN

QUALITY OF LIFE IN RESIDENTIAL CARE

Abstract. While some of the changes in life that accompany increasing age may promote higher quality of life, the increased prevalence of disease and other negative hfe events presumably operate in the opposite direction. Among the most salient negative life events 10 later life IS the development of physical or mental disability sufficiently severe to Justl!y moving from an independent life in the community to a residential care facility. This chapter first reviews the current state of knowledge concerning quality of life with increasing age. Most studies note changes, but the distinction of age effects from cohort effects is not often made Illness in later life has a major impact upon quality of life, especially if disability is one outcome. As illness and disability often lead to a move into a residentlal care facility. the quality of life in residential care becomes of broader interest. Such interest is enhanced by the strong contrast in vIews expressed by people when contemplating life in reSidential care and when actually experiencing it Various models of this apparent "disability paradox" are described, followed by an analysis of the conceptual difficulties underlying research into quality of life because of the varied definitions in current use.

The final decades of life can be a period of great satisfaction and fulfilment for many older people (Connidis, 1987; Keller, Leventhal, & Larson, 1989). Whether this state is referred to as life satisfaction (Deiner, Emmons, Larsen & Griffin, 1985) or quality of life, older people who experience such fulfilment may well feel that they are experiencing a high or even optimal level of well-being and quality of life (e.g., Jones, 1988). At the same time, the increased risks of physical illness and major social change can bring substantial changes in later life that negatively affect quality of life. One example is how patterns of relationships with family and friends can change due to deaths and relocations. Another example is the manner in which individuals whose lives centred around work patterns and relationships can experience m!Uor difficulties in adjusting to retirement. If an older person experiences illness of any magnitude or duration, these alterations in physical health can also have a major negative effect upon quality of life. Only a minority of older people, between 5 and to percent of people in most Western countries, experience cognitive and physical impairments of sufficient severity to warrant a move to some form of residential care. As a consequence of this move, they would appear to experience multiple changes in their life of such magnitude that the quality of life that they experience could only suffer. Indeed, this view of life in residential care as predominantly negative was dominant for many years (Cohn & Sugar, 1991). Cohn and Sugar report that it was only in the mid-1980s that direct research on quality of life in residential care facilities began to appear.

We begin with a review of some of the relevant literature on quality of life in the later years of life and life in residential care facilities in order to establish the current state of research. We then point out some of the conceptual and practical difficulties that are involved in the measurement of quality of life in this population. Many of these difficulties are a function of the lack of consensus as to the meaning of quality

105

E. Gullone and RA. Cummins (eds.), The Universality of Subjective Wellbeing Indicators, 105-128. © 2002 K/uwer Academic Publishers.


of life and the rapid expansion in the number of measures of quality of life. The final section then summarizes some of the current thinking on adaptation to the circumstances of life in residential care. The question of what happens to quality of life as people become older should be addressed before determining what changes with a move into residential care. In the absence of any major change in health or interpersonal relationships, it is likely that some aspects of quality of life remain stable over time in older adults. Grundy and Bowling (1999) argue that psychological well-being, physical functioning, interpersonal relationships, health and social activity are the areas of life that are the most important to older individuals and their perceptions of their lives. Maintaining their independence was central to the life satisfaction of their participants. Financial status, standard of living and housing were of less concern to their older respondents in assessing the quality of their lives. Grundy and Bowling conducted a longitudinal study of 620 of the oldest old (those aged 85 and over) who were initially living in the community in an economically-depressed area of inner London. While the distribution of "good" and "poor" scores was unchanged at three-year follow-up, all scores declined. That is, as people became older their life satisfaction declined. At baseline, almost half of the sample expressed dissatisfaction with their lives. The group that reported poor life satisfaction at baseline were more likely to have smaller social networks, have experienced more severe life events, feel old and be lonely most of the time. Very few of this group reported feeling young, whereas over half of those who were satisfied or very satisfied with their lives did feel young. The factors that differentiated those with "good" and "poor" quality of life were: level of activity, level of disability and problems with health. Similarly, from a sample of 10,263 Canadian elders living both in the community and in residential care Clarke, Marshall, Ryff and Rosenthal (2000) found that as their subjects aged, they reported less environmental mastery, personal growth, purpose in life and positive relations with others. Asakawa, Koyano, Ando and Shibata (2000) examined the effect of functional decline of aspects of the quality of life of 692 Japanese elderly over two years. They found that, after controlling for age, gender and socio-economic status, changes in functional health were related to changes in quality of life measures. Kunzmann, Little, and Smith (2000) used the Positive and Negative Affect Scale (PANAS; Watson, Clark, & Tellegen, 1988), which provides measures of one domain of quality of life under a broadly defined conceptualization. Their results also highlighted the importance of physical health as a mediator of quality of life in normal aging. Over a four-year period, they found that in those older people whose physical health remained good, positive affect increased while negative affect decreased. If health status was not controlled, then the association of positive and negative affect with age reversed. Krach, DeVaney, DeTurk and Zink (1996) investigated the way in which the oldest-old adapt to old age. They said "In terms of physical functioning, our subjects had come to terms with the expectation of physical problems with a matter-of-factness. There was no sense of hopelessness in their acceptance and no instances of panic about physical deterioration." (p. 458). While 33% of this sample stated that they were limited a great deal by health problems, only 4% rated their health as poor. Similarly, George and Clipp (1991)

MJ. POWER, M. BULLINGER AND THE WHOQOL GROUP 107

suggested that older people express satisfaction about conditions that produce dissatisfaction in people from younger age groups. Mercier, Peladeau and Tempier (1998) also found that older people with mental illnesses expressed greater life satisfaction than did younger people suffering from similar conditions. While these findings may be the result of a process of adaptation, it may equally reflect a cohort effect among people whose life experience includes a major economic depression and two world wars.

HEALTH AND AGE

The association between illness and disability and quality of life has been well documented, with the association being weaker for objective measures of health than for subjective ratings (Kozma, Stones, & McNeil, 1991). The evaluation of the older person's quality of life is also likely to show discrepancies in the extent of agreement between self report and reports by proxies, be they family or medical staff. Estimates provided by nurses and physicians have been found to correlate poorly with an older person's evaluation of their quality of life (Berlowitz, Du, Kazis & Lewis, 1995).

More recent studies by Kendig, Browning & Young (2000) and Romney, Jenkins & Bynner (1992) provide further evidence that it is the symptoms of disability that accompany illness that reduce morale and quality of life for older people in the community. Such evidence is among the more recent literature that argues against the older interpretation that increased physical symptoms such as pain are a psychosomatic consequence of low morale and depression (Blumer & Heilbronn, 1982).

While the extensive literature on quality of life and physical illness clearly demonstrates the strong relationship among illness, pain, and disability and quality of life, other domains of quality of life have also been investigated. Pearlman and Uhlmann (l988a) examined the attributes and events associated with the subjective quality of life of 126 community dwelling or hospitalised elderly people suffering from chronic diseases such as arthritis, coronary disease, pulmonary disorders, diabetes and cancer. When they were asked to list the factors that had a positive impact on their quality of life, people cited the effects of medical care or surgery, their interpersonal relationships and improvements in housing. Health, functional impairment, deterioration in interpersonal relationships and pain were all cited as having had a negative impact on the subjects' quality of life. In addition, measures of quality of life were significantly related to self-rated and comparative health, social relationships and perceptions of financial status. Self-esteem and meaning in life have also been identified as important factors (O'Connor & Vallerand, 1998), as have negative life events, loneliness, and low levels of activity (Osberg, McGinnis, Dejong, & Seward, 1987; Grundy & Bowling, 1999), and migrating as an adult (Berdes & Zych, 2000).

In an extension of their earlier study, Pearlman and Uhlmann (1991) examined quality of life in a sample of 258 American people aged 65 and over. Only 11% rated the quality oflife as "not so good" or "not good at all, poor" or "terrible". This is significant because all of the participants were recruited through doctor's records. That is, this sample had frequent medical contact and, indeed, 72.9% were limited


Quality of Life and Increasing Age of the respondents rated their lives as being "about as good as it could possibly be". Again, perceived health status was consistently associated with quality of life. On the other hand, in Grundy and Bowling's (1999) sample of people aged over 85 who were living in poverty, almost half expressed dissatisfaction with their lives.

Another study of older individuals living in the community found that functional impairment and level of activity were the only significant predictors of subjective quality of life. Married men reported higher quality of life than their single counterparts but married women did not (Osberg et aI., 1987). Fry (2000) conducted qualitative interviews with 331 people aged between 60 and 85 to ascertain the views and preferences regarding issues such as their social, financial and legal situations, work, safety and health. Factor analysis of the responses revealed four factors important to the participants. These were I) capital guarantees and assurances of protection from the social, economic and legal systems such as guarantees of assisted suicide and guarantees of protection from age discrimination and harassment; 2) aspirations and expectations for future quality of life such as a life with respect and dignity, economic independence and challenging activities; 3) fears and concerns about future life relating to dying alone, poverty, disability, elder abuse and family conflict and 4) external and environmental factors that threaten to affect quality of life. These were divided into factors over which the person has no control such as physical isolation, crime and lack of transport and those the person can control such as lack of leisure activities and interference in decision-making of adUlt children (Fry, 2000). Fry summarises the significant factors which emerged from the study as "I) a strong desire to be autonomous and independent in pursuing their chosen life-style; 2) a desire to be remembered, appreciated, and to be treated with respect and dignity by younger adults; 3) a right to empowerment as a function of being recognized for their achievements and contributions to society; and 4) a need to voice their fears and anxieties about future quality of life." (p. 373). From a sample of 10,263 Canadian elders living both in the community and in residential care, Clarke, Marshall, Ryff and Rosenthal (2000) found that as their subjects aged, they reported less environmental mastery, personal growth, purpose in life and positive relations with others. However, these variables contributed less than 2% to the variance in well-being.

It is clear from this brief review that health concerns are among the major determinants of quality of life among older individuals living in the community or during hospitalized phases of acute care. There is also an extensive literature upon individuals with long-standing disability, again with the great majority of such individuals residing in the community. In both these cases, there is a clear assumption that there is a continuum of quality of life that extends from full independence and health through stages of illness and disability. There does not appear to be a substantial argument in favour of their being a qualitative difference in these states. In a study of the quality of life of the oldest-old in Italy, Urciuoli, DelloBuono, Padoani and De Leo (1998) address this issue to some extent. They found little difference in the health status of those living in the community and those resident in nursing homes. The latter group were found to be more functionally impaired and required more assistance with activities of daily living. This was

M.J. POWER, M. BULLINGER AND THE WHOQOLGROUP 109

reflected in lower quality of life scores in the domain of self-care. However, nursing home residents did not differ significantly from their peers in the community in other domains such as depression and anxiety, social functions, sexual functions, life satisfaction or trust in God.

QUALITY OF LIFE IN RESIDENTIAL CARE

This section addresses the issue of whether the transition from life in the community represents a quantitative shift along dimensions of quality of life, or if it is more of a qualitative change. Some of the factors that are relevant to quality of life in residential care are discussed in this context. The section then concludes with a review of current data that have been published on quality of life in residential care facilities.

There are reasons for believing that the nature of quality life for residents of care facilities may be qualitatively different from that of older people who continue to live in their own homes in the community. If there is not a qualitative difference, there can be a substantial quantitative difference in quality of life. Such a view is certainly held by many older people themselves. At least one study suggests that the move to residential care is often dreaded more by older individuals who are living independently than is death itself(Loebel, Loebel, Dager, Centerwall & Reay, 1991), while caregivers of older people with dementia may share the same view of life in nursing homes (Karlawish, Klocinski, Merz, Clark & Asch, 2000).

A more objective indication is that many aspects of life clearly change when a person moves into residential care. There is the loss of the familiar home and, in many cases, an accompanying loss of the familiar neighbourhood. Any move to a smaller residence, such as the limited living space in a residential care facility, leads to a loss of familiar belongings, many of which hold substantial sentimental value to their owners. Life in residential care facilities frequently implies increased structure and regulation of life for the residents. This increase in external control can also be accompanied by increased noise and disruption to daily routines, providing an overall loss of control over many aspects of life. Rodin (1986) pointed out the importance of a sense of control in maintaining health. Ronnberg (1998) suggests that nursing homes often fail to provide a psychologically satisfying environment for residents. Lack of privacy and control and isolation from the community, coupled with physical impairment, can lead to learned helplessness and depression. Clark and Bowling (1990) investigated the quality of everyday life in long stay hospitals and nursing homes. In one nursing home, observers documented that 18% of patients' requests were totally ignored, usually because there was no staff member present when the request was made. In the geriatric ward of a hospital, staff were observed to be insensitive to the residents' needs. Staff nurses, nurse aides and domestic staff were all observed to be impatient, bad-tempered or simply rude. For instance, at meal times the domestic staff often removed plates before the resident had actually finished their meals. Such staff behaviour was not evident in either of the nursing homes taking part in the study.

Ferrell, Ferrell and Osterweil (1990) reported on the prevalence of pain in American nursing homes; pain being a significant health factor influencing quality of


life. Over 70% of their sample reported having constant or intermittent pain which, in over 75% of these cases, was substantial enough to effect participation in enjoyable activities, ambulation, posture and sleep. While pain was moderately correlated with the ability to participate in enjoyable activities (r = .50), depression was not significantly correlated with pain (p. 412).

Other forms of disruption beyond physical pain are also found in residential care. Schnelle, Alessi, AI-Samarrai, Fricker and Ouslander (1999) examined the effects of noise and light on sleep patterns of nursing home residents. Nursing home routines usually involve two or more nursing "rounds" in which nurses administer medication, check whether beds require linen changes due to incontinence of the residents or assist residents to the toilet. Nurses usually take with them a trolley full of clean linen as well as a skip for dirty linen. These routines mean that room lights are turned on and nurses converse with residents or with each other. In addition, residents often call for attention by use of buzzers that may be audible all over the nursing home. The noise of trolleys, footsteps, buzzers and voices can all be disturbing to light sleepers. During a baseline measurement, these researchers found over 70 instances in each of two nursing homes when the noise levels exceeded 50db. Lights were turned on and off between 4 and 6 times per night. Even an intervention that managed to reduce both noise and light levels could not reduce these intrusions enough to improve sleep patterns among the residents.

Kruzich, Clinton & Kelber (1992) examined the relationship between organisational variables and nursing home satisfaction. The results indicate that residents in facilities rated as attractive and clean were more satisfied, as were those living in facilities with lower levels of staff turnover. Another important contribution to satisfaction was the rigidity of the nursing home routines. Residents who had more of their personal possessions around them and had more choice in the structure of their day were more satisfied than residents who did not. However, organisational variables accounted for only II % of resident satisfaction. Resident variables, such as length of anticipated stay, social support, ability to self-care and self-rated health accounted for 17% of the variance. Additionally, the contribution of these variables differed according to the care needs of the residents. For instance, the rigidity of nursing home routines did not impact on the satisfaction of residents with high care needs while there was a significant relationship between these variables in residents with low care needs.

With these various negative characteristics, it might therefore be expected that the loss of control over so many aspects of life in residential care and other rather negative elements of life there would lead to further quantitative declines or qualitative shifts in perceived quality oflife.

ACTUAL QUALITY OF LIFE IN RESIDENTIAL CARE

The data on quality of life in nursing homes suggests a rather different picture. Pearlman and Uhlmann (1988b) examined the quality of life of elderly people living in a nursing home. Thirty-six percent reported the perception that their lives were "about as good as it could possibly be", comparable with the 37% of elderly people living in the community who gave the same rating. Just over 19% of the nursing

M.J. POWER, M. BULLINGER AND THE WHOQOL GROUP III

home group, however, rated their quality oflife as poor, as compared to 11% of the community sample. In response to open ended questions, respondents cited moving to a nursing home as having both a positive and a negative impact. Positive factors included access to care, easier access to religious services (because clergymen visit) and social activities. On the negative side, residents cited the lack of privacy, cost of care, the standard of meals, loss of independence and having to leave their pets as having a detrimental impact on their perception of their lives. Overall, there were few differences in quality of life between those who lived in the community and those living in a nursing home. Only personal relationships and place of residence were said to limit quality of life. Nonetheless, nursing home residents did not differ significantly from people living in their own homes or with family in terms of their global assessment of quality of life. For both groups, a perception of health was a major correlate of quality of life. As expected, nursing home residents were more physically impaired than their peers living in the community.

In addition, while not true of all forms of residential care, in many nursing homes there is a large number of individuals with dementia, with whom effective communication is very difficult. Do these various changes in the physical and social environment result in a qualitative change in the nature of quality of life or do they represent different points on a dimension or dimensions of quality of life?

The answer to this question very much depends upon how one defines quality of life and how quality of life is actually measured. These definitional issues must be considered, as it is clear that some characteristics of individuals may change during the course of their adaptation to life in residential care. Some of these attributes may undergo adaptation and return to levels observed in independent life prior to the move to residential care. Other attributes may change to new levels and reflect a different state from that observable in the community. Issues of quality of life in residential care therefore very much depend upon what aspect of quality of life one is discussing. It is also important to consider the source of the information on quality of life. Cohn and Sugar (1991) demonstrated substantial differences in the perceived importance of different aspects of life in residential care by the residents, their caregivers and by their families. Residents in this study rank ordered the importance to them of issues related to autonomy. ]n order from most to least important, the issues were: access to telephone, a place to be alone, availability of transportation, choice of roommates, having their keepsakes nearby; setting their own schedules; the frequency of baths; choice of foods; choice of company at meals, choice of where to eat and being able to decorate their own rooms. Thirty-eight percent of the sample commented on the importance of morale, which was conceptualized as "the residents' enthusiasm, identification with the institution and positive and negative views that may affect their lives and how they address issues and problems" (p. 35). When asked what they most missed about their former lives they mentioned socialemotional factors or activities most often. Cooking, entertaining, shopping and holidays were all rated as among the most important aspects of life that were no longer available to them. Independence and autonomy were also missed by a large number of residents. However 16% of residents said that they missed nothing about their previous lifestyle. In comparing the perceptions of residents, staff and family, Cohn and Sugar (1991) found that residents were less concerned about changes in


professional care than were the other respondents, who considered it to be very important.

Such results reflect the complexities of definition and meaning in the field of quality of life. However, before dealing with the thorny defmitional issues related to quality of life, we shall consider the question of whether the transition from life in the community to life in residential care constitutes a quantitative or a qualitative change in quality of life. The issue of whether change is qualitative or quantitative is important at both the theoretical and pragmatic levels. If a change is qualitative, then a major fundamental shift in the internal state of a person must be involved, as opposed to a shift in level with quantitative change. At the pragmatic level, qualitative change implies a shift in type or category, a process that implies the use of different measurement instruments for different relevant constructs. Quantitative change implies the use of the same measurement instruments, as long as the scale of the change is within the range of possible scores on the instrument. The transition into residential care provides a context to explore the issue in more detail.

UNDERSTANDING TRANSITIONS AS QUANTITATIVE OR QUALITATIVE

CHANGE: THE DISABILITY PARADOX

If there is a qualitative shift in perspective regarding quality of life as people move from life in the community to residential care facilities, such that the objective differences in the quality of the two circumstances come to be minimized psychologically, then some process must be active. The basic effect underlying this process of minimization has been called the "disability paradox". This effect was initially observed in cases of disability, in which subjective measurements of quality of life are often equivalent to those in non-impaired samples. For instance, Bretscher, Rummans, Sloan, Kaub, Bartlett, Borkenhagen and Loprinzi (1999) studied quality of life in hospice patients in the end-stage of terminal disease. Contrary to the general perception of the effect of hospitalisation and serious terminal illness, these patients consistently rated their quality of life at an average of 75%. This appraisal was stable from the time of admission to the study until they were no long able to respond to questions. Like Bretscher et al.'s dying subjects, Albrecht and Devlieger (1999) found that their disabled participants perceived their lives to be good. Indeed, 54.3% of their respondents, who had moderate to serious disabilities, reported their quality of life to be excellent or good. Other researchers have reported similar findings of unexpectedly high life satisfaction among people with severe physical ailments (Breevelt & Van Dam, 1991; Andrykowski, Brady & Hunt, 1993). Cummins, Gullone, and Lau (this volume) discuss the general phenomenon of high levels of perceived quality of life across a variety of populations, including ill and disabled samples, and across a variety of quality of life measures, in more depth.

It is within the context of comparatively stable periods of time that Cummins et al. discuss at length the overall level of scores on measures of quality of life. The measurement of quality of life is commonly assumed to be under stable conditions of the respondent's life. In the case of many disabilities, both developmental and acquired, there is often a long period of transition, and in the case of permanent

MJ. POWER, M. BULLINGER AND THE WHOQOLGROUP 113

disabilities, a long period of time in which there may be little or no change in the overall condition. In such cases there is also the element of stability and consequently most measurements of quality of life have both appropriate instructions to respondents and time frame for use in framing their responses.

For individuals with acquired disabilities, there are thus two issues. One is the degree of physical impairment and its concomitant associations. The second is the stability of the condition over time and how recent any major changes have been to the time of measurement of quality of life. As Koch (2000) points out, physical disability, in and of itself, is not a barrier to high quality of life. However for many people such conditions carry ramifications beyond the physical. If illness or disability prevents a person from working then there may be financial restrictions. These, together with problems of mobility, can limit social or recreational opportunities, all of which are considered to be important domains of quality of life (Albrecht & Devlieger, 2000). One perspective on the disability paradox deals with the process by which individuals undergo change while adapting to different life circumstances.

It is not only in the assessment of quality of life that this phenomenon of the adaptation of perspectives can be observed. Breevelt and Van Dam (1991) cite several studies which found cancer patients, despite reporting higher levels of physical complaints such as nausea and fatigue, did not score higher than healthy control groups on measures of depression, anxiety and other psychological variables. Cassileth, Luck, Strouse, Miller, Brown, Cross and Tenaglia (1984) also found no significant differences between groups of patients suffering from a range of conditions such as arthritis and diabetes and the general population on measures of anxiety, depression, general positive affect, emotionality and loss of behavioural or emotional control. This study did, however, find that patients who had only recently been diagnosed reported greater anxiety, depression and loss of control than did those who had been diagnosed for some time.

Schwartz and Sprangers (1999) discuss the follow-up of a study of people suffering from multiple sclerosis. In spite of a significant reduction in neuropsychological functioning, and increased neurological impairment, one group of participants reported fewer psychosocial role limitations, decreased reliance on negative coping styles and an improved sense of well-being than they had two years earlier. The other group in this study reported no change on these variables. Various theories have been proposed to deal with the process of adaptation and to explain why some people adapt well and maintain high levels of well-being and quality of life while others do not.

THEORIES OF THE DISABILITY PARADOX

There are several theories that have been brought to bear to explain the phenomenon of the disability paradox, ranging from psychoanalytic to social-cognitive. All involve the process of adaptation in some form. Crisis theory suggests that a highly adverse event, such as the diagnosis of a serious illness, is initially perceived as too great for the person's problem-solving abilities to deal with. With time, the person is able to increase, or improve the efficiency of, their problem-solving capacity so that


the event is no longer overwhelming (Breevelt & Van Dam, 1991). Cassileth et aJ.'s (1984) findings suggest that the group who had been living with their diagnosis for some time had been able to adapt to their circumstances. This implies that a person's psychological functioning is disrupted by a critical event, such as diagnosis of a serious illness, but that, with the passage of time, many people are able to return to a positive evaluation of their lives.

The psychoanalytic theory of defence mechanisms has been used to explain some of the emotional aspects of adaptation. Bahnson and Bahnson (J 996) found that cancer patients who employ defence mechanisms report lower levels of psychological distress than those who do not.

Other theories provide perspectives upon the processes involved in adaptation to new life circumstances. Judgement theories such as social comparison and adaptation level theories have also been suggested as playing a part in this process of adjustment. They have both been found to be related to well-being among people with serious medical conditions (Breevelt & Van Dam, 1991). Adaptation level theory suggests that we compare our current experience with previous ones. It presumes that we maintain a balance, a sense that our current situation is somewhere in the middle of a continuum of experience. After an extreme event, our conception of the "lower" extreme of this continuum changes. While our idea of what is "normal" is still at the midpoint, the midpoint itself has shifted towards the newlyconceived lowest extreme (Breevelt & Van Dam, 1991). One adaptation theory will be discussed in greater detail later in this paper.

Social comparison theory suggests that we compare ourselves to others in the same circumstances in order to judge our own situation (Breevelt & Van Dam, 1991). According to Gibbons (1999), we compare ourselves to others because it gives us information about where we stand, socially and personally. When a critical event such as a diagnosis of serious illness occurs we are thrown into a new situation and a new identity. We compare ourselves to others in the same situation in order to assess our progress and to gain information about what to expect. In a review of the literature on the topic, Gibbons concludes that people with serious illnesses such as cancer and rheumatoid arthritis engage in more social comparison than those without such diseases. People may either choose a downward comparison, by comparing themselves with someone who is worse off than themselves or an upward comparison by choosing a person who is doing better than they are. Downward comparison can be helpful to people who are feeling threatened by their situation. Upward comparison, however can be informative and inspiring for those who are not experiencing high levels of threat (Gibbons, 1999).

RESPONSE SHIFT

Another approach to this phenomenon is the theory of response shift (Sprangers & Schwartz, 1999). Response shift is defined as:

"a change in the meamng of one's self-evaluation ofa target construct as a result of (a) a change in the respondent's internal standards of measurement (I e. scale recalibration); (b) a change in the respondent's values (i e. the Importance of component

M.J. POWER, M. BULLINGER AND THE WHOQOL GROUP

domams constituting the target construct) or (c) a redefinition of the target construct (i.e. reconceptuaiization)" (Sprangers & Schwartz, 1999, p 1532).

115

Scale recalibration occurs when a person changes their own internal standard of measurement of some aspect of their lives. According to Jansen, Stigglebout, Nooij, Noordijk and Kievit (2000), people decide what various points on a scale represent based on their life experiences and understanding. As we have new experiences we may change our appreciation of those points. A simple example of scale recalibration would occur if a person who had always lived in the tropics were to spend a winter in Europe. Before the move they might consider a temperature of 15° C to be chilly; however, after living in a colder climate, they may regard this temperature as mild. In the same way, a person who has suffered a serious illness might still regard his or her quality of life as satisfactory because his or her idea of what is acceptable has changed as a result of the experience. A similar, if not identical process appears to occur in the adaptation level model of change. Whether the process is one of change in the evaluation of the current state or a change in the mid-point of the evaluation scale will not be easy to determine empirically.

Change in values is the second aspect of response shift. For instance, a person may see the conditions necessary for a good quality of life as good physical and psychological health and good relationships. If an event or disease robs them of their physical health they may come to value it less, instead placing more emphasis on the other aspects. They do not change their conceptualisation of quality of life but they reconsider the importance of the various components (Jansen et a\., 2000). A quote from one of Albrecht and Devlieger's (1999) study participants illustrates this change in values.

"Then, multiple sclerosis hit me like a Iightemng bolt - without warning I went into a deep depression and was about to give up hope and [ looked at my husband and kids and said: Get your head on straight. What's important here? My values were clarified by the shock The big picture came into focus. My MS wasn't as important as my family and I wasn't going to let it ruin our lives" (p. 983).

The person quoted above changed the values attributed to family and disease. A further refinement may be to perform a complete change in the priorities or values held. Such an aspect of response shift might be termed the reconceptualisation of a construct itself. It occurs when a person changes his or her own definition of a particular construct. Accordingly, a person who develops a chronic illness may no longer think that good physical health is essential for a good quality of life. Thus it may be removed from their conceptualisation and may be replaced by aspects such as spiritual health and creativity. Again, a participant in Albrecht and Devlieger (1999) provides an illustration. " ... Disability made me grow-up. My idea of success changed" (p. 983). A shift of definition of this sort may well be regarded as a qualitative change in what is considered to be important. While little, if any, research has explored the point, it may be that some individual reactions during the process of response shift may amount to qualitative changes in perceptions and values, with corresponding implications for the construct of overall quality of life.


MODELS OF RESPONSE SHIFT

Sprangers and Schwartz (1999) propose a model with five elements (See Figure 1)a catalyst, antecedents, mechanisms, response shift and perceived quality of life. The catalyst is an event that changes the person's status. An example would be a loss of physical function necessitating a move to a nursing home. The antecedents are the individual's history, personality, expectations and beliefs that they take into the new situation. Mechanisms are the person's cognitive, emotional and behavioural response to the catalyst. Given the antecedents and the presence of a catalyst people will deal with the situation in idiosyncratic ways. They may use coping strategies, compare their situation to those of others or rely on social support or spiritual beliefs. Such a change in their life situation may lead people to reconsider what is important in their lives and change their goals and expectations accordingly. These mechanisms lead to the response shift, wherein their internal standards, values and ideas about what is necessary for a good quality of life may all change. Their perceived quality of life is the outcome of this reappraisal of life in light of the catalyst event. This process becomes a feedback loop, so that the response shift is an iterative process. Thus, perceiving an unsatisfactory quality of life can itself become a catalyst, prompting the individual to engage one or more mechanisms that may produce an eventual response shift (Sprangers & Schwartz, 1999).

j Antecedents e9 • soclodemographlcs

• personality - expectations - spmtul'tde nttty

Catalyst Mechanism --_. Response Shift e9 - coping Ie changes 10

- SOCial companson - Internal standards - SOCial support - values - goal reordenng - conceptualizations - retrammg expectations - splntual pracbce

Perceived QOL

Figure I. Theoretical model of response shift and quality of life (QOL) Sprangers & Schwartz (1999)

Wilson (1999) suggests that response shifts may represent an aspect of a broader mechanism of adaptation. Carver and Scheier (2000) contend that response shift is not a distinct process but is part of normal adaptive processes that may simply be more obvious in extreme circumstances. Their theory involves feedback control processes. The first is a behaviour loop, the aim of which is to reduce the discrepancy between the self-image and reality. It consists offour components: I) an input function which conveys information to the person about their current condition.


2) a reference value which may be considered to be a personal goal. It may be a long- or a short-term goal. 3) a comparator, a function which compares the input function to the reference value and 4) an output function - the result of the comparison, which may be considered equivalent to a behaviour, either physical or psychological. If a discrepancy is apparent, then the behaviour aims to reduce the discrepancy. Ifthere is no discrepancy then the action is to continue the status quo.

A second loop runs parallel to, and interacts with, the behavioural one. It relates to affect. In this loop, the input function is the rate of discrepancy reduction in the behaviour system over time. In a sense, it is monitoring the progress of the behaviour loop and comparing it to a reference value. If progress is too slow then the system gives rise to negative affect that can motivate an increase in the rate of action. If progress is satisfactory, then the affect experienced is positive. This affect loop produces a sense of confidence or of doubt which effects the way that people respond to the challenge of a discrepancy between input functions and reference values. If people are confident of success they will continue working toward their goal. If, however, they are in doubt they may disengage from, or amend, the goal.

The reference values in the model proposed by Carver and Scheier (2000) are not fixed but may change as a result of experience. However, this reference value change is not immediate but is the result of an iterative process of effort and feedback. If the discrepancy between the reference value and the current situation is great it may motivate the person who feels confident to try harder to achieve their goal. Only with continued failure and an increasing discrepancy will the person revise their reference value. They may do this by reducing it to a more manageable and achievable level or disengage from that goal and change to a goal in another life domain. For instance, people whose goal is to maintain a full working life may be faced with difficulties after a disabling injury. Initially, they may work harder attempting to maintain their previous work level. However, if they find themselves with more and more work and depleted physical resources they may either reduce their work load, take on less demanding tasks or opt out of work altogether and focus on a new goal of improving their personal relationships.

Carver and Scheier suggest that this resetting of the reference value in both loops is equivalent to the scale recalibration aspect of response shift proposed by Sprangers and Schwartz (1999). They address the other two components of response shift -changing values and reconceptualization - by considering goal-setting as a hierarchical process. At the lowest level are concrete behavioural goals that are important because they aid in the achievement of a higher order goal linked to the idealised self. Therefore lower order goals may change if people find an alternative route to the higher goal. If higher-level goals become unattainable then the person might opt for yet another higher-order goal that maintains the idealised self. Thus goals change and the elements composing a construct are amended, perhaps to the point that an entirely different evaluation of life results. Most would regard a change of that nature to be qualitative.

As shown in Figure I, Sprangers and Schwartz (1999) suggest that social comparison is a mechanism in the process of response shift. Gibbons (1999) examined this role of social comparison. He defines social comparison as "comparing oneself, one's status and/or one's situation with that of others" (p. 1517).


According to Gibbons, the literature suggests that when negative events occur, people wish to make comparisons with others to gain insight into the impact the event has on that individual. Social comparison theory posits that our self-knowledge is enhanced through comparisons with others. This comparison can alter our perspective and our perception of our lives and ourselves. Thus, social comparison can produce a reconceptualisation of abstract constructs such as quality of life. Downward comparison can be beneficial where the person can see something that distinguishes them from the other person. In the same way upward comparison aids in an adaptive response shift where the person can see something that is similar in both themselves and the comparison person.

Although there is little direct research on the topic of social comparison in response shift, Gibbons (1999) did find studies where the outcome of social comparison appear to support the occurrence of the three components of response shift. In several studies cited by Gibbons, the presence of a disabled or chronically ill person was associated with increased reports of well-being in healthy subjects. This may be because their own good health became more salient in their assessment of their quality of life. However, these were experimental findings that may be shortlived and not last beyond the immediate experience. There is evidence that a more permanent effect can be found in disabled individuals who find greater salience in non-health dimensions of quality of life such as social support and family relationships. That is, they choose to focus on areas of their lives in which they feel that they are at an advantage compared to others. Gibbons interprets this as evidence of scale recalibration. Heidrich and Ryff (1993a, b) suggest that the extent to which individuals make social comparisons in order to place their own situation into perspective and how well individuals can integrate themselves into a new social structure are the most relevant issues in the context of adaptation to changes in physical health. It appears plausible that similar mechanisms may operate for individuals moving into residential care and adapting to the new circumstances in which they find themselves. Indeed, Mitchell and Kemp (2000) found that social cohesion and participation in social activities were among the stronger predictors of quality of life in their sample from 55 California residential facilities.

A respondent in the Albrecht and Devlieger (1999) studies appears to benefit from downward comparison.

"Compared to others, [ have It good. [ see other people with conditions like mine and feel and look a lot better than they do. Sometimes [ even see people who are in much worse shape than me; who don', have farmly or friends or can't get out and about. I'm lucky. "(pp 983-984)

Helping other people may also facilitate response shift. Schwartz and Sendor (1999) cite research that found that peer helpers reported a greater sense of personal control, less depression, a sense of calmness and enhanced self-esteem after their experience as a supporter. In fact, Schwartz and Sendor found that providing assistance had a greater positive impact on the quality of life of the helper than it did on the recipient. They found that this benefit was greater during the second year of the study than it was during the first. These researchers suggest that a process of social comparison might have facilitated such a shift, that is, comparing their situation with that ofthe peer they were helping.


There have been some investigations of response shift in quality of life among patient groups. Jansen et al. (2000) conducted an investigation of scale recalibration and change in values in a group of women with breast cancer before and after they received radiotherapy. They used a procedure called the Then Test. In this method, the respondents are asked at the time of the post-test to complete the measure with reference to how they recall themselves as having been at the time of the pretest. The difference between the actual pretest and the Then Test scores are thought to be indicative of response shift (Schwartz & Sprangers, 1999). Response shifts can be positive or negative. In the context of the Jansen et al. study, a participant who suffered negative effects from treatment would be expected to report a better quality of life on the Then Test than on the pretest. Alternatively, if treatment brought about a significant improvement, respondents would be likely to report that their condition was much worse on the Then Test. The study found evidence of scale recalibration in the expected directions for all indicators measured, although only "overall quality of life" was statistically significant. This research, however did not find any change in values over the 6 or 7 weeks of treatment.

Another study considered the issue of response shift in 289 subjects aged between 65 and 97 (Daltroy, Larson, Eaton, Phillips, & Liang 1999). They did not find support for the proposition that, as we age, we consider some level of impairment to be normal and, therefore, might understate practical difficulties. These researchers did, however, find strong evidence that self-reported increase in disability over the preceding six months, and level of dissatisfaction with ability in the preceding week, were strongly related to self-reported disability, controlling for observed function, as were gender and current joint pain or stiffness. Combined, these variables accounted for 85% of the observed variance in disability scores. Those subjects who had a recent decline resulting in a level of functioning they perceived to be unsatisfactory reported higher levels of disability than others who had not had a recent experience but had the same level of observed function. Daltroy et al. suggest that this finding is evidence in support of the adaptation level theory in that the recent experience of the respondents had caused them to recalibrate their scale in the direction of higher disability.

It may be that those subjects reporting a higher level of disability are, in fact, in the early stages of response shift while the other participants had already achieved such an adaptation. This is suggested in the lower level of dissatisfaction reported by these latter participants.

The Sprangers and Schwartz theory can also be employed to understand the findings. The recent decline in function is the catalyst. At the point of testing these participants would, in fact be in the process of employing mechanisms to address their situation but have not yet reached the point of response shift.

The Carver and Scheier model is equally applicable. The input function is the current health of the respondent. The reference value may be the goal of independence. The comparator function produces an output of perceived decline in function. The affect loop, detecting a discrepancy between the reference value and the input function, results in negative affect - the dissatisfaction with ability in the last week.


There are interesting aspects of the results reviewed here. Both Sprangers and Schwartz (1999) and Carver and Scheier (1999) propose that response shift or adaptation is a gradual process. With regard to the Daltroy et al. study, response shift should therefore not be affected by the experience of joint pain or stiffness on the day of the examination, yet this was a significant contributor to the variance accounted for in self-reported disability. In the same way, dissatisfaction with ability in the last week should not have been the highly significant contributor that it was if, in fact, scale recalibration did not occur until after a period of time. Similarly Jensen et al. found evidence of recalibration after 6 or 7 weeks. These findings suggest that scale recalibration may be a more immediate process than other components of response shift or adaptation.

Another consideration is those people who do not achieve such an adjustment. Although 54% of Allbrecht and Devlieger's (1999) disabled respondents reported a good or excellent quality of life, a significant number reported their quality of life to be poor or fair. One possible explanation from the response shift and adaptation approaches is that the latter group may have become disabled more recently than those who reported good quality of life. Additionally, their own antecedents, such as personality and self-esteem, may limit the mechanisms available to them. Gibbons (1999) suggests that individual characteristics such as depression and optimism may also be mediators of response shift. According to Carver and Scheier (2000), depression might be the result of the frustration of a goal that is deeply associated with the idealised self. For people with depression, as with others who do not appear to adapt easily, it is very difficult to disengage from a goal central to the self.

Sprangers and Schwartz (1999) note that the way in which these three elements of response shift operate has not yet been elucidated. For instance, Sprangers and Schwartz speculate that it may be necessary for at least two components to change before a response shift is produced or that there may be some hierarchical order of changes.

Whether response shift is a discrete phenomenon or part of a wider process of adjustment remains to be explored. There is little direct literature examining either approach with regard to quality of life research. It is, however, an important endeavour. Understanding how people maintain life satisfaction in the face of disability or disease can guide the development of interventions to facilitate this accommodation.

A final consideration is that such adaptation poses a methodological dilemma for researchers assessing the impact of a treatment on the quality of life of the subject using a pretestlposttest design. Various methods to overcome this problem have been proposed, such as the Then Test. For a full assessment of these techniques see Schwartz and Sprangers (1999). As far as we know, the Then Test or similar approach has not been evaluated with older people before and after a move into residential care.

Given the likely occurrence of a response shift process during the course of a move into residential care, it is correspondingly likely that life in residential care is not qualitatively different from life in the community for the majority of people who make this transition during their life. One relevant study on this point is by Reberger, Hall and Criddle (1999), who examined quality of life (as measured by the SF36) in


new admissions to aged care hostels in Western Australia. They found that the average level of well-being actually improved at three months post-admission. If the quality of life in residential care is not qualitatively different for the majority, but remains on the same dimensions as life outside of residential care, the question remains as to what dimension or dimensions of quality of life are relevant. Another question is how many people do undergo such substantial recalibration of their views and values that qualitative change in quality of life can result. These considerations bring us to the question of the nature of quality of life.

THE CONSTRUCT OF QUALITY OF LIFE

The construct of quality of life itself lacks a commonly accepted definition. This lack of clear definition for the construct of quality of life has been noted for at least two decades (Van Dam et a!., 1981), without much substantial progress in theoretical developments since then. Cummins et a!. (this volume) note that the term "quality of life" is used inconsistently and with different connotations. Kozma et al. (1991) noted that the terms well-being, quality of life, happiness, and life satisfaction all tend to be used as if they were the same construct. Current working definitions include: absence of illness, health status, subjective well-being, functional status, happiness, life satisfaction, and positive affect, while health economists place yet another and different meaning on the phrase quality of life (Joyce, O'Boyle, & McGee, 1999). DeNeve and Cooper (1998) illustrated the same conceptual overlap in using the terms subjective well-being, life satisjaction,_happiness, and quality oj life as search terms in conducting their meta-analysis. Kozma et al. point out that semantically and conceptually, these are all distinct constructs that developed over time as different researchers developed their own ideas. They demonstrate the different domains to which the various terms differentially apply (1991, p. 9). They also note that concerns over the use of these terms as if interchangeable dates to the early 1980s.

Whether terms such as happiness, morale, satisfaction and well-being are indeed synonymous or not is largely a matter of accepted definitions and usage in everyday language. At that level, many of these terms clearly can be distinguished from one another, even though consultation with a dictionary shows the links among them. A more pragmatic consideration is whether the measures that have been developed to measure quality of life under these various terms are in fact measuring the same thing. That question is a different one, and one that is amenable to empirical answers through psychometric approaches such as the use of multitrait multimethod analyses. The literature on the topic is, unfortunately, of mixed value to inform the debate. In part, this appears to be due to independent bodies of research that make little or no reference to one another and to different philosophical and methodological approaches to studying quality of life.

As one illustration, in many medical settings quality of life is simply defined as an absence of illness and symptoms of the disease under consideration. The proliferation of disease-specific quality of life scales reflects this implicit definition (Bowling, 1995). Many applied researchers in the health area adopt a working definition of quality oflife as health status (e.g., Walker & Dordrecht, 1993). Despite


this common equation, Bergner, who developed the Sickness Impact Profile that is commonly used as a measure of health status and quality of life (Bergner, 1993), clearly distinguishes the two as quite different constructs (Bergner, 1989).

If quality of life lacks a common frame of reference that defines it as a construct, can the methods that are used to measure it provide an operating definition? Edwards and Bagozzi (2000) discuss the concept of formative construct validity for characteristics which lack a theoretical definition. They use indices of socioeconomic status as an illustration of a formative construct. Quality of life, however, may even lack the widely accepted constituents that appear to be necessary for formative construct. There often appear to be connotations in the literature of the phrase "quality of life" that promote its conceptualization as a global attribute, such that the majority of efforts to measure it in practice are unsatisfying (Hunt, 1999). At the same time, the continuing use of multiple terms as if they were synonymous and interchangeable can only be supported if there is empirical evidence for doing so. For the most part, we argue that the limited available evidence does not provide convincing evidence that the scales promoted under these various terms are in fact measuring the same thing, although Kozma et at. (1991) argue to the contrary.

While the terms positive affect and satisfaction are commonly used as reflecting aspects of quality of life, these are characteristics that have been studied in their own right for many years. The substantial literatures on positive affect (Folkman & Moskowitz, 2000), happiness (Headey & Wearing, 1992; DeNeve & Cooper, 1998), and life satisfaction (Diener, 1984; Diener, Suh, Lucas, & Smith, 1999) are evidence of the conceptual and theoretical strength of many psychological attributes that have been defined outside of the mainstream view of quality of life. While measures of life satisfaction are often taken as indices of subjective well-b~g, the early work in this broad area illustrates how terms have blended into one another. The pioneering report by Campbell, Converse, and Rodgers (1976) took quality of life as equivalent to well-being. They defined well-being as encompassing happiness, satisfaction with life, general affect, stress and anxiety, and personal competence. In more recent years, Ryff (1989) has defined six domains of well-being: self-acceptance, autonomy, environmental mastery, purpose in life, personal growth, and positive relations with others and established an instrument for assessing these domains. It is clear that the definitions of Campbell et at. and Ryff ha~~,f.omparatively little in common. Note that under Ryffs definition, well-being,;~~t;itutes one aspect of quality of life, while for Campbell et at. and others, well-bt:ing is quality of life. In another perspective, quality of life is a function of psychological well-being (armel, Lindenberg, Steverink, & VonKorff, 1997).

The examples cited above show that nomological networks have developed around some of the terms that have been used as synonymous with quality of life. The fact that well-being and quality of life have been placed in all possible combinations with regard to one another clearly illustrate the hazards of the casual use of these terms and the serious lack of consensus at theoretical levels.

Further examples of the theoretical confusion over the terms can be provided. Okun and Stock (1987) identified positive and negative affect and cognitive factors related to adaptation and social comparisons as the major elements of well-being. In order to assess these and related characteristics, they recommended the use of two

M.J. POWER, M. BULLINGER AND THE WHOQOLGROUP 123

existing scales and parts of a third, with no strong recommendation for the assessment of the relevant cognitive factors. A further example of the lack of theoretical integration is provided by the work of Diener and his colleagues on life satisfaction, which is also claimed to assess subjective well-being (Diener, 1984; Diener et aI., 1999). In this body of work, however, the method of assessment is quite different from that conceptualized by Ryff (1989), in that it relies upon short, unidimensional measures of life satisfaction. It is perhaps significant that the recent review by Deiner et al. (1999) contains only one reference to the substantial body of work on well-being by Ryff. It is also clear that although these two bodies of work both clearly deal with the same topic, well-being, the methods of measurement clearly show that the traits under discussion are in fact two entirely different constructs. Further evidence of the complexity of measurement issues in this area is provided in the chapter here by Vittors0 and R0ysamb, who report that even the basic 5 items of the Satisfaction with Life Scale can be interpreted differently across different cultures and language groups.

In addition to satisfaction and affect, the other significant domain in the assessment of quality of life that is relevant to older people in residential care is that which arises from health care and medicine. Many definitions of quality of life identify physical health and psychological well-being as being the major determinants of quality of life. Health and disability have already been mentioned as important factors that affect the emotional state and quality of life of older people. The large number of disease-related measures and multidimensional measures of quality of life highlight the perceived importance of this domain. Yet as Hunt (1999) points out, functional limitations themselves do not affect all people equally. Some adapt better than others; some manage adversity in good spirits. To the extent that health-related measures of quality of life do not tap into such attributes, they are not addressing issues that should be measured if these attributes are relevant to the traits that one seeks to measure under the rubric of quality of life. It is just these characteristics that may be most relevant to issues of quality of life in residential care. Individuals in assisted living quarters by definition need assistance with some aspect of life. To lose such independence means loss of quality of life under some definitions, but perhaps not under others. The choice of measure of quality of life will therefore require a thoughtful consideration of what actual definition of quality of life is most relevant and what actual measures may best evaluate those characteristics. In addition, the likelihood that response shift processes may occur in some people and not others, or may be in the process of occurring for other people, poses an additional methodological complication.

One method of dealing with the complexity of quality of life as an attribute is to face the issues squarely and regard quality of life as multidimensional. Lawton (1983, 1991) has outlined four overlapping sectors of "the good life": psychological well-being, perceived quality of life, behavioral competence, and the objective environment. Note that this definition again illustrates a different view of the relationship of quality of life to well-being. However, Lawton's conception of quality of life is not so much multidimensional as it is universal, in that it encompasses virtually every element of internal and external life under a single domain. Such extreme breadth renders it subject to major difficulties in practical


measurement unless an explicit multidimensional approach is adopted. For example, Lawton notes that the domain of behavioral competence alone is "... able to accommodate any externally observable facet of a person" (p. 8), and there is no accepted instrument available that is based upon this model. An alternative solution to this measurement complexity lies with the measures of quality of life that explicitly take into account this complexity through the use of multiple scales. One such example is the multidimensional quality of life scale developed by Evans and Cope (1989). This instrument assesses five major domains of quality of life: general well-being, interpersonal relations, organisational activity, occupational activity, and leisure and recreational activity. The measure includes a total of 15 content scales, together with a scale to assess social desirability. This degree of sophistication is not common in the assessment of quality of life. Such a scale also clearly demonstrates that any effort to form a composite measure of quality of life that encompasses many domains into a single numerical index can only result in an essentially meaningless, non-specific global quantity.

QUALITY OF LIFE IN RESIDENTIAL CARE: RELEVANT DOMAINS

How one defines quality of life clearly has a bearing upon the issue of whether quality of life changes in either a quantitative or qualitative way when an individual moves into a residential care facility. If the objective external environment is regarded as a major component of quality of life, then it is likely that life in residential care will be seen as undergoing a substantial qualitative reduction in quality. If, on the other hand, one has a view of quality of life that encompasses primarily life satisfaction and the value system of the resident, then changes from life in the community to life in residential care may result in quantitative changes, or even no change at all.

Clearly health status and disability are relevant in the residential care system, but these are factors to which many people appear to adapt. The concept of freedom is generally not explicitly taken as an element of quality of life. The extremely broad definition advocated by Lawton (1991) undoubtedly can subsume the essence of freedom and independence through the person-environment and social-normative elements. Lawton's approach does not explicitly identify freedom and autonomy as elements of his model, yet the full expression of characteristics in his model virtually requires the presence of freedom of choice and autonomy of action. As an example, Ball, Whittington, Perkins and Patterson (2000) examined quality oflife in American assisted living facilities, equivalent to Australian hostels. They found that autonomy and control were very important to residents who valued their independence and had choice about aspects of their lives. Residents who were able to make a practical contribution to the facility, either by helping other residents or even doing chores felt less dependent than those who were unable to reciprocate for the assistance they received.

Such efforts at autonomous actions are not always encouraged in residential care facilities. Draper (1996) found that, while all of the II nurses interviewed valued the individuality and autonomy of the residents, their behaviour was quite contradictory. When, in their professional opinion, an activity would be of benefit to the resident,


they were prepared to impose that activity and limit the resident's choice. They did this by using mechanisms including compromise, massive encouragement, forcing, physical restraint and even physical force. Nurses explained this contradiction in terms of acting in the greater good of the patient or fitting in with the smooth running of the institution. Mitchell and Kemp (1987) reported low levels of rated independence in their sample of residents of care facilities, but independence was also consistently associated with ratings of quality of life.

The issues of independence and control that are implicit in the concept of freedom clearly have a bearing on both well-being and quality of life in residential care. Residential care facilities can be very restrictive environments. Are there critical elements in the social climate of care facilities that reduce autonomy, wellbeing, and quality of life? Do the presence of disability and cognitive impairment reduce freedom more than coercive actions by nurses that are intended to promote well-being? Another way of asking the question: what is the role of freedom (of movement, of social participation, of selection of food) in quality of life, as opposed to physical health and function?

These questions will be difficult to answer. To the extent that questions are phrased in terms of quality of life, the lack of a consensus definition for this term can only hinder such efforts. The multiple operational definitions of quality of life in the many different fields of research in which the term is used also will provide formidable obstacles to the emergence of any clear resolution. At the same time, there is increasing awareness among researchers of the ambiguity of the meaning of the umbrella term "quality of life" and of the importance of clearly defining what they mean by this term in a study. The increased consideration of such issues will then in tum help clarify the processes that are operating in older people as they adapt to the transition from independence to residential care. Such increased attention to measurement issues may also help us understand whether this process is one of quantitative change along the important dimensions or of a qualitative change to different important dimensions in the different environment of residential care facilities.

Address for Correspondence: Dr. E. Helmes, School of Psychology, James Cook University, Townsville, Qld 4811. Tel: +61 (07) 4781 5159, Fax: +61 (07) 4781 5117, E-mail: [email protected]

REFERENCES

Albrecht, G.L., & Devlieger, PJ. (1999). The disability paradox: high quality oflife against all odds SOCIal SCIence and MedIcine, 48, 977-988.

Albrecht, G.L. & Devlieger, PJ. (2000). Disability assumptions, concepts and theory reply to Tom Koch SocIal SCIence and MedIcine, 50, 761-762.

Andrykowski, M.A , Brady, MJ , & Hunt, J W. (\ 993). Positive psychosocial adjustment in potential bone marrow transplant recipients: cancer as a psychosocial transition. Psychooncology, 2, 261-276.

Asakawa, T., Koyano, W., Ando, T., & Shibata, H. (2000). Effects of functional dechne on quality oflife among the Japanese elderly. internatIOnal Journal oj Ageing and Human Development, 50, 319-328.

Ball, M. M., Whittington, F. J., Perkins, M. M., & Patterson, V. L. (2000). Quality oflife in assisted living facilities. Journal oj Applied Gerontology, /9,304-325.

126 QUALITY OF LIFE rN RESIDENTIAL CARE

Bahnson, C. B. & Bahnson, M. B. (1996). Role of the ego-defenses: Denial and repression in the etiology of malignant neoplasm. Annals of the New York Academy of Sciences. 125. 827-855

Berdes, M. S. P. H. & Zych, A. A (2000). Subjective quality of life of Polish, Polish-immigrant, and Polish-American elderly. InternatIOnal Journal of Aging and Human Development. 50. 385-395.

Bergner, M. (1989). Quality of life, health status, and clmical research. Medical Care 27 [3. Suppl J. SI48-S156.

Bergner, M. (1993). Development, testing, and use of the Sickness Impact Profile. In S.R.Walker & R. M. Rosser (Eds.), Quality of life assessment· Key Issues In the 1990s (pp. 95-110). Dordrecht: Kluwer Academic.

Berlowitz, D R; Du, W., Kazis, L. & Lewis, S. (1995). Health-related quality of life of nursing home residents: Differences in patient and provider perceptions. Journal of the American Genatrlcs Society. 43, 799-802.

Blumer, D., & Heilbronn, M. (1982). Chronic pam as a variant of depressive disorder: The pam-prone disorder. Journal of Nervous and Mental Disease. 170. 381-414.

Bowling, A (1995). Measuring disease: A review of disease-specific quality of life measurement scales Buckingham: Open University Press.

Breevelt, I.S. & Van Dam, F.S AM. (1991). Underreporting by cancer patients: The case of response shift. Soctal SCience and MediCine. 32. 981-997.

Bretscher, M., Rummans, T., Sloan, J., Kaur, 1., Bartlett, A, Borkenhagen, L., & Loprinzi, C. (1999). Quality of life in hospice patitents: A pilot study Psychosomatics. 40. 309-313

Campbell, A, Converse, P. E., & Rodgers, W. L (1976) The quality of Amencan life New York: Russell Sage Foundation.

Carver, C.S" & Scheier, M.F (2000) Scaling back goals and recalibration of the affect system are processes in normal adaptive self-regulatIOn. understanding "response shift" phenomena Social SCience and MediCine. 50. 1715-\722.

Cassileth, B.R" Lusk, EJ., Strouse, T.B., Miller, D.S., Brown, L.L., Cross, P.A. and Tenaglia, AN. (1984). Psychological status in chronic illness. a comparative analysis of SIX diagnostic groups. New England Journal of Medicine. 3Jl, 506-511.

Clark, P , & Bowling, A (1990). Quality of everyday life in long stay institutions for the elderly. An observational study oflong stay hospital and nursing home care. SOCial SCience and MediCine, 30, 1201-12\0.

Clarke, P 1., Marshall, V. W., Ryff, C. D., & Rosenthal, C. 1. (2000). Well-being in Canadian semors' Fmdings from the Canadian Study of Health and Agmg. Canadian Journal on Aging, 19, 139-159.

Cohn, 1. & Sugar, J. A (199l). Determinants of quality oflife in institutIOns: PerceptIOns of frail older residents, staff, and families. In 1. E. Birren, 1. E Lubben, J CRowe, & D. E. Deutchman (Eds.), The concept and measurement of quaitty of life In the frOlI elderly (pp. 28-49). San Diego, CA: Academic Press.

Conmdis,1. (1987). Life in older age: The view from the top. In V. W. Marshall (Ed.), Aging In Canada (2"d ed., pp. 451-472) Markham, ON. Fitzhenry & Whiteside.

Daltroy, L.H., Larson, M.O , Eaton, H.M., Phillips, c.B., & Liang, M.H (1999). Discrepancies between self-reported and observed phYSical function in the elderly- the mfluence of response shift and other factors. SOCial SCience and MediCine, 48, 1549-1561.

DeNeve, K M. & Cooper, H. (1998). The happy personality: A meta-analysis of 137 personality traits and subjective well-being. Psychological Bulletin, 124, 197-229.

Diener, E (1984). Subjective well-being Psychological Bulletin, 95, 542-575. Diener, E , Emmons, R. A, Larsen, R. 1., & Griffin, S. (1985). The SatisfactIOn with Life Scale. Journal

of Personality Assessment, 49, 71-75. Diener, E , Suh, EM., Lucas, R E., & Smith, H. L (1999). Subjective well-beIng: Three decades of

progress. Psychological Bulletin, 125, 276-302 Draper, P. (1996). Compromise, masSive encouragement and forcmg' a diSCUSSIOn of the mechanisms

used to lImit the choices available to the older adult In hospital. Journal of Clinical Nursing, 5, 325-331.

Edwards, J. R. & Bagozzi, R. P. (2000) On the nature and direction of relationships between constructs and measures. Psychological Methods, 5, 155 -\7 4

Evans, D. R & Cope, W. (1989). Quality of Life Questionnatre manual. Toronto Multi-Health Systems. Ferrell, B. A, Ferrell, B. R., & OsterweiL D. (1990) Pain in the nursing home. Journal of the Amencan

Geriatrics SOCiety, 38, 409-414.

M.J. POWER, M. BULLINGER AND THE WHOQOL GROUP

Folkman, S. & Moskowitz, J. T. (2000) Positive affect and the other side of copmg. American PsychologIst. 55. 647-654.

127

Fry, P. S (2000). Whose quality of life is It anyway? Why not ask seniors to tell us about It? Internallonal Journal of Agmg and Human Development. 50. 361-383.

George, L K. & Clipp, E C (1991) Subjective components of aging well. GeneratIons. 15.57-60. Gibbons, F.X. (1999). Social comparison as a mediator of response shift. Soc/Ol SCIence and Medlcme. 48.

1517-1530. Grundy, E., & Bowling, A (1999). Enhancing the quality of extended life years. Identification of the

oldest old with a very good and a very poor quality of life. Agmg & Mental Health. 3, 199-212. Headey, B. & Wearing, A (1992). Understandmg happmess' A theory of subjecllve well-bemg London:

Longman Cheshire. Heidrich, S. M., & Ryff, C D. (1993a) Physical and mental health in later life: The self-system as

mediator. Psychology and Agmg. 8. 327-338. Heidrich, S. M., & Ryff, C D (1993b). The role of social comparisons processes In the psychological

adaptation of elderly adults. Journal of Gerontology. 48.127-136. Hunt, S. M (1999). The researcher's tale: A story of virtue lost and regained. In C. R. B. Joyce, C. A

O'Boyle, & H. McGee (Eds) IndIvidual quality of life' Approaches to conceptualisation and assessment (pp. 225-232). Harwood Academic: Amsterdam.

Jansen, SJ.T., Stlgglebout, AM, Nooij, M.A., Noordijk, E.M. & Kievit, J. (2000). Response shift in quahty oflife measurement in early-stage breast cancer patients undergoing radiotherapy. Quality of Life Research. 9. 603-6 I 5

Jones, M (1988) Growmgold' The ulllmatejreedom. New York: Human Services Press. Joyce, C. R B., O'Boyle, C A, & McGee, H (Eds). (1999). ). IndIVIdual quality of life: Approaches to

conceptuabsatlOn and assessment. Amsterdam: Harwood AcademiC Karlawish, J H T., Klocinskl, J. L., Merz, J., Clark, C M, & Asch, D A (2000). Caregivers'

preferences for the treatment of patients with Alzheimer's disease. Neurology. 55. 1008-1014. Keller, M., Leventhal, E., & Larson, B. (1989). Aging: The hved experience. InternatIOnal Journal of

Agmg and Human Development. 29, 67-82. Kendig, H., Browmng, c., & Young, A E. (2000). Impacts of illness and disability on the well-being of

older people. Dlsablbty and RehabzlltatlOn An InternatIOnal Muilldlsclpimary Journal. 22. 15-22. Koch, T. (2000). The illusion of paradox' commentary on Albrecht, G.L. and Devlieger, P 1. (1998). The

disability paradox. high quahty of life against all odds. SOCial Science & Medicine, 48, 977-988. SOCIal SCIence and Medlcme, 50. 757-759.

Kozma, A , Stones, M. 1., McNeil, 1. K (1991) Psychological well-bemg m later life Toronto: Butterworths Canada.

Krach, P, De Yaney, S , DeTurk, c., & Zink, M. H. (1996) Functional status of the oldest-old in a home setting Journal of Advanced Nursmg. 24. 456-464.

Kruzich, J. M., Chnton, J F., & Kelber, S T. (1992). Personal and environmental influences on nursing home satisfactIOn. The GerontologISt. 32. 342-350.

Kunzmann, U., Little, T D, & Smith, 1. (2000) Is age-related stablhty of subjective well-being a paradox? Cross-sectional and longitudmal evidence from the Berlin Aging Study. Psychology and Agmg. 15.511-526

Lawton M. P. (1983). EnVironment and other determmants of well-being in older people. The GerontologIst. 23. 349-357.

Lawton, M. P. (1991). A multidimensional view of quality of life In frail elders. In J. E. Birren, 1. E. Lubben, J CRowe, & D. E Deutchman (Eds.), The concept and measurement ofqualrty of life m the jratl elderly (pp. 3-27). San Diego. CA: Academic Press.

Loebel, J P, Loebel, J. S, Dager, S. R, Centerwall, B. S., & Reay, D T (1991) AnticipatIOn ofnursmg home placement may be a precipitant of sUicide among the elderly Journal of the AmeT/can GeT/atncs SocIety. 39. 407-408.

MeTCIer, c., Peladeau, N ,& Tempier, R (1998) Age, gender and quality ofhfe Commumty Mental Health Journal. 34. 487-500.

Mitchell, J M & Kemp, B. J. (2000). Quality ofhfe In assisted living homes' A multidimensional analysis. Journal of Gerontology. 558. 117-127.

O'Connor, B. P. & Vallerand, R. J (1998). Psychological adJustinent variables as predictors of mortality among nursing home reSidents. Psychology and Agmg. 13. 368-374.

Okun, M. A & Stock, W. A (1987). Correlates and components of subjective well-being among the elderly Journal of Applzed Gerontology. 6. 95-112.


Ormel, 1, Lindenberg, S., Steverink, N , & VonKortT, M. (1997). Quality oflife and social production functions: A framework for understanding health effects. SocIal ScIence and MedIcine, 45, 10SI-1063.

Osberg, J.S., McGinnis, G.E., Dejong, G, & Seward, M.L. (1987). Life satisfaction and quality of life among disabled elderly adults Journal of Gerontology, 42, 228-230.

Pearlman, R. A., & Uhlmann, R. F. (1988a). Quality of life in chronic diseases: Perceptions of elderly patients. Journal of Gerontology' MedIcal SCIences, 43, M2S-M30.

Pearlman, R. A. & Uhlmann, R. F. (I 988b ). Quality of life In the elderly' comparisons between nursing home and community residents. Journal of Applied Gerontology, 7,316-330.

Pearlman, R A., & Uhlmann, R F. (1991). Quality of life in elderly, chronically ill outpatients. Journal of Gerontology: Medical Sciences, 46, M31-M38

Reberger, C., Hall, S. E., & Criddle, R. A. (1999) Is hostel care good for you? Quality oflife measures in older people moving into residential care. AustralaSIan Journal on Ageing, 18, 14S-149

Rodin, J. (1986). Aging and health: Effects ofthe sense of control. Science, 233, 1271-1276. Romney, D. M., Jenkins, C. D., & Bynner, J. M. (1992). A structural analysis of health-related quality of

life dimensions. Human RelatIOns, 45, 16S-176. Ronnberg, L. (1998). Quality oflife in nursing-home residents: an intervention study of the effect of

mental stimulation through an audiovisual programme. Age and Ageing, 27. 393-397. Rytf, C. D. (1989). Happiness is everything, or is it? Explorations on the meaning of psychological well

being. Journal of Personality and Soc/QI Psychology. 57. 1069-1081. Schnelle, J. F., Alessi, C. A., A1-Samarrai, N., Fricker, RD., & Ouslander, J. G. (1999). The nursing

home at night: Effects of an intervention on noise, light and sleep. Journal of the Amerrcan GeriatrIcs SOCIety, 47. 430-438.

Schwartz, C. E., & Sendor. M. (1999). Helping others helps oneself Response shift effects In peer support. SOCIal ScIence and MedICine. 48. 1563-1575.

Schwartz, C. E., & Sprangers, MAG. (1999). Methodological approaches for assessing response shift in longitudinal health-related quality-of-Iife research. SOCIal SCIence and MediCine. 48. 1531-1548

Sprangers, MAG., & Schwartz, C.E. (1999). Integrating response shift into health-related quality oflife research: A theoretical model. SOCIal Sc,ence and MedIcine. 48. 1507-1515.

Urciuoli, 0., Delio Buono, M., Padoani, W., & De Leo, D. (1998). Assessment of quality of life in the oldest-old living In nursing homes and at home. ArchIves of Gerontology and Geriatrrcs. Suppl6. 507-514.

Walker, S. R. & Rosser, R. M. (Eds.). (1993) Qualrty of life assessment. Key Issues In the 19908. Dordrecht: Kluwer Academic.

Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation ofbnefmeasures of pOSitive and negative affect: The PANAS scales. Journal of Personality and SOCIal Psychology. 54. \063-1070

Wilson, LB. (1999). Clinical understanding and clinical implications of response shift. SOCIal Sc,ence and Mediclne.48,1577-1588.

MJ. POWER, M. BULLINGER AND THE WHOQOL GROUP*

THE UNIVERSALITY OF QUALITY OF LIFE: AN EMPIRICAL APPROACH USING THE WHOQOL

• The WHOQOL group comprises a coordinatmg group, collaboratmg investigators in each of the field centres and a panel of consultants. Dr John Orley and Dr Rex BIllington preVIOusly directed the project. The work reported here was carried out primarily in the 15 initial field centres in which the collaboratmg investigators were: Professor H. Herrman. Dr H. Schofield and Ms B. Murphy, UniverSIty of Melbourne, Australia; Professor Z. Metelko, Professor S. Szabo and Mrs M. PibernikOkanovic, Institute of DIabetes. Endocnnology and Metabolic Diseases and Department of Psychology, Faculty of Philosophy. University of Zagreb, Croatia; Dr N. Quemada and Dr A. Caria, INSERM, Paris, France, Dr S. Rajkumar and Mrs Shuba Kumar, Madras Medical College, India; Dr S. Saxena and Dr K Chandiramani. All India Institute of Medical Sciences, New Delhi, India, Dr M. Amlr and Dr D Bar-On, Ben-Gurion Umversity of the Negev, Beer-Sheeva, Israel; Dr Miyako Tazaki, Department of Science, Science University of Tokyo, Japan and Dr Ariko Noji, Department of Community Health Nursing, St Luke's College of Nursing, Japan; Professor G. van Heck and Dr J. De Vries, Tilburg University, The Netherlands; Professor J. Arroyo Sucre and Professor L. PicardAmi, UniverSIty of Panama, Panama; Professor M. Kabanov, Dr A. Lomachenkov and Dr G. Burkovsky, Bekhterev Psychoneurological Research InstItute, St. Petersburg, Russia, Dr R. Lucas Carrasco, University of Barcelona, Spain; Dr Yooth Bodharamik and Mr Kitikorn Meesapya, Institute of Mental Health, Bangkok, Thailand; Dr S. Skevington, University of Bath, United Kmgdom, Professor D. Patrick, Ms M Martin and Ms D. Wild, UniverSIty of Washington, Seattle, USA and Professor W Acuda and Dr 1. Mutamblrwa, University of Zimbabwe, Harare, Zimbabwe In additIOn to the expertise provided from the centres, the project has benefited from considerable assistance from. Professor M. Bullinger, Dr A. Harper, Dr W. Kuyken, Professor M. Power and Professor N. Sartorius. We would also like to thank the new centres that have Joined the project, some of whom have also provided data for analyses reported in this chapter.

Abstract. This chapter examines first some general issues about the concept of quality of life, in particular, in relation to umdlmensional versus multIdimensional approaches. These issues are then examined in relation to the WHOQOL group of measures. Analyses are then summarised from a variety of datasets that have been collected with the WHOQOL measures that address the issue of universality versus cultural-specificity of aspects of quality of life. While the data proVIde clear support for universal aspects of quality of life, nevertheless, it is also argued that there is a need to respect the limitations of this umversahty.

Conceptually, the extent to which the concept of quality of life is transferable from one culture to another is unclear. Within the past two decades the field of quality of life has received increasing attention with, more recently, one development being the demand for international quality of life research that is conceptually clear, methodologically sound, and of practical application in the health field. Although areas such as anthropology have focussed on quality of life indicators across cultures and nations, this research has focussed primarily on socalled objective indicators of quality of life such as gross national product, infant mortality, life expectancy, and so on. Subjective indicators of quality of life have only recently been included in sociological surveys on the well-being of citizens.

129

E. Gullone and RA. Cummins (eds.), The Universality of Subjective Wellbeing Indicators, 129-149. © 2002 Kluwer Academic Publishers.

130 PREDICTING QUALITY OF WORK LIFE

These surveys have examined domains such as material, financial, and political aspects of well-being rather than health (e.g. Bullinger, 200 I; Cummins, 1996). In contrast, health-related quality of life focuses on dimensions of function and wellbeing that are relevant for an individual's judgement of his or her own health status. In this chapter therefore we will first examine the concept of quality of life and how it has been operationalised in a number of recent measures. We will then focus in detail on one such approach, the WHOQOL group of measures, in particular examining work to date on cross-cultural aspects of the measure.

THE DEFINITION AND MEASUREMENT OF QOL

The term "quality of life" is used in many different ways and one of the major issues that faces this area of work is how the term should be defined and conceptualised. One of the key distinctions that has been made is that between health-related and non-health-related quality of life (e.g. Spilker, 1996). The starting point for a number of the health-related definitions has been the well-known World Health Organization (e.g. 1958) definition of health as:

"a state of complete phYSIcal, mental and social well-being and not merely the absence of disease or infirmity".

The inclusion of the phrase "well-being" in the WHO definition has led some researchers to focus too narrowly on self-reported psychological well-being as the only aspect of quality of life of importance (e.g., Dupuy, 1984). However, "wellbeing" has to be seen as a term that is an important aspect of quality of life, but that is not the only aspect that needs to be considered. The challenge has been to specify the range of health-related and non-health related aspects of quality of life that should also be included, such that "quality of life" is not simply another term for "well-being" .

There is a range of generic measures of quality of life that are currently available and that are in widespread use. Such measures include the long-established global scales such as the Kamofsky Performance Status Scale (Kamofsky and Burchenal, 1949), and more recent derivatives such as the DSM-IV Global Assessment of Functioning (OAF) scale (American Psychiatric Association, 1994). These single measure global scales continue to be used in clinical audit and outcome studies because of their brevity and ease of application. They have questionable reliability because they rely primarily on proxy reports with a potential bias, and questionable validity, because they attempt to squeeze a multi-dimensional construct into a single dimension. A similar problem arises with the utility indices used by health economists such as the QAL Y and the DALY, that is, Quality or Disability Adjusted Life Years (e.g. Torrance, 1996). In addition, these measures require the identification of a disability value on a scale between 0 and 1 that is associated with different illnesses. There is no reason however why such values should be consistent across individuals with the same illness, across the same individual over time, and across different sub-groups within the population.

A significant advance over single index measures has been the more recent development of multi-dimensional approaches to quality of life assessment.

Mol. POWER, M. BULLINGER AND THE WHOQOL GROUP 131

Although even multi-dimensional measures can be used to produce single scales, their preferred use is in the form of profiles of scores across a limited number of dimensions or domains. Two of the most widely used of these measures are the SF-36 (Ware and Sherbourne, 1992) and the EuroQOL groups of instruments (EuroQOL Group, 1990). The extensive use of these measures has led to the collection of rich datasets including population norms for many cultures. There are limitations, however, with each measure. For example, the SF-36 Health Survey is best used as a mid-range assessment because it has been shown to have significant floor and ceiling effects (Ware, 1996). The limitations of the EuroQOL include the fact that, within the scale, health is defined as the absence of problems rather than being construed in a more positive sense. The curtailment of the positive end of the scale in the EuroQOL makes it problematic for general population use because responses tend to have a very skewed distribution towards the "no problem" end of the scale.

The rationale for the development of the WHOQOL, its conceptual background, the proposed uses, and the steps taken to develop the pilot version of the WHOQOL have been described in detail in several recent publications (e.g. WHOQOL Group, 1998a). There were a number of key steps, summarised as follows. First, an initial step involved an international collaborative review to establish a definition of quality of life and an approach to international quality of life assessment. An agreed definition of Quality of Life provided the starting point, in which quality of life was considered to denote:

"'mdividuals' perception of their position m life in the context of the culture and value systems in which they lIve and in relatIOn to their goals, expectatIOns, standards and concerns It IS a broad ranging concept affected m a complex way by the persons' physIcal health, psychological state, level of independence, social relationships and their relatIonship to salIent features of theIr envIronment" (e,g WHOQOL Group, 1995),

The WHO definition of health has provided an excellent starting point for defining quality of life (e.g. WHOQOL Group, 1995), but it leaves open two key questions. First, what other areas should be included in addition to the physical, mental, and social? And, second, should the conceptualisation include for example objective characteristics of the individual such as medical and socio-economic factors in addition to the individual's subjective evaluation? Although existing approaches to these two questions vary, nevertheless, there may now be an emerging consensus for both of these key issues. In addition to the physical, mental, and social aspects there is now a recognition that spiritual and religious aspects need to be included in health-related quality of life (e.g. Power, Bullinger, Harper, & WHOQOL Group, 1999; Spilker, 1996; WHOQOL Group, 1995) and a range of aspects of the individual's physical environment needs to be included in the concept of quality of life.

In relation to the second issue of the objective and the subjective, although many of the earlier measures included both objective characteristics (e.g. being able to run for a bus or walk up a flight of stairs) and subjective characteristics (e.g. rating satisfaction/dissatisfaction with level of physical mobility), the most recent measures have focussed solely on the subjective (WHOQOL Group, I 998a). It now seems to make sense that subjective and objective indicators must be kept separate in order to achieve better conceptual clarity and consistency within each measure. To give an


extreme example, how can an individual living in poverty in a village in India report a higher level of happiness and quality of life than a multi-millionaire in Wall Street? This problem has led economists such as the Nobel-prizewinning Amartya Sen (e.g. Sen, 2001) to reject subjective indicators because of their discordance with objective economic indicators. However, from a psychological viewpoint the discordance between the objective and the subjective is crucial and often provides a testament of how the human spirit can overcome and even flourish under adversity.

Following an agreed definition of quality of life, the aim of the next phase of work within the WHOQOL programme was (I) to break the definition of quality of life down into those aspects of life (facets) considered necessary for comprehensive coverage, (2) to define these facets, (3) to collect and develop items representing these facets, and, (4) to generate an international question pool from which the WHOQOL questions would be derived. This work was carried out simultaneously in fifteen different cultural setting worldwide, with coordination and technical support from the WHO coordination group in Geneva.

Focus groups in each centre generated suggestions for aspects of life that they considered contributed to its quality. The range and definition of facets were developed iteratively, such that each centre involved in the project considered and reconsidered the proposals from their own centre, from other centres, and from the coordinating team. Their contributions were valued highly, some facets were modified, and indeed a facet on "spirituality" had to be added because of the importance given to spirituality by the Asian centres within the project.

THE WHOQOL FACETS

Table I. Pilot WHOQOL Domains And Facets.

Domain I Physical I Pain and discomfort 2 Energy and fatigue 3 Sexual activity 4 Sleep and rest 5 Sensory functions Domain II Psychological 6 Positive feelings 7 Thinking, learning, memory and concentration 8 Self-esteem 9 Bodily image and appearance 10 Negative feelings Domain III Levelo/independence 11 Mobility 12 Activities of daily living 13 Dependence on medicinal substances and medical aids 14 Dependence on nonmedicinal substances (alcohol, tobacco, drugs)


15 Communication capacity 16 Work capacity

Domain IV Social relationships 17 Personal relationships 18 Practical social support 19 Activities as provider/supporter Domain V Environment 20 Freedom, physical safety and security_ 21 Home environment 22 Work satisfaction 23 Financial resources 24 Health and social care: accessibility and quality 25 Opportunities for acquiring new information and skills 26 Participation in and opportunities for recreation/leisure

activities 27 Physical environment: (pollutionlnoise/traffic/climate) 28 Transport Domain VI Spirituality/religion/personal beliefs

Overall quality of life and general health perceptions

The list of facets developed in the pilot WHOQOL is presented in Table I; the Table also shows the grouping of the facets into a set of six domains, the details of which have been presented elsewhere. A further series of steps enabled a pilot WHOQOL comprising 236 questions addressing these 29 facets of quality of life to be constructed in readiness for translation (where not already in the local language) and field testing (see e.g. WHOQOL Group, 1998a). The piloting and subsequent psychometric evaluation had several aims: first, to examine the construct validity of the WHOQOL domain and facet structure, and refine and reduce it accordingly; second, to select the optimum questions for each facet with the aim of producing a version of the WHOQOL for use in the field trials; and, third, to establish the WHOQOL's psychometric properties.

THE PILOT STUDY

A cross-sectional design was used in the first pilot phase of the development of the WHOQOL. An agreed-upon standardised study protocol was followed in the 15 centres that participated in the pilot phase. There were fifteen participating field centres in the pilot phase as follows: I. University of Melbourne, Australia 2. Institute of Diabetes, Endocrinology and Metabolic DiseaselUniversity of

Zagreb, Croatia 3. INSERM, Paris, France 4. Madras Medical College, India 5. All India Institute of Medical Sciences, New Delhi, India 6. Ben-Gurion University of the Negev, Beer-Sheva, Israel


7. Science University of Tokyo, Japan/St Luke's College of Nursing, Japan 8. Tilburg University, The Netherlands 9. University of Panama, Panama 10. Bekhterev Psychoneurological Research Institute, St. Petersburg, Russia 11. University of Barcelona, Spain 12. Institute of Mental Health, Bangkok, Thailand 13. University of Bath, United Kingdom 14. University of Washington, Seattle, USA 15. University of Zimbabwe, Harare, Zimbabwe.

In the pilot study, the centres then collected at least 300 respondents, 250 of whom were users of health care services, and 50 of whom were well, and all of whom completed the pilot WHOQOL together with providing some minimal sociodemographic information. The pilot WHOQOL was administered to a minimum of 300 respondents in each of the 15 field centres. A sampling quota was specified with regard to age (50% > 45,50%:5; 45), gender (50% male, 50% female), and health status (250 persons with a disease or impairment; 50 "healthy" respondents). A greater number of ill than well individuals were selected at this pilot stage, because the main purpose was to use differences between well and ill populations in order to select more discriminating items for the field instrument. At subsequent stages in the instrument development these proportions have been very different.

The pilot data from the centres were then used to examine the psychometric properties of the WHOQOL; following a series of psychometric tests (item frequency analyses, subscale reliability analyses, and item, facet, and domain correlation analyses), the original 236 items were condensed to 100 items to produce the so-called WHOQOL-IOO (The WHOQOL Group, 1998a). These analyses were carried out at the level of individual centres, at the level of summaries across centres, and at the level of the pooled global data, as described in Bullinger et al. (1995). The analyses presented in the present paper are based on the 100-item version of the WHOQOL.

The structure of the pilot WHOQOL has been discussed in detail elsewhere (e.g. The WHOQOL Group 1995) and is very similar to the revised WHOQOL-IOO. The structure presented in Table I shows the hierarchical nature of the WHOQOL approach to Quality of Life. At the highest level there is an overall assessment of Quality of Life, which is represented in the instrument by four questions about general health and well-being. At the next level the original 6 domains from the pilot WHOQOL (Physical, Psychological, Independence, Social, Environmental, and Spiritual) have now been condensed in more recent analyses (The WHOQOL Group, 1998a, 1998b) to give four domains, which are the Physical Health (incorporating Independence), the Psychological (incorporating Spirituality), Social Relationships, and the Environment domains. The measure therefore provides a broader view of Quality of Life than most other measures through, for example, the inclusion of the Social Relationships and Environment domains. At the next level down each of the domains is broken into a number of facets; for example, the Psychological domain includes facets that assess Positive Feelings, Body Image, Self Esteem, Thinking,


Memory and Concentration, Spirituality and Negative Feelings. Finally, there is the level of individual items, which, in the case of the WHOQOL-IOO, amount to four items per facet.

Although most of the subjects completed the WHOQOL themselves, a very small minority of subjects who had literacy problems or who had a disability that interfered with self-completion of the questionnaire, were administered the WHOQOL as a structured interview. Data collection included sociodemographic and basic health status information for each subject. Upon completion of the questionnaires, each centre mailed them to the WHO study centre for data input, plausibility checking (i.e. checking that the questionnaires represented genuine responses through, for example, checking that answers varied and were not all the same, that the questionnaires had not been spoiled in some way, that they were not all in the same handwriting), and statistical analysis.

STATISTICAL ANALYSIS

Regression analysis was used to estimate the proportion of variance explained by different domain predictors for the criterion of overall quality of life as measured by the four general questions included in the instrument. Structural equation modeIling was used to identity the interrelationship between domains and facets with regard to quality of life at each cultural level, as well as the overall data pool, using the EQS program (Bentler & Wu, 1995). In the structural modelling, the WHOQOL hypothetical hierarchy of domains was regarded as a single model to be tested in the 15 centres. Analysis of variance and post hoc Scheffe tests were used to compare means across Centres.

One of the opportunities that data collected in this way offered was the possibility to examine the actual structure and content of quality of life. Perhaps inspired by Maslow's (e.g. 1970) original hierarchy of needs and related work in personality theory, a number of influential approaches have conceptualized quality of life as a hierarchical structure or pyramid with overall well-being at the top, broad domains (such as Physical, Psychological, and Social) at the intermediate level, and then specific facets or components of each domain at the bottom (e.g. Spilker, 1990). This overall hierarchical approach was adopted by the WHOQOL Group. As a preliminary test of this predicted hierarchy, a table of facet and domain intercorrelations was produced. The most notable finding was that whereas the experts had relegated Sexual Activity to the Physical domain (facet-to-corrected-domain r = 0.16), the data showed that respondents considered sex to be part of the Social Relationships domain (r = OAI), to which it was moved.

There were very few missing values in the dataset. Overall, approximately 85% of individual items had less than 2.0% missing values. Following the guidelines set out for the scoring of the WHOQOL therefore (WHOQOL Group, 1998a), missing values were replaced with the appropriate mean variable scores in for the relevant analyses.

The data presented in Table 2 provide a summary of the samples from the 15 centres in relation to age, gender, and health status. The statistics show that there were some differences between the centres for these sociodemographic data; that is,


centres had collected opportunistic samples, which were often more than the 300 originally requested. Because these pilot data were primarily being for item selection, it was decided to use all of the available data from all centres rather than limit them to the original sampling frame.

RESULTS

Table 2. General descriptions of the sample from each of the 15 centres.

n Age ± s.d. % Female % Sick

Total 4802 43.4 ± 16.0 53.8 81.0

Bangkok 300 37.7± 15.3 61.0 83.3 Beer Sheva 344 47.3 ± 18.5 43.7 81.1 Madras 412 38.0 ± 14.3 47.1 76.0 Melbourne 300 41.3 ± 16.6 61.4 69.5 New Delhi 304 40.7 ± 14.3 49.3 83.2 Panama 300 39.7 ± 14.5 58.0 83.3 Seattle 300 47.3 ± 15.9 55.3 83.3 Tilburg 411 48.1 ± 13.9 62.5 83.5 Zagreb 300 44.6 ± 15.6 50.0 83.3 Tokyo 286 46.0 ± 20.0 53.8 80.4 St Petersburg 300 45.2 ± 12.7 49.3 82.7 Harare 300 42.9 ± 13.4 53.8 83.3 Barcelona 305 44.6 ± 16.7 49.2 83.6 Paris 323 42.3 ± 15.6 52.4 77.7 Bath 319 45.0 ± 17.4 50.9 81.2

F = 15.4, X2 = 68.2, X2 = 1844.4, P<.OOI P<.OOI P<.OOI

Summary of Preliminary Frequency, Reliability, and Correlation Analyses

A series of frequency, reliability, and correlation analyses were run on the pilot data from the WHOQOL at three different levels of analysis (see Bullinger et a!., 1996, for a detailed discussion of this approach): I. At the level of individual centres - for example, did centre 1 show problems for

any items for these analyses 2. Summarised across the individual centres - for example, how many of the 15

centres showed the same type of problem with a particular item? 3. On the pooled global data- for example, did a particular item show any

problematic properties in the pooled dataset?


That is, preliminary analyses were carried out for each centre individually, as summaries across the 15 centres, and as a single pooled dataset including data from all centres, in order to examine both general trends in the data, and potential differences between centres.

Item frequency analyses were carried out to examine the distribution of responses across the five-point rating scales for the items. Scale reliability analyses were also carried out (using the SPSS-Windows package) at the three levels of (i) centres, (ii) summarised centres, and (iii) global data; that is, any items with problematic corrected item-total correlations either in the pooled data or in more than 50% of the centres were flagged for reliability problems. These initial reliability analyses were carried out in order to identify items that were inappropriate because they correlated poorly with their own facets at values of Pearson r < 0.4. The analyses of frequency, reliability, and other item correlation problems led to a number of items being dropped at this stage in the analysis, so that they were not considered for inclusion in the field trial WHOQOL. The item elimination process is presented in WHOQOL Group (1998a). Following the exclusion of these items, the scale reliability analyses were repeated for each facet at the levels of the centres, the summarised centres, and the pooled global data. These repeated analyses were again designed to highlight any items that now possessed reliability problems because of the altered composition of the facets. These analyses were also used to provide information about the size of the item-to-corrected-facet correlations, a factor that was taken into account in the item selection procedure for the field trial WHOQOL (see later).

The item selection procedure had implications for the structure and number of facets. The elimination of five facets meant that there were now 24 specific facets and an additional one measuring overall quality of life. In deciding on the number of items to choose to represent each retained facet, the decision was taken to select four items per facet, because four is the minimum number required for the scale reliability analyses to be carried out in any further psychometric testing of the instrument. These considerations therefore led to a consensus for 25 x 4 = 100 items (including the four general items).

The final selection of items took into account a number of features of the items including the extent and the rank order of an item's loading on a particular facet in the pooled dataset, the degree of conceptual overlap between potential items (which was minimised, according to the judgement of the coordinating group, where possible), and the extent and range of problems highlighted in the earlier analyses mentioned above. The Cronbach alphas demonstrate good internal consistency for the facets with a range of 0.65 to 0.93. All facet scores range from 4 to 20, with higher scores denoting higher quality of life, except for the reverse scored facets Pain and discomfort, Negative feelings, and Dependence on medication. The revised field trial WHOQOL has come to be known as the WHOQOL-IOO. As a preliminary step to investigate whether or not similar models might be appropriate across different cultures, a series of multiple regression analyses were carried out. The aim was to examine how each of the four domains loaded onto the overall summary of quality of life, as measured by four general questions relating to the 'Overall Quality of Life and General Health' facet. The summary for one


multiple regression for the global pooled data and for each of the fifteen centres is presented in Table 3. The global data show that all four domains contribute significantly to the prediction of the overall quality of life score; furthermore they accounted for 64.1 % of the variance, demonstrating that the domains provided a good model for overall quality of life. Regression analyses show the physical and psychological domains to predict the largest proportion of variance, although both social relationships and environment domains contribute significantly. The regression equations for individual centres showed a range of accounted for variance from 51.0% (for Tokyo) to over 75% (for Seattle). All domains continue to contribute significantly to the centre-level regression equations, with the exception of the social relationships domain which did not contribute significantly to the overall quality of life facet in two of the field centres (Bangkok and Panama). There are however a number of differences in both the rank ordering of domains across Centres, and in the strength of contribution of domains in relation to different cultures.

MULTIPLE REGRESSION ANALYSES FOR DOMAINS

Table 3. Multiple regression analyses with the "Overall Quality of Life and General Health" facet score as the dependent variable and domain scores as the independent variables.

Domains

Adjusted Physical Psychologi Social Environme R2 cal relationshi nt

2s Global data set 64.1 .304 .265 .171 .242

1 Bangkok 54.0 .226 .394 .049* .218 2 Beer Sheva 57.9 .370 .201 .159 .252 3 Madras 58.2 .243 .316 .212 .131 4 Melbourne 58.0 .252 .197 .331 .185 5 New Delhi 51.9 .253 .267 .219 .157 6 Panama 70.7 .361 .272 .036* .298 7 Seattle 75.3 .328 .337 .200 .175 8 Tilburg 71.8 .460 .289 .197 .079 9 Zagreb 55.5 .238 .169 .179 .343 10 Tokyo 51.7 .298 .160 .187 .229 II St petersburg 62.1 .207 .172 .114 .451 12 Harare 70.6 .466 .284 .098 .100 13 Barcelona 61.9 .347 .181 .197 .294 14 Paris 61.7 .218 .2S0 .IS9 .348 IS Bath 6S.7 .323 .323 .216 .136

Note. Figures for Domains show final equation beta values Asterisks show non-significant values


STRUCTURAL EQUATION MODELS

The multiple regression analyses suggested that all four domain scores contribute to the assessment of quality of life for almost all of the field centres. An exploratory factor analysis also provided good support for the 4-domain solution (the EF A is presented in detail in The WHOQOL Group, 1998a). However, recently derived Confirmatory Factor Analytic techniques provide more powerful methods for testing hypothesized structures or models than is possible with exploratory approaches. The development of statistical programs such as EQS (8entler & Wu, 1995) can provide a test of whether or not the contribution of each domain score to quality of life is similar across different cultures through the comparison of a number of so-called "fit indices" by which the different models can be compared. In order to carry out these analyses, the hierarchical structure of the WHOQOL was used as a model that could be tested both at the level of the global data and replicated across groups of centres.

Table 4. Confirmatory factor analysis at domain level.

I 2 3 4 5 6 7 8 9 IO II 12 13 14 15

Individual centre analyses Global data set Bangkok Beer Sheva Madras Melbourne New Delhi Panama Seattle Tilburg Zagreb Tokyo St Petersburg Harare Barcelona Paris Bath

Multiple sample analyses

CFI .990, df==2 .977 .978 .987 .974 .876 .969 .976 .952 .983 1.00 .999 .989 .978 .989 .981

Centres 1-8 .963 Centres 9-15 .989

Note: CFI = Comparative Fit Index

The conceptual structure of the WHOQOL-I00 suggests that all four domains contribute to the overall assessment of quality of life. Therefore, these four domains would be expected to load onto one single factor (a hypothetical quality of life construct) - see Figure I. It should be noted here that in Figure 1, the overall quality of life shown in the diagram refers to a hypothetical construct, rather than the "Overall Quality of life and General Health" facet included in the multiple regression


analysis. The results of the modelling, both for the global data set and for the individual centres, are presented in Table 4. The value for the global data X2 is now trivial for a dataset of this size (X2 = 62.6 compared to 6029.2 for the independence model) and the Comparative Fit Index (CFI, an index that ranges from 0 to I with values of 0.9 or higher being considered a reasonable level of "fit" for a model) is almost unity, both statistics showing that a single factor model fits the data extremely well. (The CFI is reported here in preference to some of the other fit indices following Bentler's recommendations for large sample sizes.) The parameter estimates shown in Figure I, for the pooled global data set, indicate that all domains contribute to the overall assessment of quality of life, as predicted from the multiple regression analysis. Similarly, in 14 of the 15 centres, the comparative fit index for the single factor model is well above 0.9 (see Table 4). New Delhi was the only centre where the CFI was below 0.9. In this case, allowing error variances to covary for the environment and social relationships domains increased the CFI to .991, thereby indicating a similar model to that of the other centres.

DOMAIN-LEVEL MODELS

Physical ~ '---____ 1 -----.70

Psychological 86 ~~~~ __ r------' Quality of life

L-~S:o:c:ia~l~re:la:tl:·o:ns:h~ip~S~ ___________ .71

Environment 70 ~~==~--~------.

Figure I. Structural model representing the hypothetical construct of Quality of Life. (Parameter estimates are givenfor the global data set.)

Multiple sample analyses were then carried out to determine whether the parameters estimated in the model were invariant across centres. As EQS accepts a maximum of ten groups, two analyses were carried out. In the first, centres 1-8 (Bangkok, Beer Sheva, Madras, Melbourne, New Delhi, Panama, Seattle, Tilburg) were included, and in the second, centres 9-15 (Zagreb, Tokyo, St Petersburg, Harare, Barcelona, Paris, Bath) were included. In the two sets of multiple centre analyses carried out for this single-factor model, all group constraints were satisfactory; that is,- the loadings of all the domains were specified to be equal for each centre and none of these constraints had to be released in order to improve the fit index to above 0.9 (see Table 4). It should be noted however that in addition to equality of factor loadings,

M.J. POWER, M. BULLINGER AND THE WHOQOL GROUP 141

the error variances can also be constrained to be equal across groups; when these additional constraints were included, the two CFI values fell below 0.9 thereby indicating that equality is only approached in terms of factor loadings, not in terms of error variances (nor, of course, in terms of mean values, as mentioned earlier).

MODELS AT FACET LEVEL

For each domain, the model assumed there was only one factor, upon which all facets loaded. Again, it must be noted that the "domains" now refer to hypothetical constructs, rather than the actual domain scores as calculated from the mean of all facets. The data presented in Table 5 shows tests of the domain structures carried out for each of the domains separately, for both the global dataset and each of the individual centres. For Domain 3, all parameter estimates were constrained to be equal, to provide degrees of freedom in the model. The global data set fitted these models well for each domain (see Table 5). Parameter estimates for the global data set indicate that all facets within domains contributed significantly to the domain.

Table 5. Confirmatory factor analysis at facet level using single factor models.

I 2 3 4 5 6 7 8 9 10 II 12 13 14 IS

Individual centre analyses

Global data set Bangkok Beer Sheva Madras Melbourne New Delhi Panama Seattle TiIburg Zagreb Tokyo St Petersburg Harare Barcelona Paris Bath

Multlple sample analyses Centres 1-8

Centres 9-15

Physical

.962, df-;14

.949

.958

.953

.933 957 .957 935 .968 .978 857 .911 .914 .919 .970 942

CFI for domains Psychological SOCIal Environment

.931, dF9

.926 896 .934 .929 .982 .971 .922 .942 .912 981 .830 .989 957 .923 .934

relationships

.965,df-;2 911 1.00 .974 960 .971 .955 .931 .920 1.00 .936 991 .996 .984 .986 .831

.921, df-;20

.959

.942

.895

.914 922 .898 .913 .874 .940 .951 .939 .706 .933 .893 .886

.953, df-;112 .946, df-;71 .950, df-;16 .929, df-;157 .935, dF97 .946, dF62 984, df-;13 .933, df-;136

At the level of individual centre analysis, single factor domain models appear to be appropriate for the majority of centres (see Table 5). In Domain I for example, the Comparative Fit Index for 14 of the 15 centres is above 0.9. In the Tokyo dataset, where the CFI feU below 0.9, aUowing error variances to covary for the energy and dependence on medication facets increased the CFI to .925. Similar error


covariances amongst variables increased CFI above 0.9 in the remaining domains where a few centres fell below the level of 0.9. These are outlined in detail in Table 6.

Table 6. Error variances allowed to covary in order to increase the CFI to above 0.9.

Domain Centre Facets where error variances allowed Resulting to covary CFI

Tokyo Energy and Dependence on .925 Medication

2 Beer Sheva Positive feelings and Negative .986 feelings

St Petersburg Body Image and Negative Feelings .916

3 Bath Personal relationships and Sexual .960 activity

4 Madras Safety and Home .950 Panama Safety and Home .937 Tilburg Environment and Home .911 Paris Environment and Home .952 Bath Safety and Environment .903 Harare Safety and Home, Environment and .910

Home

In order to carry out multiple sample analysis, centres were again split into two groups: centres 1-8 and centres 9-15. For all analyses, error covariances were included where appropriate for individual centres (as shown in Table 6), but parameter estimates from facet scores to domains were constrained to be equal across centres. Multiple sample analysis for most of the domains displayed appropriate CFIs above 0.9, suggesting parameter estimates to be invariant across centres. However, two analyses fell just short with CFIs below 0.9 for the Physical and Environmental domains for the second group of Centres. In both cases, freeing the constraints for the Harare centre for a single facet (Medication on the Physical domain, and Leisure on the Environment domain) increased the model fit to above 0.9. Table 6 also shows that for all domains the unconstrained models provide a better fit than do the constrained models. However, when constraints are released individually, it is only necessary to release a few of the constraints before the level of fit approaches that of the completely unconstrained models.

THE WHOQOL-BREF STUDY

The WHOQOL-IOO allows detailed assessment of each individual facet relating to quality of life. In certain instances however, the WHOQOL-100 may be too lengthy for practical use. The WHOQOL-BREF version has been developed to provide a short form quality of life assessment that provides summary scores for four domains rather than detailed scores at the facet level (WHOQOL Group, 1998b).


At a conceptual level, it was agreed by the WHOQOL Group that comprehensiveness ought to be maintained in any abbreviated version of the WHOQOL-IOO, by selecting at least one question from each of the 24 facets relating to quality of life. The most general question from each facet (Le., the item that correlated most highly with the total score, calculated as the mean of all facets) was chosen for inclusion in the WHOQOL-BREF. Individual items selected by this method were then examined by a panel to establish whether the putative items reflected the conceptually derived operationalisations of facets of quality of life. That is to say, they constituted a cohesive and interpretable domain, with good construct validity. The WHOQOL-BREF contains a total of 26 questions (see WHOQOL Group, 1998b). To provide a broad and comprehensive assessment, one item from each of the 24 facets contained in the WHOQOL-IOO was included. In addition, two items from the Overall quality of Life and General Health facet were included.

CONFIRMATORY FACTOR ANALYSIS OF THE WHOQOL-BREF STRUCTURE

Table 7. Centres included in development o/the WHOQOL-BREF.

Data from original pilot Data from original Data trom new centres of the centres field testing the field testmg the WHOQOL n WHOQOL-IOO n WHOQOL-IOO n Bangkok, Thailand 300 Bangkok, ThaIland 435 Hong Kong 856 Beer Sheva, Israel 344 Beer Sheva, Israel 464 LeIpzIg, Germany 527 Madras, IndIa 412 Madras, IndIa 567 Mannhelm, Germany 483 Melbourne, Australia 300 Melbourne, Australia 350 La Plata, Argentina 421 New Delhi, IndIa 304 New Delhi, India 82 Port Alegre, Brazil 82 Panama City, Panama 300 Panama City, Panama 117 Seattle, USA 300 Seattle, USA 192 Tllburg, The 411 Tilburg, The 799 Netherlands Netherlands Zagreb, Croatia 300 Zagreb, Croatia 96 Tokyo, Japan 286 Tokyo, Japan 190 Harare, Zimbabwe 300 Harare, Zimbabwe 149 Barcelona, Spain 303 Barcelona, Spain 558 Bath, England 319 Bath, England 105 St Petersburg, Russia 300 Paris, France 323 Total 4802 4104 2369

The hypothetical structure of the WHOQOL-BREF is based on a 4 domain solution with the addition of an additional higher order factor on which all four domains load. As noted previously, Confirmatory Factor Analysis is a method used to test whether data fit a hypothetical model. In both the dataset relating to the original pilot and the dataset relating to the field trial of the WHOQOL-100, an acceptable comparative fit index (CFI) was achieved when the data were applied to the four domain structure using confirmatory factor analysis (CFI = .906 and .903, respectively). The so-called "field trial" was designed to test various aspects of the instrument such as sensitivity to change, further psychometric properties, and so on with the new measure. In the dataset including new centres that field tested the


WHOQOL-IOO (see Table 7), the initial Comparative Fit Index was 0.870, suggesting that alterations to the model were necessary. When three pairs of error variances were allowed to covary (i.e. Pain and Dependence on Medication, Pain and Negative feelings, Home and Physical environment) and two items were allowed to cross-load on other domains (i.e. Safety allowed to load on the Global domain and Medication allowed to load negatively on the Environment domain), the comparative fit index increased to .901. This suggests that all data sets fitted the hypothetical four domain structure reasonably well.

COMPARISON BETWEEN WHOQOL-IOO AND WHOQOL-BREF SCORES

There were high correlations between domain scores based on the WHOQOL-IOO and domain scores calculated using items included in the WHOQOL-BREF. These correlations ranged from .89 (for domain 3) to .95 (for domain I).

Table 8. Internal Consistency Of The WHOQOL-BREF Domains.

Cronbach alpha Orig. data Field data New data (n = 4802) (n = 3882) (n = 2369)

Physical Health .82 .84 .80 Psychological .75 .77 .76 Social relationships* .66 .69 .66 Environment .80 .80 .80

Note. * = Only 3 items, therefore Cronbach alphas may not be reliable.

Internal Consistency

Cronbach alpha values for each of the four domain scores ranged from .66 (for domain 3) to .84 (for domain I), demonstrating acceptable internal consistency (see Table 8). Cronbach alpha values for domain 3 should be viewed with caution, however, because they were based on three scores (i.e. the Personal relationships, Social support and Sexual activity facets), rather than the minimum of four generally recommended for assessing internal reliability.

FIELD TRIAL STUDIES WITH THE WHOQOL-BREF

The WHOQOL-BREF analyses presented in the previous section were analyses primarily derived from the WHOQOL-IOO datasets, rather than based directly on data collected with the WHOQOL-BREF itself. One of the problems with such analyses is that they are likely to show closer concordance between the analyses than is likely to be obtained when entirely new datasets are analysed. In this section therefore preliminary analyses will be presented from data collected with the WHOQOL-BREF itself. These data have been collected in different field trials using different study populations (ranging from healthy individuals, to clinically depressed


individuals, to patients with cancer), and so are not collected within a consistent and systematic sampling frame from the different centres. Nevertheless, the data provide a further opportunity to examine cross-cultural factors using an internationally derived instrument.

Six centres have currently collected datasets with over 500 respondents using the WHOQOL-BREF. These centres are in Germany (N=2408), Israel (N=75 I), Norway (N=1047), Spain (N= 659), Japan (N=1453), and India (N=1456), which provide an interesting range of European and Asian cultures. Substantial sample sizes are of course recommended for the types of analyses involved in multivariate methods such as confirmatory factor analysis (e.g. Tabachnick and Fidell, 2001).

The analyses showed that when the hierarchical 4 domain structure was tested on the global data from the 6 centres combined and when tested on the 6 centres individually, the overall CFI of 0.871 was reasonably good. Only a few minimal changes, which allowed two or three variables to cross-load on other domains (e.g. Mobility to load on the Environment domain as well as the Physical health domain), brought the CFI to above 0.9.

The pattern however for the individual centres was somewhat more complicated. The individual CFI values for Germany, Israel, Spain, and India were similar to that for the overall analysis. Again, a few minimal changes to the model allowing two or three understandable cross-loadings bring the fit indices to above 0.9 for these centres, an approach that also worked for the Norwegian data. However, the poorest fit was obtained for the Japanese dataset where the CFI was 0.762. This value suggested that the four domain hierarchical model as it stands is not an acceptable model for the Japanese WHOQOL-BREF dataset, the question being therefore why this might be the case. Although earlier analyses of the Japanese WHOQOL-IOO data had suggested that the 4 domain solutions were satisfactory, even with these data there was evidence from the multiple regression analyses (see Table 3 above) that the domains accounted for a lower percentage of the variance in overall quality of life than for other centres.

In order to explore the problems for the Japanese dataset, the EQS program (using the Lagrange Multiplier Test) offers preliminary suggestions about for example which variables need to be loaded onto domains other than their own in order to improve the degree of fit. Examination including a number of understandable cross-loadings, together with the subsequent item regression coefficients, highlighted an interesting problem; namely, that 3 items were particularly problematic and higher loadings indicated domains other than their usual ones. These three items were: Energy (Do you have enough energy for everyday life) which normally loads onto the Physical domain but which seemed to load primarily onto the Psychological domain for Japan; Mobility (How well are you able to get around) which also loaded on the Psychological domain higher than on the Physical domain; and Mood (How often do you have negative feelings such as blue mood, despair, anxiety, depression) which loaded primarily on the Physical domain rather than on the Psychological domain. With these altered loadings, plus some linked error covariances being allowed to covary, then the general 4 domain structure fits at above 0.9 on the CFI. However, the key points raised by these analyses are the differences as well as the similarities in relation to cross-cultural issues. Although


there may be a more abstract level at which universality is a feature of quality of life like in the areas of emotion and linguistics (e.g. Power and Dalgleish, 1997), at a more fine-grained level there are importance differences between cultures as well. The finding for example that negative affect may be experienced in a primarily somatic form in Asian cultures and therefore load more on the Physical than the Psychological domains has been reported in epidemiological studies (e.g. Jenkins, Kleinman and Good, 1991); it provides direct evidence of the differences obtained both in our own and in other researchers' data.

DISCUSSION

The analyses presented in this chapter indicate that it has been possible to develop measures of quality of life that are reliable and valid for use in a range of cultures. The initial development of the pilot WHOQOL pooled input both at a conceptual level and in relation to specific items from culturally diverse centres. A general instrument was developed through an iterative process that included an agreed defmition of quality of life, agreed definitions of the facets, the generation of a large item pool reflecting those definitions, and, finally, an agreed set of items for the pilot WHOQOL.

The first phase analyses of the item response distributions, item-facet reliability analyses, and examination of item correlations with other facets showed that some items (as is usual) had to be eliminated based on psychometric criteria from the item pool. In addition, the item analyses suggested that some facets should not be retained in the field trial instrument either because responses were, for example, too skewed, or, in the case of the Activities as a Provider/Supporter and Work Satisfaction facets, because the facet demonstrated poor reliability and validity across cultures. It must be noted that although facets related to sensory functioning, communication, and burden of care for others have been dropped from the core WHOQOL-IOO, it would be possible to develop add-on modules designed for either specific populations (e.g. those with sensory or communication dysfunctions) or for specific cultures in which these items could be included, so long as they met criteria specified by the WHOQOL Group. The development of the core WHOQOL-IOO provides a first step in defining the core set of items needed to assess quality of life, but it is not intended to suggest that other aspects of quality of life should be excluded. For example, in some clinical studies it will be worthwhile to consider the addition of a diseasespecific or treatment-specific WHOQOL and/or national questions if these are culturally relevant.

The analyses presented in this chapter represent an intermediate stage in the development of the WHOQOL measures of assessment. The WHOQOL-IOO has been found to be a reliable and valid instrument that can be used in a diverse range of cultures. There are, however, a significant number of questions yet to be addressed, that were not part of the pilot testing of the instrument. One of the main limitations of the data presented here is that they are cross-sectional. Longitudinal data are now necessary to investigate the test-retest reliability of the instrument in populations who have not experienced significant life changes. In addition, it is necessary to collect longitudinal data from populations who have experienced

M.J. POWER, M. BULLINGER AND THE WHOQOL GROUP 147

significant life change and thereby assess the sensitivity or responsiveness of the instrument to change. Such studies are well under way in several centres with a variety of populations, but final published data are not yet available. Given the anticipated widespread use of the WHOQOL, it will be necessary to examine how a range of physical, psychological and social interventions impact on both general and specific aspects of quality of life. In addition, it is now necessary to put the WHOQOL-IOO and the WHOQOL-BREF further to the test as instruments in their own right (as opposed to as "extracted" ones). New datasets should be collected using other designs (e.g. clinical trials) as well as in a range of new cultures which were not represented in the first sets of centres. Such studies and extensions are currently underway and will be reported in due course. In the meantime, the WHOQOL-IOO presents an innovative approach and a major advance both in the background methodology used for the development of a reliable and valid crosscultural instrument, and in the provision of an instrument that measures a broad range of domains of quality of life.

FINAL COMMENTS AND CONCLUSIONS

The WHOQOL took as its starting point the classic WHO definition of health. Based on a number of influential conceptualisations of quality of life and related areas, an overall hierarchical structure was proposed that included Overall Quality of Life, a set of specific Domains, and specific Facets which reflect aspects of each of those Domains. The empirical work to date provides good support for such a structure, with confirmatory factor analyses offering support more for the four-domain than the six-domain hierarchical solutions. While we acknowledge criticisms of the WHOQOL-IOO scale (Hagerty et aI., 2001), more recent work is further refining the scale. Moreover, some matters are not in dispute, such as the hierarchical nature of quality of life such that a scoring system based on domain-level summary scores can provide a useful profile of scores, for example, in individual clinical use of the scales. In addition, it may be useful to produce an overall summative index from the scale. It is important to note though that the analyses show that the measure should not simply be used to provide an overall score, but that domain-level scores are also important.

The collection of data from a large number of different cultures has allowed the question to be asked of whether or not there is universality in the overall sense of well-being and quality of life. Although the term "quality of life" itself does not translate well into all languages, our analyses across a wide variety of cultures suggest that there are universal aspects of this concept that may well be linked in to other universals in areas such as language, emotion, and social relationships (e.g. Power & Dalgleish, 1997). This universality relates to the dimensionality of the quality of life concept. Regarding the initial item formulation, it was unexpected to find so much agreement across centres on the reduction from several thousand items down to the 236 of the Pilot WHOQOL. This universality continued in the domain and facet structure in which little cultural difference was found in the structural equation models. Comparable dimensionality however does not imply identity of WHOQOL profiles across persons and cultures; thus, the claims that there are


universal features of quality of life do not deny that there are also important differences between cultures. The claims that there are universals for language and emotion obviously do not mean that we all speak the same language or that we all experience the same emotions: the claims for universals in quality of life must be understood in the same way.

Address for correspondence: Professor Mick Power, Dept of Psychiatry, Royal Edinburgh Hospital, Edinburgh EHIO 5HF, UK; e-mail: m;@Srvl.med.ed.ac.uk

REFERENCES

American Psychiatric Association (1994) Dlagnosllc and Stallsllcal Manual of Mental Disorders (Fourth Edition). Author, Washington, DC.

Bentler, P. M., & Wu, E. J. C. (1995). EQSfor WindOWS Users Guide. Encino, Multivariate Software, Inc, CA

Bullinger, M (2001). The challenge of cross-cultural quality oflife assessment Psychology and Health (In press)

Bullinger, M., Power, M. 1., Aaronson, N. K., Cella, D. F., & Anderson, R. T. (1996) Creating and evaluating cross-cultural instruments. In B. Spilker (Ed.), Quality of Life and Pharmacoeconomlcs m Climcal Trials (2nd Ed.). Lippincott-Raven, Hagerstown, MD

Cummins, R.A. (1996). The domains oflife satisfaction: An attempt to order chaos. SOCial Indicators Research, 38, 303-332

Dupuy, H.J. (1984). The Psychological General Well-Being (PGWB) Index. In N K Wenger, M.E Mattson, C.D. Furberg. & J. EIinson (Eds), Assessment ofqualtty of life m cltmcal trials of cardIOvascular therapies. New York: Le Jacq.

EuroQOL Group (1990). EuroQOL - a new facility for the measurement of health-related quality of life. Health Poltey, 16. 199-208

Hagerty, M.R., Cummins, R.A., Ferris, AL., Land, K.. Michalos, AC., Peterson, M , Sharpe, A., Sirgy, 1., & Vogel, J (200 I). QUality oflife indexes for national policy Review and agenda for research. SOCial Indicators Research. 55. 1-91.

Jenkins, J.H., Kleinman, A and Good, B.J. (1991). Cross-cultural studies of depression. In 1. Becker & A KleInman (Eds.), Psychosocial Aspects of DepreSSIOn _Hillsdale. N.J. : Erlbaum

Kamofsky, D., & Burchenal, J (1949). The Climcal EvaluatIOn of ChemotherapeutIc Agents m Cancer New York: Columbia Press.

Maslow, AH. (1970). MotIVatIOn and Personalzty (2nd Ed.). New York: Harper and Row. Power, MJ., Bullinger, M., Harper, A, & WHOQOL Group (1999) The World Health Organlzahon

WHOQOL-100' Tests of the universality of qualIty oflife In 15 different cultural groups worldwide. Health Psychology, 18. 495-505.

Power, M.J., & Dalgleish, T. (1997). CogmtlOn and Emotion: From Order to DIsorder. Hove. Psychology Press

Sen, A (2001). Economic progress and health. In D.A. Leon & G. Watt (Eds.), Poverty. Inequality, and Health: An InternatIOnal Perspective. Oxford: Oxford Umversity Press.

Spilker, B. (1990). Introduction In B. Spilker (Ed ), Qualtty of Life Assessments m Clmlcal Tnals. New York: Raven Press.

Spilker, B. (1996). Introduction. In B.Sptlker (ed.), Quality of Life and Pharmacoeconomlcs m Clmlcal Trials (2nd Ed.). Lippincott-Raven, Hagerstown, MD.

Tabachnick, B.G. and Fidell, L.S. (2001). Usmg Muillvarlate StatistIcs (4th Ed.). Boston: Allyn and Bacon.

Torrance, G. W. (1996) Designing and conducting cost-utility analyses. In B.Spilker (ed.), Quality of Life and Pharmacoeconomlcs in Clzmcal Trials (2nd Ed.). Lippincott-Raven, Hagerstown, MD.

Ware, 1. E. (1996). The SF-36 health survey. In B Spilker (Ed.), Quality of Life and Pharmacoeconomlcs m Clzmcal Trials (2nd Ed.). Lippincott-Raven, Hagerstown, MD.

Ware, J.E., & Sherbourne, C.D. (1992). The MOS 36-item short-form health status survey (SF-36): I. Conceptual framework and item selection. Medical Care. 30. 473-483.


The WHOQOL Group (1995). The World Health OrganizatIon Quality of Life assessment (WHOQOL). Position paper from the World Health Organization. Social SCIence and Med,cine, 41, 1403-1409.

Psychological The WHOQOL Group (1998a) The World Health Organization Quality of Ltfe Assessment (WHOQOL): Development and general psychometric properties. Social SCIence and MedIcine, 46, 1569-1585.

The WHOQOL Group (1998b). Development of The World Health Organization WHOQOL-BREF Quality of Life Assessment. PsychologIcal MedIcine, 28, 551-558

World Health Organization (1958). The F,rst Ten Years of the World Health OrgamzatlOn. Geneva: World Health Organization.

GEORGIA POMAKI AND STAN MAES

PREDICTING QUALITY OF WORK LIFE: FROM WORK CONDITIONS TO SELF-REGULATION

Abstract. Quality of work has been frequently defined in terms of work conditIOns. Work conditions, described in theoretical models (as the Job Demand-Control-Social support model, the Effort-Reward Imbalance model and the Vitamin model) are presented as important predictors of wellnesslhealth outcomes. Although empirical findings have clearly illustrated the predictive power of these models, limItations and inconsistent results support the exploration of additIOnal, complementary perspectives. We suggest a person-centered, self-regulatory approach to quality of work life. Personal goals are presented as the core predictor of wellness and health. Within MotivatIOnal Systems theory (MST), personal goals help employees direct and organize behavior The strategies and processes involved in goal pursuit are predIctive of goal attainment The opportumty to attain goals or the frustration of one's goals is the key to health and wellness. Although there are several theoretical models and theories describing the cognitIve and emotional processes involved ID the pursuit of personal goals, empmcal research concentrating on such phenomena at the workplace has been scarce. An overview of studies investigating the relationshIp between personal goals and wellnesslhealth outcomes among employees is presented and discussed. Although most of the studies stem from different theoretical models, we focu~ on goal processes that are common with MST processes in an attempt to provide constructive and systematic conclusions. Goal processes were significantly predictive of wellness indicators and work -related outcomes in crosssectional as well as longitudinal studies. More attentIOn should however be paId on the assessment and operationalization of goals and the choice of the appropriate goal level. Personal goals and the processes involved in goal pursuit are certainly worthy of further investigation as part of sound, integrative models.

INDICATORS OF QUALITY OF WORK LIFE

The majority of people spend between one third and one half of their active life at work, According to some authors, people's quality of life is determined by their perceptions concerning their performance on four functional areas: physical and occupational function, psychological state, social interaction and somatic sensation (Schipper, Clinch, & Powell, 1990), Although not explicitly considered by these authors, the link between quality of life and quality of work life is obvious. From a conceptual perspective, CaIman (1984) was the first to suggest that quality of life is determined by the gap between the person's expectations and actual achievements, This defmition describes a factor that helps individuals shape their perceptions concerning their performance on the aforementioned functional areas, The definition can thus be applied to quality of work life, implying that not only objective indicators of professional achievement (as career and job performance) playa role, but so does the extent to which these achievements meet the person's needs and expectations. In other words, quality of work life is a subjective concept. A further distinction is important. While one can concentrate on outcomes of quality of work life - job satisfaction, intention to turnover, absenteeism, health related variables, burnout - in view of interventions it is equally important to identify predictors of these outcomes.

151

E. Gullone and R.A. Cummins (eds.), The Universality of Subjective Wellbeing Indicators, 151-173. © 2002 Kluwer Academic Publishers.

152 PREDICTlNG QUALITY OF WORK LIFE

Over the past two decades, there have been numerous investigations concerning the prediction of quality of work life, both in and outside of the discipline of psychology. We first turn to several literature reviews, which summarize the effects of job characteristics on work-related outcomes, psychological well-being, and physical health. We then consider the models that might help organize these data. This is followed by a person-centered approach related to the prediction of wellness and health at the workplace, namely Motivational Systems theory. Finally, we present and discuss empirical findings concerning the predictive value of personal goals, as the core component of this theory and give suggestions for future goal research on employee health.

Work-related outcomes, such as job satisfaction, intention to turnover, remaining in work, organizational commitment, job performance, absenteeism, and workrelated accidents have been significantly predicted by work characteristics (Lund & Borg, 1999; Roe, Zinovieva, Dienes, & Ten, 2000; Lu, Kao, Cooper, & Spector, 2000; Iverson & Maguire, 2000; Hui & Lee, 2000; Vahtera, Kivimaki, Uutela, & Pentti, 2000; Greiner, Krause, Ragland, & Fisher, 1998; Fried, BenDavid, Tiegs, A vital, & Yeverechyahu, 1998). In a review of absenteeism research, Harrison and Martocchio (1998) argued for a strong contribution of work characteristics in the absenteeism process. Absenteeism was also found to be significantly related, in a time period of a few months to a year, to various withdrawal behaviors and job performance. Studies investigating the relationship between work characteristics and psychological well-being have found strong associations with outcomes, including depression (Dwyer & Mitchell, 1999), self-esteem (Schonfeld, 2000), physical, psychological and social functioning (Stansfeld, Bosma, Hemingway, & Marmot, 1998; Sundquist & Johansson, 2000), emotional exhaustion (Feldt, Kinnunen, & Maunao, 2000), general stress levels (Rose, Jones, & Fletcher, 1998), mental health (Lu, Kao, Cooper, & Spector, 2000), psychiatric morbidity (Cropley, Steptoe, & Joekes, 1999), and accident proneness (Kirschenbaum, Oigenblick, & Goldberg, 2000). Finally, regarding physical health outcomes, Schnall, Landsbergis and Baker (1994) concluded in a review of job strain and cardiovascular disease (CVD) that there is considerable evidence to suggest a strong association between job characteristics and CVD outcomes, such as CVD morbidity and mortality, myocardial infarction, angina pectoris, coronary heart disease (CHD), and all-cause mortality. Tennant (2000) also concluded that job stress consistently predicts CHD events and CHD-adverse risk factors.

A review by Kasl (1996) on the same topic suggested that there has been considerable empirical support for the association between work-related factors and coronary heart disease, mainly in studies conducted within sound theoretical frameworks. Several recent studies have also supported the association of job characteristics with coronary heart disease events (Wamala, Mittleman, Horsten, Schenck, & Orth, 2000), and mortality due to coronary heart disease (Alterman, Shekelle, Vernon, & Burau, 1994).

In summary, in most of the cross-sectional, longitudinal and cross-cultural studies investigating the impact of adverse work conditions on employee wellness and health an association indeed has been confirmed. There is, however, little agreement with respect to which specific aspects of the work environment predict

G. POMAKI AND S. MAES 153

these wellness/health outcomes. As a consequence, it has been difficult to demonstrate how changes in the work environment, either positive or negative, can influence employee wellness and health. This suggests to us that improved and testable models that describe pathways between work conditions and outcomes are needed. We turn to such models in the following section.

MODELS OF QUALITY OF WORK LIFE

The J-DCS model

The Job Demands/Control/Social Support model (J-DCS), developed initially by Karasek (Karasek, 1979; Karasek & Theorell, 1990: J-DC) and later extended with the social support component by Johnson and Hall (1988: J-DCS) is one of the most influential models concerning occupational stress. It has been tested in over 100 empirical studies (c.f. Barnett & Brennan, 1995). The basic premise of the model is the strain and iso-strain hypothesis. which states that psychological strain and physical ill health can be predicted by the main effects of increased job demands, lack of control and lack of social support. According to the J-DCS model, the combination of demands and control result in four different classifications of jobs represented by the four quadrants of the model. These are labeled "high strain" jobs (high demands and low control), "low strain" jobs (low demands and high control), "active" jobs (high demands and high control) and "passive" jobs (low demands and low control) (Landsbergis, Schnall, Deitz, Friedman, & Pickering, 1992). Jobs with high control are hypothesized to be healthier, especially in low strain work environments, where one has both low demands and high control.

In addition, the model has two diagonals, the strain diagonal and the learning diagonal. Learning can inhibit strain, in such a way that resource-taxing situations are converted into opportunities for growth and learning, by the exercising of control over one's own work activities (Storr, Trinkoff, & Anthony, 1999).

Another way of looking at the model is to treat control and social support as buffer factors that reduce the detrimental effects of high demands on health outcomes (buffer-hypothesis). In this case, the interaction between demands, control and social support predicts the outcomes. To summarize, the J-DCS model has been described by means of three hypotheses: 1. The (iso-)strain hypothesis (demands, control, and social support predict strain). 2. The learning hypothesis (demands, control, and social support predict personal

growth). 3. The buffer hypothesis (control and social support buffer the negative effects of

demands on health).

In a review of 51 studies on the J-DC or the expanded J-DCS model and physical health outcomes, Van der Doef and Maes (1998) explored the first and third hypotheses. The review revealed that the "strain" hypothesis predominates in studies of all cause-related (i.e. mortality, cardiovascular disease related and other noncardiovascular disease) health outcomes, such as musculoskeletal symptoms and


pregnancy outcomes. In contrast, the "buffer" hypothesis proved to be most prevalent in research on self-reported psychosomatic complaints. In the case of both the strain hypothesis and the buffer hypothesis, the empirical findings are equivocal. Working in a high (iso-) strain job appears to be associated with elevated risk of cardiovascular disease and negative pregnancy outcomes, and a greater number of somatic and psychosomatic complaints. Conclusions concerning other physical outcomes seemed premature, considering the limited number of studies. The buffer hypothesis is supported in the few studies on cardiovascular disease outcomes and in some studies on somatic and psychosomatic complaints (Verhoeven, Maes, Kraay, & Joekes, in press).

In a second review of 63 studies Van der Doef and Maes (1999) explored the relationship between the J-DC(S) model and outcomes reflecting psychological wellbeing, such as depression, anxiety, job satisfaction, and burnout. The same two hypotheses were considered. While the literature provides considerable evidence supporting the strain and iso-strain hypotheses, support for the moderating influence of job control and social support is less consistent. The conceptualization of demands and control proved to be a key factor discriminating supportive from non-supportive studies. Only those conceptualizations of job control that corresponded to the specific demands of a given job were found to moderate the impact of high demands on well-being (Verhoeven et aI., in press).

Detailed methodological questions have been successfully raised in several reviews concerning the J-DC(S) model (Carayon, 1993; van der Doef & Maes, 1998; Fletcher & Jones, 1993; Karasek, 1989; Landsbergis et aI., 1992; Kristensen, 1995; Kristensen, 1996; Theorell & Karasek, 1996). One common critique is that the model explains only a small percentage of variance in health outcomes or measures of psychological strain. It is also criticized as being too simple. Inconsistency of evidence for the J-DCS model may also be attributed to a number of factors: inappropriate interaction terms used; inadequate operationalization of the concepts of job demands, control and social support; the use of occupational level analysis in some studies and the use of homogeneous samples in others; overlooking employment grade issues; gender differences; the possible moderating effects of personality variables that determine the vulnerability to stressors or the ability to benefit from the buffer factors. Finally, methodological shortcomings have been suggested, including a conceptual overlap between the stressor and the outcome variable, and possible common-method variance effects (van der Doef & Maes, 1999). Searle, Bright, and Bochner (1999) concluded that lack of support of the buffer hypothesis, "may reflect problems with the theoretical model", especially due to a potential socioeconomic bias.

Although there is a plethora of research investigating the relationship between the model's three basic components with a variety of outcomes, there is serious lack of research concerning the underlying mechanisms. The three model components could be related to basic human needs, the attainment of which is a prerequisite to effective human functioning. According to the Self-determination theory (Deci, Connell, & Ryan, 1989), there are three basic needs that individuals strive to fulfill -competence, autonomy, and relatedness. Demands and competence, control and autonomy, and social support and relatedness could be related constructs. The J-DCS


model components may therefore be related to basic human needs, and derive their central significance in employee quality of work life implicitly from these relationships. This hypothesis may be worthy of further investigation.

The Effort-Reward Imbalance model

The Effort-Reward Imbalance model (ERI) that was developed by Siegrist (1996) is composed of personal as well as environmental components. The model encompasses broader psychosocial parameters that could explain contemporary workplace processes and phenomena and bring new insights to the J-DCS model (Theorell et aI., 1996). The aim of the ERI model is to identify work-related stressors as well as coping strategies that can predict strain and ill health. In the most recent version, the model specifies the environment and the person as the two sources of stress (Peter & Siegrist, 1999). Regarding the role of the work environment, the imbalance between high efforts spent at work and low rewards obtained predicts sustained stress experiences and consequently adverse health outcomes. Efforts are defined as the strivings of the individual to meet the demands and obligations of the job. Rewards can be obtained by means of three systems: money, esteem, and career opportunities, including job security. Regarding the role of the individual, the coping pattern of overcommitment is introduced as a possible source of strain, as it represents the tendency towards excessive striving, motivated by an overarching need for approval and esteem.

However, it is assumed that the goal of the individual· is to gain equilibrium and/or reduce stress, so effort-reward imbalance cannot be sustained in the long run. Peter and Siegrist (1999) emphasize that under certain conditions overcommitted individuals tend to persist investing in their jobs, independent of their high effort-low reward status, a condition that will eventually predict ill health. In a series of studies, Siegrist and his co-workers investigated the relationship between the model components and health outcomes. Using cardiovascular risk or disease as the health outcome, five studies have fully or partially supported the model (Peter et aI., 1999).

The Vitamin model

Another model that has been linked to the J-DCS model and, by introducing the concept of non-linearity, provided a useful perspective, is the Vitamin model developed by WaIT (1990). Here, just as vitamins can be beneficial for physical health up to, but not beyond a certain level, so too can job characteristics. On the basis of this assumption, the observed associations between traditional job characteristics, such as job demands, control and social support, and well-being are expected to be non-linear, with decrements in well-being being found at extremely low or high values of the job characteristics. According to the Vitamin model, job characteristics are differentially related to employee health, and their effects can be dependant on employees' individual characteristics, such as abilities. preferences and goals. Though this concept has not been a crucial part of the model, WaIT has in a way introduced the idea of individual differences in the traditional work stress research. In line with the suggestions by WaIT (Vitamin model) and Siegrist (ERI


model and the role of overcommitment) on the importance of individual differences, is the notion that a match between the stressor and the buffering factor might improve the predictive power of control as a buffering factor. Van der Doef and Maes (1999) concluded that support has been provided for the buffer hypothesis in studies where demands and control were operationalized at a comparable level of specificity. Empirical evidence has been inconsistent concerning the match suggestion (Dormann & Zapf, 1999; Kivimaki & Lindstrom, 1995; Searle, Bright, et aI., 1999; van der Doef, Maes & Diekstra, 2000). Perhaps, however, the matching of the buffer factor to the stressor requires a more individualized approach. It would be interesting to investigate how different individuals use different work characteristics as buffer factors against the same stressors. For example, for one employee time pressure may be balanced by the opportunity to expand his/her knowledge and skills, whereas for another more tangible rewards are needed, such as a good salary and benefits.

Individual differences, such as the work-related coping style of overcommitment, work-related self-efficacy and personal goal attainment possibilities could also act as buffer factors in contemporary workplaces (De longe, Bosma, Peter, & Siegrist, 2000; Gardner & Pierce, 1998; Roberson, Korsgaard, & Diddams, 1990). From this perspective it seems that the l-DCS model emphasizes only one of two possible sources of variance in emotional, behavioral and physiological arousal. The first source, considered above, refers to the impact of the work environment and work characteristics on employee wellness and health. The second source of variation in wellness and health is concerned with the individual, and hislher role at the workplace, and here motivational aspects of work and employee life, and the opportunity to set and attain personal goals may be important.

SELF-REGULATION: AN APPROACH TO QUALITY OF WORK LIFE

Personal goals as motivational sources

While traditional occupational stress research has concentrated on the influence of the work environment's pressures on the individual, little attention has been paid to the pressures that the individual sets on himlherself, and the processes that are involved in addressing these pressures. Both of these aspects are related to the concept of personal goals.

Ford (1992) has defined goals as thoughts about desired consequences that the individual would like to achieve, or undesired consequences that the individual would like to avoid. Personal goals constitute strong motivational components that drive human behavior. Goal pursuit includes psychological processes that are responsible for the initiation, direction, and persistence of behavior and provides the foundation for learning, skill development, and behavior change (Ford, 1992).

Personal goals can be seen as a set of intrinsic demands that are posed on the individual, or that the individual poses on him/herself. Research has shown that personal goals and the processes that are related to goal pursuit are significantly and consistently predictive of people's behavior and health (Brunstein, 1993; Emmons,


1992; Affleck et al., 1998). However, more research on goals and health especially in a work-related context is necessary.

The nature of the relationship between environmental pressures (i.e., work characteristics) and individual-imposed pressures (i.e., personal goals) is of special interest. Although this topic has not - to our knowledge - been explicitly investigated, we certainly encourage research that can shed light on the exact way these two parameters can interact in order to predict employee health and performance at work. Models such as the J-DCS model and the ERI model could be linked with personal goal attainment. Along these lines, we would like to suggest that an adverse work environment characterized by work overload, lack of autonomy and social support may hinder the setting and pursuit of personal goals. Adverse work characteristics can therefore influence health directly and indirectly, by reducing opportunities for employees to attain their goals. Another example of an adverse work environment is one characterized by the imbalance between employees' invested efforts and received rewards. In this case, employees may invest their energy and efforts in their attempts to fulfill the organization's standards in order to acquire the desired rewards. Two consequences may result from such sustained behavior; first, ill health is predicted; second, the pursuit of personal goals is hindered. Again, the work environment can influence employee health directly and indirectly through personal goal frustration. The relationship between personal goal attainment and job stress models may be worthy of further investigation.

MOTIVATIONAL SYSTEMS THEORY

Model-based research is important for our understanding of quality of work life. Hence, motivational aspects of workplace phenomena cannot be thoroughly investigated in the absence of a relevant theory. Self-regulation theories could provide the framework for the investigation of motivation. Though a number of selfregulation theories exist, to our knowledge no model has been both specified and tested explicitly in a work-related context. According to Maes and Gebhardt (1999, p. 344), "self-regulation can be defined as a sequence of actions and/or steering processes intended to attain a personal goal". These actions or processes have been described by Karoly (1993, p. 25): "self-regulation may be said to encompass up to five interrelated and iterative component phases: (I) goal selection, (2) goal cognition, (3) directional maintenance, (4) directional change or reprioritization, and (5) goal termination".

While Control theory (Powers, 1973; Carver & Scheier, 1982) could also be used as a framework for this purpose, here we concentrate on the application of Motivational Systems theory (Ford & Ford, 1987). Motivational Systems theory (MST) is an integrative framework that describes, "how motivational processes interact with biological, environmental, and nonmotivational psychological and behavioral processes to produce effective and ineffective functioning in the person as a whole" (Ford, 1992, p.12). Its major advantage is that it integrates a number of concepts derived from theories of self-regulation into a comprehensive conceptualization of self-regulatory processes (Karoly, 1993).


Personal goals as a strong motivational force are of great importance in this theory. A taxonomy of goals has been developed by Ford and Nichols (Ford et al., 1987). This is a standardized classification scheme that can facilitate the process of classifying individuals in relation to their goals, and that allows one to make comparisons across individuals and social groups. The taxonomy concentrates on goals as desired consequences of behavior and not as behaviors themselves. At the highest level of abstraction, the taxonomy is divided into two types of goals: goals that represent desired consequences within individuals, namely affective, cognitive, and subjective organization goals, and goals that represent desired consequences with respect to the relationship between people and their environments, namely social relationship and task goals.

Human functioning (see Figure I) is regulated by four general functions: (a) biological functions; (b) governing (directive) functions, which include directive, regulatory, control cognitions; (c) arousal functions, which include emotional, attention, consciousness and activity arousal processes; and (d) transactional functions, or motor, communicative, ingestive, eliminative, and sensory perceptual actions, as well as environmental components.

In general, the directive function determines the intentions and the goals to be set. The processes of goal choice and goal attainment often require evaluative thoughts, such as self-evaluative thoughts, standards, rules, and values. The regulatory cognitions contribute to the selection among different options and the evaluation of information that can lead to goal attainment. The control cognitions connect current information with the person's knowledge, skills, and capabilities, by means of problem formulation, problem solving and plan execution processes. The arousal functions regulate the individual's emotional and other resources. Finally, the transactional functions connect the person with the environment (Ford, 1992).

Ford and Nichols (1991) argued that there are three mechanisms for cognitive regulation: (I) feedback mechanisms that monitor and evaluate progress towards goal attainment, (2) feedforward mechanisms that regulate expectancies about one's capabilities for effective functioning and about the responsiveness of the environment, and (3) activation of control processes that develop plans and problemsolving strategies.

Within this framework, intentions and goals are formed and organized in two ways: as a nested hierarchical structure, where a goal can serve as a means to the attainment of other goals, and as a value hierarchy, where goals are ordered according to their importance and value to the person. Effective functioning requires a strategic approach of attainable short-term goals in combination with a continuous overview of the long-term goals that can give meaning to current behavior. Another interesting issue refers to goal alignment. When alignment among goals is reached, motivation is significantly enhanced, as the person has more than one reason to exert a specific behavior episode. On the other hand, goal contlict can have detrimental effects on the individual's motivational status.

G. POMAKI AND S. MAES

Governing (COGNITIVE) FUNCTIONS

Goal evaluation processes: Goal relevance Goal importance Goal attainability Emotional salience

Feedforward Feedback cognitions

Information processing

AROUSAL FUNCTIONS

BIOLOGICAL FUNCTIONS

Figure I. Motivational Systems theory (Ford. 1992): Functions

159

Apart from the governing functions that organize cognitions and specify actions, human behavior, dynamic at its core, is subjected to change, self-developing, and self-maintaining processes. According to the Living Systems Framework (LSF; Ford & Ford, 1987) a number of processes illustrate these characteristics: (a) change and development processes, (b) stability maintaining processes, (c) incremental change processes and (d) transformational change processes.

Ford and Nichols have outlined several dimensions along which people may vary in their general orientation to goal setting: (a) active-reactive, (b) approachavoidance, and (c) maintenance-change. Empirical evidence regarding the approachavoidance goal orientation, where a person may tend to conceptualize goals as


positive consequences to be reached or negative consequences to be avoided has shown a significant relationship between these goal orientation and well-being outcomes (Elliot & Sheldon, 1997).

At a situational level of analysis, effective functioning is represented by achievement. Achievement is defined as the attainment of a personally or socially valued goal in a particular context. At the personality level of analysis, effective functioning is represented by competence. Competence is defined as the attainment of relevant goals in specific environments, using appropriate means and resulting in positive developmental outcomes.

Achievement is the outcome of a goal-directed behavior that is combined with affective energizing, possession of relevant skill and capability beliefs, and a responsive environment. All of these aspects are necessary for achievement to occur. If one of them is missing, goal attainment becomes difficult.

Effective functioning is central in MST and is anchored directly to motivational processes, such as feedback and feedforward mechanisms. To say that behavior is regulated by feedback mechanisms is to assume the existence of reference standards for behavior, which in most theories of self-regulation are identified with goals. Goals are therefore a fundamental component in most motivation theories (Carver & Scheier, 1982; Emmons, 1989; Cropanzano, James, & Citera, 1993; Klein, 1989; Edwards, 1992; Hyland, 1988).

In the previous section several theoretical points have been made. Work conditions have been consistently shown to predict well-being, health and workrelated outcomes. In view of interventions, theoretical models have described the specific pathways between work conditions and outcomes. Although the Job Demands-Control-Social support model concentrates on the importance of work conditions, related models (Effort-Reward Imbalance model, Vitamin model) have successfully introduced the individual as another source of stress. Even within the JDCS model research has shown that a more individual-focused approach can raise its predictive power. In other words, there is strong theoretical and empirical evidence suggesting that there may be two major predictors of employee health: the work environment as well as the person. In search of a theoretical model that can help organize research, a self-regulatory perspective has been presented. More specifically, Motivational Systems theory, a broad integrative framework, focuses on how individuals strive to pursue their personal goals within a specific context. Personal goals act as motivating forces, as internal demands. The content of goals, the processes involved in goal pursuit and finally goal attainment or frustration may be crucial for employee health. Personal goals research may provide a personcentered approach and improve the prediction of wellness/health outcomes. In the following section, we examine the empirical evidence regarding the link between goals and wellness/health within the work context and give suggestions for future research.

PERSONAL GOALS AT THE WORKPLACE

There has been considerable theoretical expectation, though less empirical evidence for self-regulation in the workplace. Conceptual diversity and methodological


shortcomings have inhibited a systematic investigation of goals and self-regulation. In the following section, we will discuss studies on personal goals that are workplace-related. The purpose of this overview is to examine the present empirical evidence concerning personal goals and personal goal processes and to provide insight and suggestions for future research. Research into personal goals is growing. An organized approach to the existing attempts at investigating goals may start a fruitful discussion on such topics as the accurate and meaningful assessment of goals, the selection of the appropriate goal level, the differentials across populations, and the use of relevant models.

The studies were selected from different journal databases (PsychLit, Current Contents), from the literature lists included in relevant articles and from personal communication with researchers in the field covering a period of fifteen years (1986-2001). The criteria for this selection were the following: (a) goals were assessed among a working population, (b) these goals had to be personal goals of the employees and not imposed organizational goals, (c) the association between personal goals and some outcome variable was investigated (studies that examined only the content of goals across the life span or in relation to gender and age were not included), and (d) studies were non-experimental. In total, 12 studies fulfilled the criteria (Barrick, Mount, & Strauss, 1993; Brown, Cron, & Slocum, 1997; Christiansen, Backman, Little, & Nguyen, 1999; Karoly & Ruehlman, 1996; Leithwood, Menzies, Jantzi, & Leithwood, 1996; Noe, 1996; Phillips, Little, & Goodine, 1997; Probst, Baxley, Schell, Cleghorn, & Bogdewic, 1998; Roberson, 1989; Roberson, 1990; VandeWalle, Brown, Cron, & Slocum, 1999; O'Neill & Mone, 1998).

A close look at these studies revealed that they focused on the predictive power of goal processes in relation to two distinctive outcomes: (l) wellness and (2) workrelated outcomes. Two central issues will be discussed in relation to the studies: (a) the theoretical constructs that they investigated, and (b) the methodology that they followed in assessing personal goals. It is important that studies on personal goals result from and feed back to theory and models. The studies that are reported here stem from different theoretical backgrounds and research areas, such as work values, performance related goal-setting, leadership, career management, expectancy value theory, equity theory, and personal projects models. In order to assess the studies within a common framework, the constructs that have been investigated in these studies will be linked to the main components of the Motivational Systems Theory. For this purpose we will first describe the studies and then attempt a link to MST (see Table I for an overview of the studies). With regard to methodology, there seem to be three goal elicitation procedures (see below), and we ask whether there is an association between the choice of method for goal assessment and the success in predicting the study outcomes.

Tabl

e 1.

Ove

rvie

w o

f em

piri

cal s

tudi

es

Em

piri

cal s

tudi

es

Goa

l rel

ated

var

iabl

es

Com

para

ble

vari

able

s M

ST

~ b

y th

e em

piri

cal f

indi

ngs

1. B

arri

ck, M

ount

, et a

l (19

93)

Aut

onom

ousl

y se

t go

als*

; G

oal

Dir

ecti

ve c

ogni

tion

s co

mm

itm

ent

2. B

row

n, C

ron

et a

l (19

97)

Pos

itiv

e &

. ne

gati

ve

anti

cipa

tory

em

otio

us;

VoH

tiou

s;

Goa

l-dl

reet

ed

beha

vior

; G

oal

att

ain

men

t 3.

Chr

isti

anse

n, B

ackm

an,

et

Goa

l co

nflic

t;

Mea

ning

; S

tres

s;

Em

otio

nal a

rous

al

Beh

avio

ral o

utpu

t T

rans

acti

onal

act

ions

E

mot

iona

l aro

usal

al

(199

9)

Com

mun

ity;

Str

uctu

re;

Eff

icac

y; P

eno

nal

D

irec

tive

cog

niti

ons

Iden

tity

; P

osit

ive

impa

ct;

Neg

ativ

e R

egul

ator

y co

gnit

ions

im

pact

; P

rou

ess

4. K

arol

y &

Rue

hIm

an (1

996)

V

alue

; S

elf-

effi

cacy

; S

ocia

l co

mpa

riso

n;

Sel

f-m

onit

orin

g;

Plan

ning

; S

elf-

rew

ard;

S

elf-

erit

idsm

; G

oal

eonO

iet;

P

osit

ive

arou

sal;

Neg

ativ

e ar

ousa

l

Reg

ulat

ory

cogn

itio

ns

Dir

ecti

ve c

ogni

tion

s E

mot

iona

l aro

usal

5. L

eith

woo

d, M

enzi

es,

et a

l C

apab

ilit

y be

lief

s;

Con

text

be

lief

s;

Reg

ulat

ory

cogn

itio

ns

(199

6)

Em

otio

nal a

rous

al

6. N

oe (1

996)

C

aree

r go

al f

ocus

; C

aree

r go

al s

trat

egie

s;

Reg

ulat

ory

cogn

itio

ns

Dls

tane

e fr

om e

aree

r go

al

7. 0

'Nei

ll &

Mon

e (1

998)

G

oal-

rela

ted

dead

line

s; G

oal c

lari

ty

Dir

ecti

ve c

ogni

tion

s

8. P

hilli

ps, L

ittl

e, e

t al (

1997

) G

oal

cont

ent;

Goa

l ch

alle

nge;

Eqj

oym

ent;

R

egul

ator

y co

gnit

ions

C

ontr

ol;

Sup

port

b

y

othe

rs;

Su

pp

ort

/Hin

dra

nee

by

th

e or

gani

zati

on;

Com

mit

men

t;

Str

ess;

O

ther

s'

view

o

f im

port

ance

; Se

nse

of c

ompe

tenc

y

Out

com

es

Job

perf

onna

nce

Pos

itiv

e ou

tcom

e em

otio

ns N

egat

ive

outc

ome

emot

ions

W

ell-

bein

g

Dep

ress

ion

Anx

iety

Bum

out

Dev

elop

men

tal

beha

vior

Jo

b pe

rfor

man

ce

Job

sati

sfac

tion

O

rgan

izat

iona

l co

mm

itm

ent

Inte

ntio

n to

turn

over

Jo

b sa

tisf

acti

on

?J '"0 1':l " n -l Z

a ;;:)

c: ~ ::j

-< o "Tl ~ C

"Tl

ttl

9. P

robs

t, B

axle

y, e

t al (

1998

) O

ppor

tuni

ty fo

r go

al a

ttai

nmen

t R

egul

ator

y co

gnit

ions

10. R

ober

son

(198

9)

Pro

babW

ty

of

succ

ess;

S

orro

w;

Joy;

D

irec

tive

cog

niti

ons

Inst

rum

enta

lity

; U

nhap

pine

ss;

Res

pons

e E

mot

iona

l aro

usal

de

pend

ency

; T

Ime

avai

1abl

e;

Goa

l ne

arne

ss;

Con

ting

eney

; G

oal

com

mit

men

t II

. Rob

erso

n (1

990)

G

oal

cont

ent

(pos

itiv

e V

I. n

egat

ive

goal

s G

oal c

on

ten

t &

ta

sk

VI.

no

n-ta

sk);

P

roba

bWty

o

f D

irec

tive

cog

niti

ons

succ

ess;

Neg

ativ

e co

nseq

uenc

es o

f go

al

atta

inm

ent;

Goa

l co

mm

itm

ent;

Lac

k o

f de

adli

nes;

T

ime

unti1

go

al

atta

inm

ent;

R

ole

dim

ensi

ons

(mea

ns);

Ins

trum

enta

lity

; T

ime

avai

labl

e: P

roba

bili

ty o

f suc

cess

with

ou

t act

ion

12.

Van

deW

alle

, B

row

n, e

t al

G

oal

orie

ntat

ion;

Goa

l le

vel;

Int

ende

d G

oal o

rien

tati

on

(199

9)

effo

rt; I

nten

ded

plan

ning

C

ontr

ol c

ogni

tion

s

* The

var

iabl

es in

bol

d in

dica

te th

at th

ey s

igni

fica

ntly

pre

dict

ed th

e st

udy

outc

omes

Tea

chin

g qu

alit

y

Goa

l-di

rect

ed

beha

vior

Job

sati

sfac

tion

Job

perf

orm

ance

0 ."

0 :s: ~ ~ c::l r.n s:: ~ Vl

0'1

W


OVERVIEW OF THE STUDIES

Goals and work-related outcomes

The majority of the studies (eight of twelve) investigated work-related outcomes (job performance, job satisfaction, organizational commitment, intention to turnover). The study by Barrick, Mount, et al. (1993) found a mediating effect of autonomous goal-setting and goal commitment on the relationship between conscientiousness and performance on the job. Goal commitment was, however, a weaker predictor. Noe (1996), in a study on career goals and their relationship with employee development and job performance, found that although career goal focus and career goal strategies were not related to developmental behavior and job performance, after controlling for relevant study variables distance from career goal predicted positively developmental behavior. O'Neill and Mone (1998) found that having deadlines and setting specific and clear goals was predictive of higher job satisfaction, and lower intention to turnover. No association was observed with organizational commitment.

Phillips, Little, et al. (1997), in a paper concerning gender issues in public administration, discussed research findings from a study on the relationship between managers' personal goal processes and job satisfaction. Perceived supportiveness of the organizational culture towards the managers' personal goals was strongly related to job satisfaction for women, although not for men. Perceived hindrance of personal goals provided by the organizational culture was associated with lower job satisfaction for male managers. No information on the statistical analyses is offered. Probst, Baxley, et al. (1998) found that after controlling for a number of organizational characteristics, participants who reported that their organizations gave employees the opportunity for autonomy and goal attainment also indicated higher job performance (teaching quality). Roberson (1989; 1990) conducted two studies investigating work-related personal goals and their association with job satisfaction and goal-directed behavior, using the Work Concerns Inventory (WCI), a thorough goal elicitation and appraisal instrument.

In a study of 172 employees (Roberson, 1989), goal commitment, contingency and time available appeared as significant predictors of goal-related behavior. Interesting relationships were also observed among the goal processes, including the central role of goal commitment. In another study of 150 employees (Roberson, 1990) job satisfaction was significantly predicted by several goal processes such as perceived probability of success, lack of deadlines and proportion of avoidance goals, after controlling for demographic variables (i.e., tenure, age, education and job level). Additional analysis was conducted for two different types of goals, task versus non-task goals. The results indicated the importance of including both task and non-task goals in goal studies. In addition, according to Roberson, a distinction should be made between approach and avoidance goals. The pursuit of a higher number of avoidance goals, in comparison to approach goals seemed to be predictive of decreased job satisfaction. In general, the best predictor of job satisfaction was perceived probability of success.


VandeWalle, Brown, et al. (1999) investigated the association between goal orientation (learning versus performance) and goal processes (intended goal level, intended effort, intended planning) and actual goal-related behavior (sales, objectively measured) among salespeople. All goal-related processes were predictive of actual sales reported. In addition, goal processes mediated the relationship between goal orientation and sales performance. Learning goal orientation was the only predictor of goal level, and intended effort. Learning, as well as performance goal orientation predicted, contrary to the study's hypothesis, intended planning.

Goals and wellness outcomes

Four studies focused on general wellness indicators (anxiety, depression, well-being, burnout). Brown, Cron, et al. (1997) found a series of associations linking goal processes (anticipatory emotions, anticipated and actual strategic planning and effort, goal attainment) to positive and negative emotions. The strengths of this study lie in the longitudinal design and the use of an objective measurement of goal attainment. A limitation is that personal goals were operationalized as personal stakes, i.e., the impact that the respondents expected their job performance (on a specified task) to have on professional and family aspects of their life. Christiansen, Backman, et al. (1999) investigated personal goal processes as predictors of subjective well-being, by using a goal elicitation procedure suggested by Little calIed Personal Projects Analysis (PPA; 1989). Stress, induced during goal pursuit, positive impact, personal identity and progress were the most significant predictors of subjective well-being in this study. Karoly and Ruehlman (1996) investigated the association between personal goal processes, differential pain experience, and well-being among managers, by means of the Goals Systems Assessment Battery (GSAB), an instrument based on Ford's Motivational Systems Theory. Contlict between work and non-work goals explained the difference between the persistent pain group and the episodic or no-pain groups. Goal processes (self-criticism, goal contlict, negative and positive arousal) explained considerable variance in anxiety, but less so in depression, in addition to the pain-related variables.

Finally, Leithwood, Menzies, et al. (1996) used another goal processes instrument based on Ford's Motivational Systems Theory, measuring personal goals and four consequent processes (i.e., capability beliefs, context beliefs, emotional, and arousal processes). According to the findings, leadership had both a direct and indirect association with burnout, through organizational support and personal goal processes. Teachers' goal cognitions (context and capability beliefs) were significantly related to burnout, while personal goals were negatively related to burnout. Emotional arousal processes were associated with capability beliefs, but were unassociated with burnout. One of the limitations of this study lies in the operationalization of the personal goals construct, which could represent the degree of adherence to organizational goals, rather than the individuals' personal goals. It could also be suggested that here might be a mediating and/or reciprocal relationship between capability beliefs and emotional processes that is worthy of further investigation.


EMPIRICAL SUPPORT FOR THE THEORETICAL CONSTRUCTS

Motivational Systems Theory focuses on goal content, goal hierarchies, goal-setting (e.g., short-term vs. long-term goals) and change processes, goal evaluations (e.g., importance, attainability, emotional salience), mechanisms of cognitive regulation (e.g., feedback and feedforward) and standards for goal attainment. The cognitive (directive, regulatory and control cognitions), emotional (emotional, attention, consciousness and activity arousal processes), and transactional functions (motor, communicative, ingestive, eliminative, and sensory perceptual actions) determine the processes involved in human functioning. Capability beliefs and context beliefs are of central importance to goal pursuit and goal attainment. MST also identifies inherent and more stable patterns, such as goal orientations, personal agency belief patterns, emotional patterns, and non-emotional affective states.

The constructs that have been supported by the empirical findings reported here, in terms of their predictive power of the study outcomes, refer to processes involved in directive, regulatory, and control cognitions, emotional arousal, goal orientations, and goal content. Other MST constructs- such as transactional functions, goal standards, goal hierarchies, goal-setting and change processes, and cognitive regulation mechanisms have not yet been tested.

These goal processes were predictive of general wellness indicators (i.e., depression, anxiety, burnout), and work-related outcomes (job satisfaction, job performance, organizational commitment, and intention to turnover). However, a more careful examination of the empirical findings showed a lack of consistency in the prediction of outcomes. In most studies only a small number of the hypothesized relationships between goal processes and outcomes were supported by the data. Reasons for the inconsistency of the results can be found in the assessment and operationalization of personal goals and the type and level of goal under investigation.

There are also several interesting issues that emerge from the empirical studies. Firstly the relationship between goal processes and outcomes may change with various types of goals. Roberson (1990) suggested that both task and non-task related goals should be taken into account when investigating personal goals at the workplace; it is possible that different processes are involved in the pursuit of task versus non-task goals.

Secondly, there may be a differential role of approach and avoidance goals. Approach goals represent striving towards a desired outcome (i.e., things an individual strives to achieve). Avoidance goals represent striving away from an undesired outcome independent of whether they are positively or negatively formulated (i.e., things an individual strives to abolish, avoid, or prevent from happening). According to Roberson (1990), employees with more avoidance goals reported decreased job satisfaction. It is worthy of note that there is some additional research, mainly among college students, that has shown that a higher percentage of avoidance goals is associated with lower well-being (Elliot, Sheldon, & Church, 1997). A higher number of approach goals has also been observed to be associated with lower scores on depression (Coats, Janoff-Bulman, & Alpert, 1996). Avoidance goals have also been associated with unfavorable goal processes: such goals have


been rated higher on difficulty than approach goals, less important, and as having brought less happiness when accomplished (Coats et aI., 1996).

Thirdly, in the studies that have already been discussed and in many other goaloriented investigations that do not fall within the area of work-related goals, different goal terms and constructs have been assessed. One difference is how broadly they have been conceptualized. Goal constructs such as life tasks ("a problem you are working on at this moment of your life"), and personal strivings ("something you are trying to do") are more vaguely defined, while current concerns (when you have decided to do something but you are not sure whether you are going through with it) and personal projects (taking up actions that may lead you to attain your goal) refer to more specific life events and action taken relevant to these events. Another difference lies in the time perspective. A part of goal constructs concerns goals that are meant to be attained within a limited period of time, when others are projects of a lifetime. The time factor can be facilitative in research, especially when designing goal assessment instruments. It is important to know what is the connection between a specific goal construct used in a study and time.

Another paragon that differentiates goal constructs from each other is the position they employ in the goal hierarchy. Life tasks, for example, refer to the core self, problems that individuals are working on across the lifespan, and can be more relevant in critical life periods. On the other hand, current concerns are goals that demand a person's energy and efforts on an everyday basis, and although important, they refer to goals that are lower in the goal hierarchy, closer to action units.

GOAL ELICITATION

As mentioned in the introduction of the goal studies, there are three ways of assessing personal goals at the workplace. One way includes a goal elicitation process, where respondents are asked to state and evaluate their personal goals on a number of goal processes. Another has been to directly ask respondents to evaluate their goals, without first requiring explicit goal formulation. The disadvantage of the latter method is that goals can often be organization-imposed, rather than selfdefined. In this case, goal evaluations can be biased. The third method may be seen as a compromise between these two. It involves asking participants to select goals that are relevant to them from a specified list. Participants can also formulate goals that are not included in the list. Again, what usually follows is another group of questions investigating relevant goal processes.

The goal elicitation procedure is important. All five studies, which used such a procedure, found strong associations between the goal-related processes and the study outcomes, while of the five that used no elicitation procedure, but only an evaluation procedure of the goal-related processes, four found but a few significant associations between goal processes and outcomes. In the two studies assessing goals using an already specified list, only a few associations were observed between goal processes and outcomes.

In conclusion, it seems that the inclusion of a goal elicitation procedure in assessing personal goal processes can result in a more powerful and consistent prediction of outcomes. Assessment methods that do not include an elicitation

168 PREDICTfNG QUALITY OF WORK LIFE

procedure depend heavily on participants' memories of the goals they have set. Lastly, list-based procedures, although more helpful to the participants, can be indoctrinating, in the sense that participants tend to adopt the goals presented to them.

CONCLUSION

The above discussion of the studies on personal goals at the workplace allows several conclusions. First, in both cross-sectional and longitudinal studies, considerable empirical support has been found for several goal constructs and goal processes, in terms of their predictive power of wellness and work-related outcomes. Goal variables explained significant variance in the study outcomes, in some cases over and above both personality variables and work characteristics. Inconsistency in the empirical support of these constructs can be attributed to goal elicitation methods, the use of various goal constructs, lack of attention to goal content, and a limited use of a sound theoretical framework as well as a theory-based goal measuring instrument.

Second, there are a number of theoretical constructs that have not attracted sufficient research attention. This may be due to the degree of difficulty in operationalizing and assessing them, as they frequently refer to dynamic processes (e.g., change processes and feedback mechanisms). Furthermore, goal content has received less research attention than processes involved in goal pursuit (Little, 1999). The effect that the content of goals can have on well-being is emphasized by Ryan et al. (1996). Individuals who focus on extrinsic goals are more likely to suffer from poor well-being than are individuals who focus on goals that represent the needs for autonomy, competence, and relatedness (Ryan, Kuhl, & Oeci, 1997). Another issue that has been neglected is the relationship between different levels of goals. Goal theories have emphasized the existence of a hierarchical structure among goals. The relationship between motives, or higher order goals and lower level personal goals is certainly worthy of further investigation.

Concerning the empirical studies, several points can be made. First, approach goals seem to be positively related to well-being - the higher the percentage of approach goals, the higher the subjective well-being reported. Second, autonomous goals seem to be more positively correlated with goal attainment and effort invested than controlled goals. Third, goal attainment appears to be an obvious and probably powerful mediating variable in the relationship between personal goal processes and wellness outcomes. Behavioral outcomes may also provide further insight into the role that personal goals play in self-regulation. Fourth, for personal goals, gender differences should be taken into account.

GENERAL CONCLUSION AND DISCUSSION

There is ample evidence that work characteristics are linked to important wellness/health consequences, varying from job satisfaction, burnout, turnover, and absenteeism to physical ill-health. Unlike in many other areas of quality of life


research, several theoretical models concerning quality of work have been tested, including the Job Demands-Control-Social Support model (Karasek, 1979), the Effort-Reward Imbalance model (Siegrist, 1996) and the Vitamin model (Warr, 1990). While these models have been successful in predicting wellness/health consequences, they mainly concentrate on the impact of the work environment on the individual and not enough on the way individuals define their work environments in terms of their personal goals and expectations.

Consequently, this chapter introduced self-regulation theory and specifically Motivational Systems Theory (Ford, 1992) as a theoretical framework for research into these motivational aspects. While some research has been carried out from a personal goal perspective, it is difficult to get an integrated picture of the existing empirical literature, because of the lack of a common theoretical framework. In other words, theory-based research in the field of personal goals is crucial. Work characteristics have not been explicitly linked to employees' ability to pursue and attain their personal goals. In addition, little information is provided in empirical studies about the pathways and the reasons underlying the observed relationships between goals and outcomes. There are several theoretical frameworks available and it is time that goal researchers start to constructively exchange opposing views and comments. Wyer, Carver and Scheier (1999) have begun this successfully by discussing various approaches on self-regulation, parallel to their suggestion focusing on control theory.

Klein (1989) suggested an integrated control theory model of work motivation that explicitly incorporates feedback, goal setting, expectancy, and attribution theories, and can be extended to include several other theories (e.g., social learning theory). In addition, Hyland (1988) suggested a meta-theoretical framework based on control theory but enriched with concepts from a number of motivation and goal theories. Motivational Systems Theory (Ford, 1992) is another example of an integrative attempt, perhaps the most inclusive to date. A number of theorists and researchers (Ford et aI., 1987; 8andura, 1989; Carver et aI., 1982; Latham & Locke, 1991; Klein, 1991; Emmons, 1997; Austin & Vancouver, 1996; Powers, 1973; Robertson et aI., 1990; Roberson, 1989; Affleck et aI., 1998; Maes & Gebhardt, 1999) have presented arguments suggesting that the goal construct is fundamental to human functioning and consequently to wellness and health. We suggest that these integrative models be used to relate the findings of existing studies in order to give an impetus for future research.

Acknowledgments We would like to thank Arne Mooers for his valuable comments.

Correspondence should be addressed to: Georgia Pomaki, Faculty of Social Sciences, Health Psychology Section, P.OBox 9555, 2300 RB Leiden, The Netherlands. E-mail: [email protected] Tel: ++31 71 5273995, Fax: ++31 715274678.


REFERENCES

Affleck, G., Tennen, H., Urrows, S., Higgins, P., Abeles, M., Hall, C., Karoly, P., & Newton, C. (1998). Fibromyalgia and women's pursuit of personal goals' A daily process analysis. Health Psychology, 17,40-47

Alterman, T., Shekelle, R. B., Vernon, S., & Burau, K. D. (1994). Decision latitude, psychologic demand, job strain, and coronary heart disease in the Western ElectrIc Study American Journal of Epidemiology, 139,620-627

Austin, J. T & Vancouver, J. B. (1996). Goal constructs in psychology: Structure, process, and content. PsychologIcal Bulletm, 120, 338-375

Bandura, A (1989). Self-regulatIOn of motivation and action through mternal standards and goal systems. In L.APervin (Ed.), Goal concepts in Personalzty and SOCIal Psychology (pp. 19-85). Hillsdale: Lawrence Erlbaum Associates, Publishers.

Barnett, R. C. & Brennan, R. T. (1995). The relationship between job experienc.es and psychological distress: A structural equation approach. Journal ofOrgamzallonal BehaVIOr, 16,259-276.

Barrick, M. R., Mount, M. K., & Strauss, J. P. (1993). Conscientiousness and performance of sales representatives: Test ofthe mediating effects of goal settmg. Journal of Applzed Psychology, 78, 715-722.

Brown, S. P., Cron, W. L., & Slocum, J. W. (1997). Effects of goal-directed emotIOns on salesperson volitions, behavior, and performance: A longitudinal study. Journal of Marketmg, 61, 39-50.

Brunstein, J. C. (1993) Personal goals and subjective well-being. A longitudinal study. Journal of Personalzty and SOCial Psychology, 65, 1061-1070.

Caiman, K. C. (1984). Quality ofhfe in cancer patients. An hypothesis. Journal of MedIcal EthiCS, 10, 124-127.

Cantor, N. (1994). Life task problem solving' SituatIOnal affordances and personal needs. PersonalIty and SOCial Psychology Bulletm, 20, 235-243.

Carayon, P. (1993). A longitudinal test of Karasek's Job Stram Model among office workers. Work and Stress, 7,299-314.

Carver, C. S. & Scheier, M. F. (1982). Control theorY' A useful conceptual framework for personalitysocial, clinical, and health psychology. PsychologIcal Bulletin, 92, 111-135.

ChrIstiansen, C. H., Backman, C., Little, B. R., & Nguyen, A (1999). OccupatIons and well-being: A study of personal projects. American Journal of OccupatIOnal Therapy, 53, 91-100.

Coats, E. J., Janoff-Bulman, R., & Alpert, N. (1996). Approach versus avoidance goals: Differences in self-evaluation and well-being Personaltty and SOCIal Psychology Bulletm, 22, 1057- I 067.

Cropanzano, R., James, K., & Citera, M. (1993) A goal hierarchy model of personality, motivation, and leadership In L L.Cummings & B. M Staw (Eds ), Research m OrgamzatlOnal BehavIor (pp 267-322). London: JAI Press Inc.

Cropley, M , Steptoe, A, & Joekes, K (1999). Job stram and psychiatric morbidity. Psychological Medlcme, 29, 1411-1416.

De Jonge, J., Bosma, H., Peter, R., & Siegrist, J. (2000) Job strain, effort-reward Imbalance and employee well-being: a large-scale cross-sectional study SOCial Science and Medicme, 50, 1317-1327.

Deci, E. L., Connell, J. P., & Ryan, R. M (1989). Self-detenninatlOn in a work organization. Journal of Applted Psychology, 74,580-590.

Donnann, C. & Zapf, D. (1999). Social support, SOCIal stressors at work, and depressive symptoms: Testing for main and moderating effects WIth structural equations in a three-wave longitudinal study. Journal of Applted Psychology, 84, 874-884

Dwyer, D. S. & Mitchell, O. S. (1999). Health problems as detenninants of retirement. Are self-rated measures endogenous? Journal of Health Economics, 18, 173-193

Edwards, J. R. (1992). A cybernetIc theory of stress, coping, and well-being in organizations. Academy of Management Review, 17,238-274.

Elliot, A J & Sheldon, K. M. (1997). Avoidance achievement motivation. A personal goals analysis Journal of Personaltty and SOCial Psychology, 73, 171-185

Elliot, A J., Sheldon, K. M., & Church, M. A (1997). Avoidance personal goals and subjective wellbeing. Personality and SOCial Psychology Bul/etm, 23, 915-927

Emmons, R A (1989). The personal strIving approach to personality. In L.APervin (Ed.), Goal concepts m Personalzty and SOCial Psychology (pp. 87-126) Hilsdale Lawrence Erlbaum Associates.


Emmons, R. A (1992). Abstract versus concrete goals: Personal striving level, physical illness, and psychological well-being. Journal oj Personality and Social Psychology, 62, 292-300.

171

Emmons, R. A. (1997). Motives and life goals In R.Hogan, J. Johnson, & S. Briggs (Eds.), Handbook oj Personality Psychology (pp 485-512). San Diego: Academic Press.

Feldt, T., Kmnunen. U., & Maunao, S (2000). A mediational model of sense of coherence in the work context: A one-year follow-up study Journal oj OrganizatIOnal BehavIOr, 21, 461-476.

Fletcher, B. C. & Jones, F. (1993). A refutatIOn of Karasek's demand-discretion model of occupational stress with a range of dependent measures. Journal oJOrganizational BehavIOr, 14,319-330.

Ford, M. E. (1992). Motivating humans: goals, emotIOns, and personal agency belieJs. California: Sage Publications.

Ford, M. E. & Ford, D. H (1987). Humans as self-constructing liVing systems: a developmental perspecllve on behaVIOr and personality. Los Angeles, California: Erlbaum

Ford, M. E. & Nichols, C W. (1991). Usmg goal assessments to identify motivational patterns and faCilitate behavioral regulation and achievement. In M.L.Maehr & P. R. Pintrich (Eds.), Advances In

Motivation and Achievement (pp. 51-84). Greenwich: JAI Press Inc. Fried, Y, BenDavid, H A., Tiegs, R. B., Avital, N., & Yeverechyahu, U. (1998) The interactive effect of

role conflict and role ambiguity onjob performance. Journal oJOccupatlOnal and OrganizatIOnal Psychology, 71 Part I, 19-27.

Gardner, D G. & Pierce, J. P (1998). Self-esteem and self-efficacy within the organizational context - An empirical examination Group and OrganIZatIOn Management, 23, 48-70

Greiner, B. A, Krause, N., Ragland, DR., & Fisher, j. M. (1998). Objective stress factors, accidents, and absenteeism in transit operators: A theoretical framework and empirical evidence Journal oJ OccupatIOnal Health Psychology, 3, 130-146.

Harrison, D. A. & Martocchio, J. J. (1998). Time for absenteeism A 20-year review of origins, offshoots, and outcomes. Journal oj Management, 24, 305-350.

Hui, C. & Lee, C. (2000). Moderating effects of organization-based self-esteem on organizational uncertainty: Employee response relationships. Journal oj Management, 26, 215-232.

Hyland, M. E. (1988). Motivational control theory: An integrative framework. Journal oj Personality and SoclQI Psychology, 55, 642-651.

Iverson, R. D. & Magulfe, C. (2000) The relationship between job and life satisfaction: Evidence from a remote mining community. Human RelatIOns, 53, 807-839.

Johnson, J. V. & Hall, E (1988). Job strain, workplace social support, and cardiovascular disease: A cross-sectional study of a random sample of the Swedish working populatIOn. American Journal oj Publlc Health, 78, 1336-1341.

Karasek, R A (1979). Job demands, job decision latitude, and mental strain: Implications for job redesign Administrative SCience Quarterly, 24, 285-311.

Karasek, R A (1989). Control in the workplace and its health-related aspects. In S.L.Sauter, J. J Hurrell, & C L. Cooper (Eds ), Job control and worker health (pp. 129-159) Chichester. John Wiley & Sons, Ltd.

Karasek, R A. & Theorell, T. (1990). Healthy work: Stress, productiVity, and the reconstruction oj working life. Basic Books

Karoly, P. (1993). Mechanisms of self-regulation: A systems view. Annual ReView oJPsychology, 44, 23-52

Karoly, P. & Ruehlman, L. S. (1996) Motivational Implications of pain: Chronicity, psychological distress, and work goal construal in a national sample of adults. Health Psychology, 15, 383-390.

Kasl, S. V. (1996). The influence of the work environment on cardiovascular health: A historical, conceptual, and methodological perspective. Journal oj OccupatIOnal Health Psychology, I, 42-56

Kirschenbaum, A., Oigenblick, L., & Goldberg, A. I. (2000). Well being, work environment and work accidents SoclQI Science and MediCine, 50, 63 I -639.

Kivimaki, M. & Lindstrom, K. (1995). Effects of private self-conSCIOusness and control on the occupational stress-stram relationship Stress Medicine, II, 7-16.

Klein, H J. (1989) An integrated control theory model of work motivation. Academy oj Management ReView, 14, 150-172.

Klem, H J. (1991). Control theory and understanding motivated behavior' A different conclusion. MotivatIOn and EmotIOn, 15, 29-44.

Kristensen, T. S. (1995). The demand-control-support model: Methodological challenges for future research. Stress MediCine, II, 17-26.


Kristensen, T. S. (1996). Job stress and cardiovascular disease' A theoretic critical review. Journal of OccupatIOnal Health Psychology, 1,246-260.

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.

Latham, G. P. & Locke, E A (1991) Self-regulation through goal setting. OrganizatIOnal BehavIOr and Human DecISIon Processes, 50,212-247

Leithwood, K., Menzies, T , Jantzi, D , & Leithwood, J (1996). School restructuring, transformational leadership and the amelioration of teacher burnout. Anxiety, Stress and Coping: An InternatIOnal Journal,9, 199-215

Little, B. R. (1989). Personal projects analysis Trivial pursUIts. magnificent obsessions. and the search for coherence. In D.Buss & N. Cantor (Eds.), Personality Psychology Recent trends and emerging d/YectlOns(15-31). New York: SpnngerVerlag.

Little, B. R. (1999) Personality and motivation: Personal action and the conative evolution. In LA Pervin & O. P John (Eds.), Handbook of Personality Theory and Research (2 ed, pp 501-524). New York, US: The GUilford Press.

Lu, L., Kao, S F., Cooper, C. L, & Spector, P E. (2000). Managerial stress. locus of control, andjob stram in Taiwan and UK. A comparative study. InternatIOnal Journal of Stress Management, 7, 209-226.

Lund, T. & Borg, V. (1999). Work environment and self-rated health as predictors of remaining In work 5 years later among Danish employees 35-59 years of age. Experimental Aging Research, 25, 429-434.

Maes, S. & Gebhardt, W. (1999). Self-regulatIOn and health behavior. The Health Behavior Goal model. In M.Boekaerts, P. R. Pintrich. & M. Zeidner (Eds ), Handbook of Self-RegulatIOn (pp. 343-368). San Diego, CA: Academic Press.

Noe, R. A (1996). Is career management related to employee development and performance? Journal of Organizational BehavIOr, 17, 119-133.

O'Neill, B. S & Mone, M A (1998). Investigating equity sensitivity as a moderator of relations between self-efficacy and workplace attitudes. Journal of Applied Psychology, 83, 805-816.

Peter, R. & Siegrist. J (1999) Chronic psychosocial stress at work and cardiovascular disease: The role of effort-reward imbalance. InternatlOnaLJournal of Law and Psychiatry, 22, 441-449.

Phillips. S D., Little. B. R., & GoodlOe, L. A (1997). Reconsidering gender and public admimstration: five steps beyond conventional research. Canadian Public AdministratIOn Administration Publlque Du Canada, 40, 563-581.

Powers, W. T. (1973). Feedback: Beyond behaViorism. SCience, 179, 351-356. Probst, 1. c., Baxley. E. G., Schell, B. 1., Cleghorn, G. D , & Bogdewic, S P. (1998). Organizational

environment and perceptions ofteachmg quahty in seven South Carolina family medicine reSidency programs. AcademIC MediCine, 73, 887-893

Roberson, L. (1989). Assessing personal work goals in the organizational setting: Development and evaluation of the Work Concerns Inventory. OrgamzatlOnal BehaVIOr and Human DeCision Processes, 44, 345-367.

Roberson, L. (1990). Prediction of job sattsfaction from characteristics of personal work goals Journal of OrganizatIOnal BehavIOr, /1,29-41.

Roberson, L., Korsgaard, M. A, & Diddams, M. (1990) Goal characteristics and satisfaction: Personal goals as mediators of situational effects on task satisfaction Journal of Applted SOCial Psychology, 20,920-941

Robertson, R. 1. & Powers, W. T. (1990) IntroductIOn to Modern Psychology The Control-Theory view. Gravel SWitch, Kentucky: The Control Systems Group, Inc.

Roe, R A, Zinovieva, I. L, Dienes, E., & Ten, H. L. (2000) A comparison of work mottvation in Bulgaria, Hungary, and the Netherlands: Test of a model Applied Psychology An InternatIOnal ReVIew Psychologle Appliquee Revue Internatlonale, 49, 658-687

Rose, 1., Jones, F., & Fletcher, B. C (1998). Investtgating the relationship between stress and worker behaviour. Journal of Intellectual DIsability Research, 42, 163-172.

Ryan, R M .• Kuhl, J , & Deci, E. L. (1997) Nature and autonomy: An organizational view of SOCial and neurobiological aspects of self-regulation in behavior and development Development and Psychopathology, 9, 701-728.

Ryan, R. M .• Sheldon. K. M., Kasser, T., & Deci. E. L. (1996) All goals are not created equal - An organismic perspective on the nature of goals and their regulatIOn. In P.M GoIlwltzer & J. A. Bargh (Eds.), Psychology of ActIOn (pp 7-26) New York The GUIlford Press


Schipper, H., Clinch, 1., & Powell, V (1990) Defimtions and conceptual issues. In B Spilker (Ed ), Quality of lifo assessments In clinical tnals. (pp. 11-25) New York Raven Press.

173

Schnall, P. L., Landsbergls, P. A., & Baker, D B. (1994) Job strain and cardiovascular disease. Annual ReVIew of Public Health. 15, 381-411

Schonfeld, I. S. (2000) An updated look at depressive symptoms and Job satisfaction in first-year women teachers Journal of OccupatIOnal and Orgamzatlonal Psychology, 73 Part 3, 363-371.

Searle, B. J., Bright, 1. E H., & Bochner, S. (1999). Testing the 3-factor model of occupatIOnal stress. the impact of demands, control and social support on a mail sortmg task. Work and Stress, 13, 268-279

Siegrist, J (1996). Adverse health effects ofhigh-effortllow-reward conditIOns Journal of OccupatIOnal Health Psychology, 1,27-41

Stansfeld, S. A, Bosma, H, Hemingway, H., & Marmot, M. G. (1998). Psychosocial work characteristics and social support as predictors ofSF-36 health functioning The Whitehall 11 study Psychosomatic Medicine, 60, 247-255

Storr, C. L., Trinkoff, A. M., & Anthony, J. C (1999) Job strain and non-medical drug use. Drug And Alcohol Dependence, 55, 45-5\.

Sundquist, J & Johansson, S E (2000) High demand, low control, and impaIred general health: working conditions in a sample of Swedish general practitioners. ScandinaVian Journal of Public Health, 28, 123-\31.

Tennant, C. (2000). Work stress and coronary heart disease. Journal Of Cardiovascular Risk, 7, 273-276. Theorell, T & Karasek, R A. (I996) Current Issues relatmg to psychosocial job stram and cardIOvascular

disease research Journal of Occupational Health Psychology, I, 9-26. Vahtera, J., Kivlmakl, M., Uutela, A, & Pentti, J (2000). Hostility and ill health. role of psychosocial

resources in two contexts of working life Journal of Psychosomatic Research, 48, 89-98. van der Doef, M. & Maes, S. (1998). The Job Demand-Control( -Support) model and phYSical health

outcomes A review of the strain and buffer hypotheses. Psychology and Health, 13, 909-936. van der Doef, M. & Maes, S. (1999) The Job Demand-Control(-Support) model and psychological well

being. a review of20 years of empirical research Work and Stress, 13, 87-114 van der Doef, M., Maes, S., & Diekstra, R. (2000). An examination ofthe job demand-control-support

model with various occupational strain indicators. Anxiety Stress And COPing, 13, 165-185. VandeWalle, D., Brown, S. P., Cron, W. L., & Slocum, 1. W (1999). The influence of goal orientation

and self-regulation tactics on sales performance: A longitudmal field test. Journal of Applied Psychology, 84, 249-259.

Verhoeven, C., Maes, S , Kraay, V., & Joekes, K The Job Demand-Control-SoCial Support model and wellnesslhealth outcomes A European study. Health Psychology (m press)

Wamala, S P., Mittleman, M. A, Horsten, M., Schenck, G K, & Orth, G K. (2000). Job stress and the occupational gradient in coronary heart disease risk in women - The Stockholm Female Coronary Risk StUdy. SOCIal Science and MedICine, 51, 481-489.

Warr, P B. (1990). DeCision latitude, Job demands, and employee well-being Work and Stress, 4, 285-294

Wyer, R S Jr., Carver, C. S., & Scheier, M F (1999). Perspecttves on behavIOral self-regulatlOn_ Mahwah, NJ: Lawrence Erlbaum Assoc Publ.

ANNA NIEBOER AND SIEGWART LINDENBERG

SUBSTITUTION, BUFFERS AND SUBJECTIVE WELL-BEING: A HIERARCHICAL APPROACH

Abstract. In quality of life research, social aspects have become recognized as important, next to physical aspects such as livmg conditions and health. However, objective indicators of physical and social aspects often correlate poorly with indicators of subjective well being. We argue that this may be due to buffer and substitution effects People produce their own well-being and there are multiple means for realizing the same ultimate goals. Multiple means allow the formation of buffers which cushion the negative effect of loss and they allow substitution when some means become costlier or inaccessible. In order to trace buffer and substitutIon effects, one needs a theory that specifies these multiple means and the goals they serve (goal hierarchy) One such theory is Social Production Function theory which we use to trace buffer and substitution effects. We test the hypotheses on these effects with data on 1094 Dutch respondents Even though the data are cross-sectional and thus do not allow us to demonstrate buffer and substitution effects on the individual level, there are good indications in the data that such etJects exist

INTRODUCTION

With increasing welfare, subjective well-being becomes increasingly important as a topic of research. This is a fairly new development and quite different from the concern, after the Second World War, with basic aspects of quality of life, especially in developing countries. One new direction was to include social conditions along with research into physical conditions. It is important to look at both people's social and physical living conditions because, presumably, people's social relations and health influence their subjective feelings of well-being (Argyle, 1996; Baumeister & Leary, 1995). There appeared to be a problem, however. The objective indices are often not strongly associated with subjective feelings of well-being (Argyle, 1999; Cummins, 1996; 2000; Diener, 1984; Diener, Suh, Lucas, & Smith, 1999). Why are people still fairly satisfied with their lives even when they lack important resources such as a paid job or a spouse? One plausible answer to this question is that people are much more active in the production of their own subjective well-being than is generally assumed in the quality of life literature. They have mUltiple ways of realizing well-being; that is, they have buffers and they can substitute one means for another. One problem with this answer is that it can only be seriously pursued if we have a theory that allows us to pinpoint the multiple means for achieving subjective well-being and helps us say something about which means are more essential (Le., less substitutable) than others (Diener & Lucas, 2000). In other words, what we need is a theory in which means to achieve subjective well-being are ordered hierarchically. Each layer in the hierarchy contains multiple means for realizing goals on a higher level which, in turn, are multiple means for still a higher level of goals. Thus, such a hierarchy pinpoints alternative means for the realization of the same higher-order goal. These alternatives allow substitution when loss occurs or

175

E. Gullone and R.A. Cummins (eds.), The Universality of Subjective Wellbeing Indicators, 175-189. © 2002 Kluwer Academic Publishers.

SUBSTITUTION, BUFFERS AND SUBJECTIVE WELL-BEING 176

when certain means become relatively too costly. It is theoretically particularly interesting to consider substitution between classes of means such as substituting a strong emphasis on affection for the pursuit of status after retirement.

Multiple means also allow the formation of buffers. The idea of buffers is quite simple. The realization of higher-order goals is subject to decreasing marginal returns. For example, having friends is important for realizing affection. But having many friends may add only a fraction of extra affection beyond the level realized by having a few friends. The same can be said about different means. Having a paid job may create a certain level of status. Adding volunteer work may increase one's status only marginally beyond the level already achieved by paid work. Even though the increase in higher-level goals (and in subjective well-being generally) may only be marginal, people are likely to pursue this increase when the means are relatively easily available. People who make friends easily may as well maintain many friends, and people who have many talents and much energy may add voluntary work to their paid job as well. Because the effect of the extra means is marginal, it creates buffers: When some of the means fall away, the overall well-being is not much affected. For theorizing about buffer effects, it is also necessary to identify the higher-order goals that may be served by multiple lower-order means. In other words, for both substitution and buffer effects, we need a theory on hierarchically ordered goals and means.

In the following, we will present such a theory (the so-called Social Production Function Theory, or SPF theory for short, see Lindenberg, 1996; Ormel, Lindenberg, Steverink, & Von Korff, 1997; Ormel, Lindenberg, Steverink, & Verbrugge, 1999) and formulate some hypotheses about buffers and substitution. We will then test these hypotheses with data gathered specifically for the purpose of assessing quality of life with SPF theory.

THEORY

SPF theory, developed by Lindenberg (1986, 1991, 1993, 1996) basically asserts that people produce their own well-being by trying to optimize achievement of universal goals within the constraints they are facing. Lindenberg distinguishes a number of universal and (hierarchically ordered) instrumental goals, which allows specificity about how individuals try to achieve well-being and reduces ad hoc specifications of needs and wants. Overall subjective well-being is a function of physical well-being and social well-being. For social well-being, three universal instrumental goals are specified: status, behavioral confirmation, and affection. Status refers to a relative ranking (mainly based on control over scarce resources such as money and education). Behavioral confirmation refers to one's belief that they have done "the right thing" in the eyes of relevant others. Affection includes love, friendship and emotional support, and is to a large extent provided in caring relationships (intimate, family and friendship relations). For the production of physical well-being, two universal instrumental goals have been distinguished: comfort and stimulation (see Wippler 1990). Comfort means the absence of deleterious stimuli (Le., physiological discomforts such as pain, hunger, thirst or cold). Stimulation refers to activation which produces arousal, including mental and

177 A. NIEBOER AND S. LINDENBERG

sensory stimulation and physical effort. Human beings seem to prefer a certain level of activation, although prolonged levels of high activation or physical effort become unpleasant. All five instrumental goals are assumed to have decreasing marginal value for the production of well-being.

An important characteristic of SPF-theory is that goals are hierarchically structured with the help of production functions. General universal goals (Le., physical and social well-being) are at the top; then there is a layer of universal instrumental goals (stimulation and comfort for the production of physical wellbeing, and status, behavioral confirmation and affection for the production of social well-being), then there are layers of means (resources) that are specific to cultures, groups and circumstances.'v If a person lacks the necessary resources for the realization ofa higher level goal, then the production of these resources can become an instrumental goal in itself For example, somebody may direct her activities toward making money in order to be able to buy a house in the future. Given that the realization of the goal is in the future, such an activity can also be seen as investment behavior.

Instrumental goals are viewed in SPF theory as substitutable depending on their relative costs. If, for example due to unemployment, opportunities and resources for the achievement of status are decreasing, a person is likely to increase the production of affection and behavioral confirmation if that production is relatively easier. In terms of expressed preferences, this may show up as an increased interest in norm-conforming behavior and investment in personal relationships. Table I gives an overview of the hierarchical levels of social well-being.

Table J. The hierarchical levels of social well-being

First-order Status Behavioral Affection Instrumental Confirmation Goals (control over (what you get (what you get from

scarce from "doing the others who care resources) right things") about you)

Examples of Occupation, Compliance with Intimate interaction, means by which Excellence in external and Providing support, goals can be sports or work internal norms Unilateral transfers achieved Examples of Education, Social skills, Spouse, Empathy, Resources / Social class, Competence Attractiveness Constraints Unique skills

Multi-functional activities, especially those that combine production and investment, and those that serve multiple higher-order goals, are clearly the most efficient kinds of activity (Lindenberg, 1996; Nieboer, 1997). People will have the tendency to engage in activities that combine the production of physical well-being and social well-being. To illustrate the mechanism of multi-functionality, one may think of what constitutes a good partner. In terms of SPF, a good partner is a partner who is stimulating, creates comfort for the other, raises the status of the other, confirms his


(or her) actions and opinions, and cares about the other. Because people seek out multifunctionality, they will attempt to find and maintain partners that are multifunctional for them. The same can be said about the work context. Work is only in part a means for making money. For a large part, it can be means to realize all five universal instrumental goals. Thus, even voluntary work should be a source for well-being rather than a drag or net sacrifice.

HYPOTHESES

The major heuristic guidance of SPF theory comes from the universal instrumental goals. They guide our search for multiple means, both for buffer effects and for substitution possibilities. Substitution possibilities are first of all located within means to realize a particular universal instrumental goal (say, status), and, if this substitution is not possible, we look for it within possibilities to substitute between universal instrumental goals. If, for example, people lack the necessary resources to obtain status they will focus on affection and behavioral confirmation (and, barring these possibilities, even comfort and stimulation). There is no change in values involved (even though it may subjectively be experienced as a lower value placed on status). People will search for the most efficient way to produce well-being and they will thus shift between means according to relative costs (i.e., shadow prices), just as they would shift between any other kind of substitution goods. Of course, there is an (unspecified) minimum level for each of the universal instrumental goals below which no substitution will take place. Thus, everybody needs a minimum amount of stimulation, comfort, affection etc. In sum, the hypothesis is:

H I (Substitution) If means of productIOn for a specific universal Instrumental goal become relatively more costly or unavailable, people wIll try to shIft to other means.

With regard to buffer formation, SPF theory guides us to mainly look at two aspects. These are; first, the relative cost of means of production, and secondly the different degrees of decreasing marginal returns. Let us take these up in tum. As mentioned above, when people have many talents and much energy they can increase their status (and probably also other goals) by adding voluntary work to their paid job, even though the increase in status may only be marginal. In this case, it is relatively cheap for people to add this extra means of production. In general, the assumption of subjective well-being maximization (or the more general assumption that people try to improve their condition) leads to the result that people keep producing more and more of a universal instrumental goal until the marginal return is equal to the marginal cost. When that cost is low, people get virtually saturated with the achievement of a particular universal instrumental goal. This also means that, as a side effect, a buffer is created against great losses in well-being when certain means of production fall away. In terms of a hypothesis, we can say:

H2 (Buffer) The relatIvely cheaper the means of productIOn, the more likely that people become saturated with regard to the universal instrumental goals. Thereby they form buffers against loss of subjective well-being.

Not every instrumental goal becomes as easily saturated as another does. There are good reasons to assume that there are differences between them. There may be


psychological reasons for differences in saturation between the different universal instrumental goals that have not been well explored yet. However, at least for social well-being, the internal logic of SPF theory guides us to look at a very specific aspect: to what degree can universal instrumental goals also be used as lower-order means. For example, status contributes directly to social well-being. Yet, status can also be used as a means to realize other means and universal instrumental goals. For example, a high status person is likely to get credit more easily, and will be agreed with more often (Le., receives more behavioral confirmation, see Wagner and Berger, 1993) and will have a wider range of interactions (Le., more stimulation, see Homans 1951). Status is in that sense akin to money. Because of this role on different levels in the hierarchy, the marginal return for increasing status is likely to decrease slowly. By contrast, behavioral confirmation is much less useful as a lower-order means even if it still does help to get credit and a modicum of status. Finally, affection may be useful besides being its own reward in terms of social well-being, but because it is intentionally not contingent on behavior, it is still less useful as a lower-level means than behavioral confirmation. Thus, the marginal return of affection in terms of well-being should level off much more than the one for behavioral confirmation, which, in tum, would level of faster than the one for status. Buffer formation should occur most for affection and least for status, with behavioral confirmation falling in-between. The hypothesis on the shape of the functions then would be:

H3 (Marginal rate of return) The marginal returns for affectIOn in terms of subjective well-being decrease faster than those for behavioral confirmation. In tum, the latter decrease faster than those for status

The above hypotheses will be tested in the following section.

METHOD

Participants

The study population consisted of2,668 persons aged 18 to 65 years. We randomly drew a sample from Dutch postal addresses and asked eligible subjects by letter to cooperate when approached for a phone interview a few days later. To avoid overrepresentation of women, the youngest male member of the household who was currently at home was interviewed. If not present, the youngest female was interviewed (Hess, 1994). The interviews lasted approximately 25 minutes. All interviews were carried out by well-trained social science students. Useful data are available for 1,094 subjects (41% of the study population). The response rate was 59% (52.5% female). Of all respondents 82.9% reported having a partner, 68.5% had paid work and 31.4% was involved in volunteer work.


Measures

The SPF-IL (Social Production Function Instrument for the Level of well-being) is a multidimensional instrument to measure the instrumental goals that enable people to realize well-being (for a description of the development of the instrument see Nieboer, Lindenberg, Boomsma, & Van Bruggen, forthcoming).

For affection, questions are asked about the extent to which people feel liked, loved, trusted and accepted, understood, empathized with, know that their feelings are reciprocated, feel that others are willing to help without expecting something in return, feel that their well-being is intertwined with others, and feel that others like to be close and hug them (Van Bruggen, 2001). Eighteen items were selected to assess the nine different aspects of affection. Each aspect included a positive and a negative item. Examples of items are: "Is it difficult for others to put themselves in your shoes?", "Do people really like you?", "Do you feel that people don't care enough about you?". The 18 items were coded on a 4-point scale with the categories "never", "sometimes", "often", or "always" (range 0-3). The scores of the items were recoded in order to have higher scores reflect higher levels of affection. The indicator was transformed to a range of 0-100 (dividing it by the maximum scale score and multiplying it by 100). Cronbach's alpha for the affection scale was .79.

Behavioral Confirmation. The level of behavioral confirmation was measured with respect to six aspects, feeling that you: do good things, do things well, are a good person, are useful, are part of a functional group, and contribute to a common goal. Examples of items are: "Do others think that your contribution is too small?", "Do people think that you make the right choices?", "Do you feel useful to others?". The instrument consists of 12 items; range of indicator 0-100. Alpha for internal reliability for behavioral confirmation was. 71.

Status. The level of status refers to six aspects: the feeling of being treated with respect, being independent, self-realization, achievement as compared to others, influence, and reputation. Examples of items are: "Do people think that you do better than others?", "Do people look down on your achievements?", "Do people think you are influential?". The instrument consists of 12 items; range of indicator DIDO. Cronbach's alpha was .60.

Comfort. The level of comfort refers to the absence of feelings of discomfort such as pain or stress. Respondents were asked: "How often do you have pain?", "In the past few months did you feel: ... fit", " ... perfectly healthy". The instrument consists of 8 items; range of indicator 0-100. Cronbach's alpha for the level of comfort was .86.

Stimulation. The level of stimulation refers to mental and physical activation. Respondents were asked, for example, "Do you find your life boring?", "Are your activities challenging to you?", "Do you really enjoy your activities?". The instrument consists of8 items; range of indicator 0-100. Alpha for internal reliability for the level of stimulation was .80.

The overall level of subjective well-being (i.e., utility) is measured with respect to life satisfaction, positive and negative affect. Cantrill's Ladder (1965) is used to assess satisfaction with life and reflects a general, cognitive evaluation of a person's overall well-being. "On a scale of I to 10, how satisfied are you with your life as-a-


whole now?" A IO-item version of the PANAS (Watson, Clark & Tellegen, 1988) was used to assess positive and negative affect. Positive affect consists of 5 items: During the past 3 months, how often did you feel ... excited, enthusiastic, alert, inspired, and determined. Due to low scalability of the item on feeling excited, it was removed from the analyses. Cronbach's alpha was .62. Negative affect consists of 5 items: sad, upset, afraid, nervous, scared. Cronbach's alpha was .73. An overall score of subjective well-being was used by adding the standardized scores of satisfaction with life, positive and negative affect, transforming the indicator to a range of 0-10. It is assumed that people can be compared on this score.

STATISTICAL ANALYSES

In order to determine the diminishing marginal returns of the level of affection, behavioral confirmation and status on the overall level of subjective well-being, quadratic terms are used. The quadratic effect of affection, behavioral confirmation and status is calculated after the variables are centered in order to avoid multicollinearity which would distort the main effects of these variables. This means that the square of each one of these variables is calculated after subtracting the means from the scores on these variables (Aiken & West, 1991).

OPERATIONAL HYPOTHESES

Since we did not have longitudinal data, we were restricted to cross-sectional tests. This means that we will not always be able to distinguish clearly between substitution and buffer effects. The test of the three hypotheses will thus be limited. Still we believe that the results are well worth being considered due to the fact that the data set is quite large and that it has been gathered with carefully operationalized concepts of SPF theory. Two pilot studies have been used to create the operationalizations of the five universal instrumental goals (Nieboer et aI., forthcoming). The way we will test the three hypotheses is as follows. One way to test Hypothesis I is to divide the sample into two status groups: high and low. If people with low status substitute, they will depend more heavily than the high status group on one or more of the other four universal instrumental goals for realizing their subjective weB-being. Thus, one operational version of hypothesis I is:

HI' (Substitution) Low status people depend more heavily for their subjective wellbeing on one or more of the other four universal instrumental goals (stimulation, comfort, behavioral confirmation and affection) than high status people.

A way to test buffer effects is to look at goods that, according to SPF theory, are very important for subjective weB-being: multifunctional goods. As mentioned earlier, work and a partner are multifunctional goods (see Nieboer, Lindenberg, & Ormel, 1998, Nieboer et al. forthcoming). Paid work is a major multifunctional source of well-being and, to a lesser extent, so is voluntary work. Similarly, having a partner is a major source of well-being, and, to a lesser extent, so is having many friends. The argument about buffer effects is as follows. According to hypothesis H2, people for whom engaging in work is not very costly (in terms of talent and


effort) should not just engage in paid work but also in voluntary work even if voluntary work only increases well-being marginally ("buffer formation"). One can then expect an absolute and a relative buffer effect. If those who do engage in voluntary work lose their paid job, their well-being is higher than if they had lost their paid job and not engaged in voluntary work (absolute buffer effect). Also, voluntary work makes the gap in well-being between having a paid job and being unemployed smaller than it would have been without voluntary work (relative buffer effect). Thus, we will test the following hypotheses:

H2' (Buffer formation) There should be a sIzable percentage of people who engage both m paid work and in voluntary work Their level of well-being should be generally higher than the level of well-being for people wIth a paId job but no voluntary work.

H2" (Buffer effect) People who are unemployed with voluntary work have a higher level of well-being than those who are unemployed without voluntary work (absolute buffer effect), The difference in well-being between employed and unemployed should be less among those who engage in voluntary work than among those who are not involved in voluntary work (relative buffer effect).

Of course, for H2" it is not possible to rule out a substitution effect. People who are out of paid job may as a reaction to unemployment engage in voluntary work. Our cross-sectional data do not allow us to determine the timing of voluntary work. For this reason, it is wise to also look at a situation in which the buffer effect is more likely than substitution: not having a partner. People who have a partner and for whom making friends comes easy are likely to do both: have a partner and many friends. If such people lose their partner, they still have their friends and thus have a higher well-being than if they had lost the partner without having many friends to fall back on (absolute buffer effect). Also, having many friends cushions the blow of losing one's partner (relative buffer effect). People who lose their partner are unlikely to make many new friends as a reaction to the loss of a partner (see Nieboer, 1997) and thus, it is likely that we are dealing with true buffer effects (and no substitution effects). The operational hypotheses to be tested are:

H2'" (Buffer formation) There should be a sIzable percentage of people who have both a partner and many frIends. Their level of well-being should be generally higher than the level of well-bemg for people wIth a partner and only a few friends

H2"" (Buffer effect) People without a partner but who have many friends have a hIgher level of well-being than people without a partner and only a few friends (absolute buffer effect). The difference in well-being between people with or without a partner should be less among those with many friends than among those with only a few fnends (relative buffer effect).

Hypothesis 3, which concerns the shapes of the production function for status, behavioral confirmation and affection, can be tested by looking at the quadratic terms. A negative quadratic term points to a concave utility function, and the less a factor deviates from linearity the less its quadratic term should contribute to the explanation of the variance of well-being. The operational hypothesis for H3 is thus:

H3' (Marginal rate of return) (a) The quadratic terms for affection, behavioral confirmation and status are all negative, (b) the negative quadratic term for affection contributes more to the explanation of subjective well-being than that for behavioral


confirmatIon which, in tum, contributes more to thIs explanation than the negative quadratic term for status.

RESULTS

Descriptives

Table 2 gives an overview of the descriptive statistics of people's overall, subjective well-being and the level of affection, behavioral confirmation, status, comfort and stimulation.

Table 2. Descriptive statistics of well-being, affection, behavioral confirmation, status, comfort and stimulation.

Well-being Affection Behavioral confirmation Status Comfort Stimulation Valid N (listwise)

Substitution

N 1089 \093 \092 \093 1093 1092 \084

M 6.88

73.83 77.95 69.43 70.06 77.20

SO 1.23

10.33 9.27 9.77

19.07 14.26

The results concerning the test of Hypothesis HI' are shown in Table 3.

Table 3. Multiple regression of subjective well-being on affection, behavioral confirmation, conifort and stimulation by status

Affection Behavioral

confirmation Comfort

Stimulation R square for equation Note: * p< .01; ** p< .001

Status low (n = 614) high (n = 474) Beta Beta .25** .11 * .10* .10*

.24**

.31 **

.44

.30**

.29**

.29

The respondents are divided into two groups. First, there are people with low levels of status who only have limited access to means that provide status. Second, there are people with high levels of status (whatever the source of that status). The regression analyses reveal, as predicted, that people with low status-levels use the


other instrumental goals (i.e., affection, behavioral confirmation, comfort and stimulation) much more (or more effectively) for the production of their overall level of subjective well-being. In the case of high-status, the other goals contribute much less to people's well-being (explained variance .44 versus .29). Hypothesis HI was therefore supported by the data. If people lack means of production for a specific instrumental goal, they try to substitute through realizing other instrumental goals.

Table 4. Mean for well-being by paid work.

Well-being M SO

No paid work 6.7"

Paid work 7.0"

1.4 1.1 N 343 746

Note:" t-testp< .001

Table 5. Mean for well-being by paid work and volunteer work.

Well-being M SO N

Notes:

No paid work No volunteer Volunteer

6.6"b 6.8c

1.4 1.3 228 115

I. a,b,c,d LSO multiple comparison testp< .001 2. One-way Analysis of variance F=7.4; p< .001

Paid work No volunteer

6.9ad

1.2 518

Volunteer 7.2bcd

1.0 228

3. Anova for interaction effect of paid work*volunteer work, controlled for main effects of paid work and volunteer work; well-being (FpaJd work· volunteer work=0.5; p = ns)

Buffer effects

We see from Table 4 that people with a paid job indeed have a significantly higher level of well-being than people without a paid job. From Table 5, we see that there is a sizable percentage (31 %) of people with a paid job who also engage in voluntary work. They also have a higher level of well-being than the one's without voluntary work. This confirms hypothesis H2' on buffer formation. Table 5 also reveals that people without paid jobs who have voluntary work have a higher level of well-being than people who are unemployed without voluntary work. This confirms the absolute buffer effect of hypothesis H2". However, the relative buffer effect of hypothesis H2", tested with ANOVA as an interaction effect, is rejected. This may be due to the substitution effect as discussed earlier. We could not find a smaller difference between employed and unemployed among those engaged in voluntary work than among those who are not involved in voluntary work.


Well-being M

SO N

Notes:

Table 6. Mean/or well-being by partner.

Well-being M SO

No partner 6.3" 1.4

N 187 Note. a t-testp< .001

Partner 7.0· 1.2 902

Table 7. Mean/or well-being by partner andfriends.

No partner Few friends 6.1 abc

1.7 84

Many friends 6.5 ade

1.1 102

Partner Few friends 6.9bd

1.2 383

I. ..b,c,d.e LSD multiple comparison test p< .05 2. One-way Analysis of variance F= 17.8;p < .00]

Many friends 7.0cO

1.1 510

3. Anova for interaction effect of partner*friends, controlled for main effects of partner and friends; well-being (Fpartner*fnends = 3.0; P < .05)

Table 6 shows the results for affection. Having a partner makes a significant difference for the level of well-being. Table 7 shows that, again, there is sizable percentage of people who, even though they have a partner, also have many friends (57%). Their level of well-being is higher than the one for people with few friends (buffer formation), confirming H2'''. People without a partner who have many friends have a higher level of well-being than the people without a partner and only a few friends. Moreover, the interaction effect is significant (i.e. the difference between people with and without a partner is less among those with many friends than among those with only a few friends). This confirms both the absolute and the relative buffer effects of hypothesis H2"".


MARGINAL RATES OF RETURN

Table 8. lntercorrelations for well-being and ajfoction, behavioral confirmation, status and their quadratic terms.

Affection Behavioral confirmation Status

affection squared behavioral confirmation squared status squared

Note. '" p< .001

Well-being .45* .40* .35*

-.27* -.21 * -.11 *

Table 9 Multiple regression analyses of well-being on the quadratic terms of affection, behavioral confirmation and status.

Well-being

affection squared behavioral confirmation squared status squared

Note. * p< .001

If .080

F 31.473*

Beta

-.205* -.114* -.025

From Table 8, we see that the correlation of the quadratic terms with well-being are all negative and highest for affection and lowest for status. Separate regression analyses for affection and status showed that in addition to the linear regression model, the quadratic term of affection contributed 2% of explained variance in wellbeing, but only 1 % for status. A similar picture emerges from the regression analyses in Tables 9 and 10. The quadratic term for affection is significant and larger than for status, with behavioral confirmation in between. Even though the effects are not very strong in terms of the explained variance, they do confirm Hypothesis 3'. This also implies confirmation for the idea that people are likely to create more buffers for affection than for behavioral confirmation and status, in that order.


Table 10. Multiple regression analyses otwell-being on affection, behavioral confirmation, status and their quadratic terms

Well-being step I Affection Behavioral confirmation Status Comfort Stimulation

F 159.200***

step 2 .436 103.850*** affection squared behavioral confirmation

squared status squared Note. t p< .10; * p< .05; ** p< .01; *** p< .001

CONCLUSION

Beta step 1 Beta step 2

.185*** .156***

.087** .087**

.086*** .085**

.262*** .254***

.298*** .298***

-.061 * -.047t

-.040

The goal in this paper was to explain why people are still fairly satisfied with their lives even when they lack important resources such as a paid job or a partner. We argued that such objective indicators are often not strongly associated with subjective feeling of well-being, because of substitution processes and buffer effects. Social Production Function (SPF) theory helps us trace these effects. According to this theory, subjective well-being is the result of people's success in obtaining affection, behavioral confirmation and status for social well-being, and comfort and stimulation for physical well-being. People have multiple means for reaching each of these goals and they also use a variety of means simultaneously, thus building buffers against loss of subjective well-being should a particular means become inaccessible for whatever reason. Differences in marginal returns make it likely that people build more buffers for affection than for behavioral confirmation and least buffers for status. An indirect confirmation of this effect can be seen in a study on the elderly by Steverink (2001). She found that elderly people lose their means of production for social well-being over time in a specific order: status first, then behavioral confirmation. Affection (along with comfort) lasts the longest as a means to produce social well-being. At least in part, this may be due to the differences in buffer formation between the three means. People also are resourceful in substituting across different life-domains and different forms of well-being. For example, a loss of status can be compensated by focusing more on behavioral confirmation and affection and even by putting more emphasis on stimulation and comfort. The empirical test of these effects confirmed the hypotheses, even though the crosssectional data put severe limitations on such a test. The results are also consistent with what we found in a pilot study (Nieboer & Lindenberg, 2000).

By specifying a hierarchical structure of goals and means for the production of subjective well-being, SPF theory opens possibilities to research more deeply the


relationship of objective and subjective indicators of quality of life. People's ability to build buffers and to substitute has important consequences for how they deal with changes in their "objective" conditions. Clearly, people are by and large not passive victims of objective circumstance. At the same time, objective conditions that block buffer formation and substitution do make people unable to cope with changing circumstances. Thus, when we consider objective conditions, we might profitably focus on those that influence both buffer formation and substitution.

Addresses for Correspondence: Anna Nieboer, University of Maastricht, Department of Health Organisation, Policy and Economics, Universiteitssingel 40, 6229 ER Maastricht, the Netherlands, e-mail [email protected]

REFERENCES

Aiken, L.S. & West, S.G. (199 I). Mulltple regressIon: testing and interpreting interactIOns. Newbury Park: Sage Publications.

Argyle, M. (1996). Subjective well-being. In A Offer (Ed ). In pursUIt of the qualtly of life. (pp. 18-45). Oxford: Oxford University Press.

Argyle, M. (1999). Causes and correlates of happiness. In D. Kahneman, E. Diener, & N Schwarz (Eds.), WeI/-being the foundatIOns ofhedomc psychology (pp. 353-373). New York: Russell Sage Foundation.

Baumeister, R.F & Leary, M.R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. PsychologIcal Bul/etln, 117, 497-529.

Cantril, H. (1965). The pattern of human concern. New Brunswick: Rutgers University Press. Cummins, RA (1996). The domains of life satisfaction. An attempt to order chaos Social IndIcators

Research, 38, 303-332. Cummins, RA (2000). Objective and subjective quality oflife: An interactive model. SocIal IndIcators

Research, 52, 55-72. Diener E. (1984). Subjective well-being. Psychological Bul/etln, 95, 542-575. Diener, E. & Lucas, R.E. (2000). Explaining differences in societal levels of happiness: Relative

standards, need fulfillment, culture, and evaluation theory. Journal of Happiness Studies, J (I), 41-78. Diener, E., Suh. E.M., Lucas, R.E., & Smith, H.L. (1999). Subjective well-being: Three decades of

progress. Psychological Bulletin, 125(2), 276-302. Hess, T. (1994). Random IndlVldual samples within households or how Inadequate methodologies lead to

a waste oj time and money. Paper presented at the 47th ESOMAR Marketing Research Congress. Homans, G. (1951). The Human Group. London: Routledge and Kegan Paul. Lindenberg S. (1986). The paradox of privatization in consumption. In A Diekmann A & P. Mitter

(Eds.), Paradoxical Effects of Social BehaVior. Essays in Honor of Anatol Rapoport. HeidelberglWien: Physica-Verlag.

Lindenberg S. (1991). Social well-being, fertility and female labour market behaviour. In J.1. Siegers, 1. De Jong-Gierveld, & I Van Imhoff (Eds.), Female Labour Market BehaVIOur and Fertlltly: A RatIOnal ChOIce Approach. Berlin! New York: Springer Verlag.

Lindenberg S. & Frey B.S. (1993). Alternatives, frames, and relative prices; a broader view of ratIOnal choice theory. Acta Sociologlca, 36, 191-205.

Lindenberg, S.M. (1996). Continuities in the theory of social production functions. In S.M. Lindenberg & H.B.G. Ganzeboom (Eds.), Verklarende soctologle: opstellen voor Reinhard Wippler (pp. 169-184). Amsterdam: Thesis.

Nieboer, A.P. (1997). Life-events and well-bemg: A prospective study on changes m well-being of elderly people due to a serious tllness event or death of the spouse. Amsterdam, NL: Thesis Publishers.

Nieboer, A.P. & Lindenberg, S. (2000). SubsfltutlOn and subJectlve well-bemg: Why do we need new measurement Instruments? Paper presented at the Third Conference ofthe International Society for Quality of Life Studies. July 20-22, Girona, Spain.


Nieboer, AP., Lindenberg, S.M., & Ormel, J. (1999). Conjugal Bereavement and Well-being of Elderly Men and Women: A preliminary Study, Omega, 38(2), 113-141.

Nieboer, AP., Lindenberg, S., Boomsma, A, & Van Bruggen, AC. (forthcoming). D,mensIOns ofwellbeing and theIr measurement: The SPF-/L Scale.

Offer, A (Ed.). (1996). In pursUIt of the quality of life. Oxford: Oxford UniverSity Press. Ormel, 1., Lindenberg, S., Steverink, N. & Verbrugge, L.M. (1999). Subjective well-being and social

production functions. SocIal Ind,cators Research, 46, 61-90. Ormel, J., Lindenberg, S.M., Steverink, N., & Von Korff, M. (1997). Quality of Life and Social

Production Functions: A Framework for Understanding Health Effects, SocIal SCIence & MedICine, 45(7), 1051-1063.

Soclaal en Cultureel Rapport (1994). Rijswijk: Sociaal en Cultureel Planbureau. Steverink, N. (200 I) When and why frail elderly people give up independent living: the Netherlands as an

example. Ageing and SOCIety, 21, 45-69. Van Bruggen, AC. (200 I). IndIVidual productIOn of SOCIal well-being: an exploratory study. Ph.D-thesis

University of Groningen, Amsterdam, NL: Thela Thesis. Wagner, D.G. & Berger, 1. (1993). Status characteristic theory: The growth ofa program. In J. Berger &

M. Zelditch (Eds), Theoretical Research Programs (pp. 23-63). Stanford CA: Stanford University Press.

Watson, D., Clark, L.A., & Tellegen. A (1988). Development and validation of brief measures of positive and negative affect: The PANAS Scales. Journal of Personallty and SOCIal Psychology, 54(6), 1063-1070.

Wippler, R. (I990). Cultural resources and participation in high culture. In M. Hechter, K.D. Opp, & R Wippler (Eds), SoclQllnstltullOns TheIr Emergence, Maintenance, and Effects (pp. 187-204). New York: Aldine de Gruyter.

Due to a Heywood case we had to constrain the error variance of item 5, leaving us with 4 degrees of freedom

2 As both reliability and model fit are somewhat sensitive to number of observations (Jackson, 200 I) we also did a regression which included sample size, but thiS inclusion did not alter the pattern revealed in Table 3.

3 Because of high multicollinearity, multiple regression made no sense for the raw score variables

,,, Utility (U) is achieved via physical well-being (PW) and social well-being (SW), thus the utility function is U = f (PW, SW). Social well-being is produced by three means: Status (S), behavioral confirmation (BC) and affection (A). Thus, the production function for social well-being is SW = f(S, BC, A). Each of these factors can, in turn, be an instrumental goal, produced by other factors. The lower we go in the hierarchy, the more context or domain-specific the production function will be.

Date post:	11-Sep-2021
Category:	Documents
Upload:	others
View:	2 times
Download:	0 times

The Universality of Subjective Wellbeing Indicators: A Multi-disciplinary and Multi-national...

Documents