Jackson 2004

8/19/2019 Jackson 2004

1/101

The Flow Scales Manual

Susan A. Jackson, PhDThe University of Queensland

Robert C. Eklund, PhDThe University of Western Australia

8/19/2019 Jackson 2004

2/101

© Copyright 2004, Robert C. Eklund and Susan A. Jackson

All rights reserved.

Reproduction or use of any portion of this publication by any mechanical, electronic,or other means without written permission of the publisher is prohibited by law.

Library of Congress Card Catalog Number: 1885693516

ISBN: 1885693516

Copyeditor: Greg LeathermanCover Design: Scott Lohr/40 West StudiosManaging Editor: Geoffrey C. FullerProduction Editor: Jamie PeinProofreader: Geoff FullerPrinted by Publishers Graphics

10 9 8 7 6 5 4 3 2 1

Fitness Information Technology, Inc.

P.O. Box 4425, University AvenueMorgantown, WV 26504 USA800.477.4348304.599.3483 phone304.599.3482 faxEmail: [email protected]: www.fitinfotech.com

8/19/2019 Jackson 2004

3/101

Acknowledgments

We would like to acknowledge the critical insights of the pre-eminent flow scholar,Mihaly Csikszentmihalyi, in guiding our work. We trust this manual will assist effortsto conduct quality empirical research in the area of flow and so contribute to theunderstanding of this important construct developed by Csikszentmihalyi. We are cog-nizant of Csikszentmihalyi’s (1992) warning that any measure of flow will only providea partial reflection of this complex experience, and by no means do we suggest the flowscales be used other than to provide one means of examining the construct from a mul-tidimensional perspective. We trust that the scales will be a useful reflection of thereality that is this complex psychological state (Csikszentmihalyi, 1992).

We would also like to recognize the expert advice and contribution of Herb Marsh tothe development of the scales. Marsh (Jackson & Marsh, 1996; Marsh & Jackson,1999) provided statistical expertise during the development of the original flow scales,and valuable input to the further research described in this manual.

We would also like to thank the other researchers who have contributed to the devel-opment of the flow scales. There are too many to list here, but their contributions canbe found in the flow research described in this manual and elsewhere. We would liketo particularly thank Jay Kimiecik for insights and questions that guided the develop-ment of the scales since their inception. While we acknowledge the people who haveassisted our research in various ways, we take responsibility for any errors in judgmentcontained in this work.

Finally, we would like to thank the many participants who took the time and effort toanswer the scales, as well as acknowledge future respondents who will assist the furtherdevelopment and refinement of these instruments.

• iiiACKNOWLEDGEMENTS

8/19/2019 Jackson 2004

4/101

8/19/2019 Jackson 2004

5/101

Responsibilities of the Purchaserof the Flow Scales

Part 1: User’s Scientific Responsibilities

1. I will carefully read this test manual before administering the flow scales.

2. I wil l have had sufficient training using the flow scales prior to administering these

test instruments to test participants (i.e., if a professional, I am in possession of amasters degree or doctorate in sport psychology or a related discipline; if a student,I am working under the supervision of a professional with sufficient training in psy-chometrics).

3. I will have read and will adhere to the ethical principles outlined in the AmericanPsychological Association’s Ethical Principles of Psychologists and Code of Ethics(APA, 1992), especially section 2 (“Evaluation, Assessment, or Intervention”) andsection 6 (“Teaching, Training Supervision, Research, and Publishing”). I have also

reviewed Standards for Educati onal and Psychological Testing (AERA, 1999).4. I understand that when administering the flow scales I will adhere to the standards

required for the protection of human subjects as specified by APA and my institu-tion’s human subjects/ethics review boards. These standards include, but are notlimited to

• the protection of the confidentiality of participants and scores,

• the rights of the participants to receive feedback regarding the purposes of theflow scales,

• the potential uses of the test results, and

• the methods of test score feedback to individual test participants.

5. I understand that before making modifications to the flow scales I will carefullyreview the test manual so as to determine the appropriateness of the modification(s).

• vRESPONSIBILITIES

8/19/2019 Jackson 2004

6/101

8/19/2019 Jackson 2004

7/101

Table of Contents

Introduction and Information for Users of the Flow Scales

SECTION I: DESCRIPTION OF THE FLOW SCALES

Chapter 1. Introducti on to the Flow Construct..............................................................3

Chapter 2. Description of the Flow Dimensions Assessed by the Flow Scales....................7

Chapter 3. Descripti on and Use of the Dispositional Flow Scale-2 (DFS-2) and the Flow

State Scale-2 (FSS-2) ................................................................................................13

SECTION II: THE DEVELOPMENT AND VALIDATION OF THE FLOWSCALES

Chapter 4. Development of the Original Flow Scales ..................................................25

Chapter 5. Development of the Revised Flow Scales (DFS-2 and FSS-2) ......................29

Chapter 6. Psychometric Properties of the Flow Scales..................................................45

Reliability

Construct Validity

Factorial Structure of the Scales

Questionnaire Intercorrelations

Criterion Validity

Relationships with Other Psychological Constructs

Relationships with Performance

SECTION III. SCORING PROFILES ON THE DFS-2 AND FSS-2 ANDSUMMARY

Chapter 7. Scoring Profiles of D ifferent Activi ty Groups on the Flow Scales..................59

Chapter 8. Summary of the Features of the Flow Scales ..............................................71

• viiTABLE OF CONTENTS

8/19/2019 Jackson 2004

8/101

References

About the Authors

APPENDIX A ....................................................................................................79

The DFS-2APPENDIX B ....................................................................................................83

The FSS-2

APPENDIX C ....................................................................................................87

Flow Scale Scoring Keys

Example Profile Sheet

TABLE OF CONTENTSviii •

8/19/2019 Jackson 2004

9/101

Introduction and Information forUsers of the Flow Scales

The flow scales are self-report instruments designed to assess the construct of flow , oroptimal experience . The scales were designed and validated primarily in physical activi-ty settings; therefore, most of the data discussed in this manual was collected fromparticipants in physical activity settings. Nonetheless, we present information in this

manual regarding the psychometric properties of the scales from initial data collectedin the areas of music and the creative performing arts, in addition to sports, exercise,dance, and yoga.

Flow is a construct that both excites and mystifies those seeking to understand andexperience it. Because it represents those moments when everything “comes together”for the performer, it is a much sought-after state. It is not, however, an easy state toachieve for most people; thus it can be perceived as being out of reach and somewhatmysterious. Nonetheless, even though flow cannot be controlled and produced ondemand, it can be understood and is attainable for most people.

The first section of this test manual provides the reader with a description of the flowscales. We begin with a description of flow in Chapter One, where its significance andrelationship to other psychological constructs is discussed.

The flow scales assess nine dimensions of flow. These dimensions are described andillustrated in Chapter Two. From these dimensions, two versions of the flow scales weredeveloped. These two versions—the Dispositional Flow Scale-2 (DFS-2) and the FlowState Scale-2 (FSS-2)—are described in Chapter Three. A rationale for having two ver-sions, plus a description of intended usage of the two types of scales, is presented in this

chapter. Detail is provided about the scoring procedures and the interpretation of scores in Chapter Three. Chapter Three also discusses appropriate audiences andpotential research settings for the flow scales.

• ixINTRODUCTION

8/19/2019 Jackson 2004

10/101

8/19/2019 Jackson 2004

11/101

tions regarding cross-cultural test development and usage can be found in Duda andHayashi (1998) and Gauvin and Russell (1993).

We recommend that wherever possible, the DFS-2 and FSS-2 be used in the formatpresented in this manual. However, we recognize that minor adaptations may be

required to facilitate use of the flow scales across diverse settings. Such adaptationsshould be carefully considered and examined for their consequent psychometric prop-erties. As with translations, any adaptations to the flow scales should consider not onlylinguistic issues but also the conceptual underpinnings of the flow scale items.

• xiINTRODUCTION

8/19/2019 Jackson 2004

12/101

8/19/2019 Jackson 2004

13/101

Section I:Description of the Flow Scales

• 1

8/19/2019 Jackson 2004

14/101

8/19/2019 Jackson 2004

15/101

Chapter 1.Introduction to the Flow Construct

As an optimal psychological state, flow represents those moments when everythingcomes together for the performer. Flow is often associated with high levels of perform-ance and a very positive experience. Csikszentmihalyi (1975) developed the conceptafter investigating the experiences of diverse groups during times when everythingcame together during performance of one’s chosen activity. These activities included

surgery, dancing, chess, and rock climbing. Despite such diversity in settings, there wasconsiderable consistency of responses regarding what was felt during moments thatstood out as being special in some way for the performer.

Since his initial investigations where the term “flow” was chosen to denote these spe-cial absorbing experiences, Csikszentmihalyi (e.g., 1990, 1997) continues to examinethe flow construct and how it is experienced. Flow has been examined in settings thatrange from daily living (Csikszentmihalyi, 1997) to major scientific discoveries(Csikszentmihalyi, 1996). Remarkable consistency has been found in the described

flow experiences of individuals across diverse settings. Flow is regarded as a special psy-chological state, one that brings the recipient much enjoyment. Flow can occur atdifferent levels of complexity, but by definition flow is intrinsically rewarding regard-less of whether it involves a simple game of throw and catch or a complicated anddangerous gymnastics routine. Csikszentmihalyi (1975) referred to the different levelsof flow experience as micro and macro flow. Micro flow experiences were postulated tofit the patterns of everyday life, whereas macro flow was reserved for experiences asso-ciated with higher levels of complexity and demand on the participant. Flow occurswhen one is totally involved in the task at hand. When in flow, the performer feels

strong and positive, not worried about self or of failure. Flow can be defined as an expe-rience that stands out as being better than average in some way, where the individualis totally absorbed in what she or he is doing, and where the experience is very reward-ing in and of itself. This definition covers several characteristics of flow. Thesecharacteristics are more fully described in the next chapter.

• 3CHAPTER ONE

8/19/2019 Jackson 2004

16/101

Crucial to the definition of flow is Csikszentmihalyi’s (1990) concept of challenge-skillbalance. Flow occurs only when the individual moves beyond his or her average expe-rience of challenge and skills and where there is an investment of psychic energy into atask. The challenge-skills balance concept is best described graphically. We present a

model of flow defined in this way in Figure 1.1. When the challenges perceived arematched by a belief in having the skills to meet the challenge, flow can occur.

This challenge-skill balance concept provides an understanding of other experiences inaddition to flow. Understanding how flow relates to other psychological constructs wasan important consideration in developing the flow scales. When evaluating a tooldeveloped to measure a psychological construct like flow, it is appropriate to subject thetool to a construct validity approach as described by Marsh (e.g., 2001) in his thoroughapproach to validation of the self-concept construct. Part of this process involves assess-ing the relationship between the construct of interest—in this instance, flow—andother relevant constructs. Referring back to the challenge-skill balance model of flow,

it can be seen in Figure 1.1 that when challenges outweigh skills, anxiety is predicted.Conversely, when skills outweigh challenges, relaxation closely followed by boredom ispredicted. An absence of significant challenges or skill requirements in a situationbrings on a state of apathy.

SECTION ONE: DESCRIPTION OF THE FLOW SCALES4 •

Challenge

Skills

High

Low

Low

Anxiety

Apathy

Flow

Relaxation -Boredom

High

Figure 1.1. Adapted from Model of the Flow State in Jackson & Csikszentmihalyi (HumanKinetics, 1999). Reprinted with permission.

8/19/2019 Jackson 2004

17/101

Flow represents optimal experience, and Csikszentmihalyi (1990) uses these two termsinterchangeably. We consider the study of optimal experiences to be as important asfocusing on problems, or negative experiences. The resurgence of the positive psychol-ogy approach (Seligman & Csikszentmihalyi, 2000) demonstrates considerable

support for the significance of understanding positive human experiences.Csikszentmihalyi (1990) provides a compelling argument for why flow experiences areimportant. Csikszentmihalyi argues that flow experiences lead to growth and complex-ity in consciousness.

Csikszentmihalyi (1990) argues that after a flow experience, the self becomes morecomplex through the processes of differentiation and integration. A full explanation of this idea is beyond the scope of this manual; however, the interested reader is referredto Csikszentmihalyi’s writings (e.g., 1990, 1993). Briefly, “differentiation” refers toincreases in distinctiveness, and “integration” to growth in communication and cohe-

siveness. Differentiation occurs through the seeking out of new challenges, whileintegration is linked to an increase in skills to meet those challenges. Complexity stemsfrom the optimal development of both differentiation and integration. Flow leads topersonal growth and growth in complexity, and provides a rewarding and inspiringprocess to get there.

• 5CHAPTER ONE

8/19/2019 Jackson 2004

18/101

8/19/2019 Jackson 2004

19/101

8/19/2019 Jackson 2004

20/101

higher skill demands. The structure of sports and competition levels provide continu-al opportunities for extending oneself. For many people, physical activity (be itcompetitive or recreational) provides one of the most concrete opportunities for settingand striving for personal challenges. Challenges and skills, however, can be modified in

any activity, making flow an accessible experience across all domains of functioning.Acti on-awareness merging. When people are asked to describe what it feels like to bein flow, ideas about action-awareness merging surface. As described by a cyclist in thefollowing paragraph, the performer feels as though he or she is one with the activitybeing performed.

“It just doesn’t seem like you’re sitting on a bike. You feel like one piece of machinery working together . . . like you’re part of this machine that you wereborn with, and it’s how you move.”

How does this experience come about? Through total absorption in what you aredoing. Such involvement can lead to a perception of oneness that brings harmony andpeace to an active engagement with a task.

Athletes often describe a sense of effortlessness and spontaneity associated with thisflow dimension of action-awareness merging. Feelings of automaticity are described byathletes, whose well-learnt routines enable them to process subconsciously and pay fullattention to their actions. The unity of consciousness apparent in this flow dimensionillustrates the idea of growth in complexity that results from flow experiences.

Clear goals. Goal setting is a process that, when undertaken correctly, helps move a per-former toward flow. Knowledge of objectives, performance preparation and planning,awareness, and understanding the fine details required for a successful outcome all helpto set the stage for flow. Once in this state, individuals describe knowing clearly whatit is they are supposed to do. When in flow, this clarity of purpose occurs on a moment-by-moment basis, keeping the performer fully connected to the task and responsive toappropriate cues.

Sports provide an excellent setting for actions bound by clear goals and rules. Thestructure of pre-set action allows more attention to be focused on immediate tasks.Personal goals can also be set and continually monitored against this backdrop of in-built goals for action. In fact, it is vital that athletes plan for their performance so that,when the time comes, there is clarity of focus on the particular goals relevant to indi-vidual performers and performances. As a runner describes in the following paragraph,without such goal preparation, flow is unlikely:


8/19/2019 Jackson 2004

21/101

“If you stand up on the block and just expect it to happen, and haven’t thoughtabout what you want to think about, it won’t happen.”

Goals are a necessary part of achieving something worthwhile in any endeavor. Thefocus that goals provide to actions also means that they are an integral component of

the flow experience.

Unambiguous feedback. Hand-in-hand with clear goals comes the processing of howperformance is progressing in relation to these goals. Paying attention to feedback is animportant step in determining whether one is on track toward goals that have been set.When in flow, feedback is easier to receive. The performer receives clear, unambiguousinformation that he or she processes effortlessly, keeping performance heading in theright direction. When in flow, athletes speak of “knowing clearly what to do” and hav-ing “everything click.”

Feedback can come from many sources. For athletes, one of the most important sourcesof feedback is kinaesthetic awareness, or knowing the spatial location of one’s body.This awareness is the internal information an athlete needs to optimize his or hermovements. Recognizing how the quality of a performance relates to an ideal perform-ance enables athletes to know, on a continuous basis, whether their movements arewhat they want them to be. Feedback can come from a range of external sources—fromthe environment in which the performance is occurring to the information providedby competitors or spectators. It is not necessary for feedback to always be positive forflow to occur or continue. When in flow, the nature of clear and immediate feedback

means that adjustments can be made to either keep a performer in flow or enable theperformer to achieve the state of flow. When receiving feedback associated with a flowstate, the performer does not need to stop and reflect on how things are going. Momentby moment, information is seamlessly integrated into performance.

Total concentrati on on the task at hand. This fifth characteristic defines the flow state.When in flow, one is totally focused in the present on a specific task being performed.There are no extraneous thoughts, and the distractibility that often accompaniesinvolvement on any task is wonderfully absent. Experiencing such clear moments pro-vides much satisfaction, which in turn leads to the growth in complexity that resultsfrom flow experiences.

Being totally connected to the task at hand epitomizes the flow state, and is one of itsmost, often, mentioned characteristics. This connectedness relies on a present-centered

• 9CHAPTER TWO

8/19/2019 Jackson 2004

22/101

focus. The past and the future are not roads to flow; flow resides in being able to be in the present moment

An interesting aspect of the concentration experienced in flow is that even though it iscomplete and intense, this type of concentration is spontaneous. In contrast to usual

experiences, no effort is required to keep the mind on task when in flow. In the follow-ing quote, an athlete illustrates this contrast.

“Well, it’s a total focus. But it’s a balance really, between a total focus and a totalrelease in a way. Because you are totally focused . . . on the other hand, it is alsohappening on its own. It is like it is a total automatic process. “

Sense of control. Another frequently mentioned flow characteristic is a feeling of beingin control. One athlete described this feeling as an “unshatterable” self-esteem. Othershave described a sense of infallibility when performing in flow. This empowered feel-

ing frees one from the all-too-frequent fear of failure that creeps into performance.Failure thoughts are nowhere to be found during flow, enabling the individual to takeon the challenges at hand.

Control, like the challenge-skills relationship, is a delicately balanced component of flow. Although the perception of control is inherent to the experience, absolute situa-tional control does not actually exist in an experiential sense. One must experiencechallenge to experience flow. Challenge does not exist under conditions of absolutecontrol. Hence, the experience of total control most likely moves an individual awayfrom the experience of flow and on to relaxation or boredom. The possibility of keep-ing things under control keeps flow active. Like flow itself, the sense of control usuallylasts only a short time. This relates back to keeping at the cutting edge of the challenge-skills balance within a situation. If the feeling of being in control keeps goingindefinitely, then the scales have tipped in favor of skills over challenge, and flow is lost.

Loss of self-consciousness. Most people live their lives surrounded by evaluations of howthey are doing. Emanating from many sources, one of the most insistent is from theself. In situations of importance, it is difficult to stop constantly evaluating how we aredoing in the eyes of others; however, this evaluation is necessary for flow. When an

individual is no longer concerned with what others think of him or her, this individ-ual has lost self-consciousness.

Athletes often find it difficult to lose self-consciousness. The very public nature of theactivity affords many opportunities for evaluation. In any activity, we face criticism —both from others and ourselves — which turns attention away from the task and onto


8/19/2019 Jackson 2004

23/101

the self. The ego, that part of our self that questions, critiques, and prompts self-doubt,needs to be quieted for flow. We can think of flow as unselfconscious action. It is lib-erating to be free of the voice within our head that questions whether we are living upto the standards that we perceive are important to be met.

Transformati on of time. Deep moments of flow seem to transform our perception of time. For some, the experience is that time stops. For others, time seems to slow. Andstill for others, time seems to pass more quickly than expected. These sensations comeabout through the intense involvement of a flow experience. Because nothing else isentering our awareness during flow, we may be surprised to find that significant timehas passed while in the state of flow. The intensity of focus may also contribute to per-ceptions of time slowing, with a feeling of having all the time in the world to executea move that is in reality very much time-limited. Thus, there seems to be a close linkbetween depth of concentration and time transformation.

The latter dimension may be the one least frequently experienced. The lack of associ-ation between time transformation and the other flow dimensions in the sportsresearch conducted to date is discussed in subsequent chapters. It may be that thenature of the sports activity, where time is often part of the infrastructure or part of thechallenge, is not easily lost. Another possible explanation is that this dimension occursonly when the flow experience is very deep. There may be a fleeting perception of thisdimension when it does occur. When this dimension is experienced, it is one of the lib-erating dimensions of flow, to feel free from the time dependence under which we livemost of our lives.

Autotelic experience. Csikszentmihalyi (1990) coined the term autotelic experience todescribe the intrinsically rewarding experience that flow brings to the individual. Asdescribed by Csikszentmihalyi, the word is derived from two Greek words that describedoing something for its own sake: auto = self, and telos = goal. Flow is such an enjoy-able experience that one is motivated to return this state. Once experienced, flowbecomes a much sought after state. Csikszentmihalyi described this dimension as theend result of the other eight flow dimensions. Athletes endorse the enjoyment compo-nent of flow strongly, with descriptions like “such a rush,” “exhilarating,” and “a buzz.”

For many, flow is the defining motivation to keep pushing towards higher limits, asdescribed by a cyclist in the following paragraph.

“There is no experience in sports that is as exhilarating or rewarding as being inflow. That’s what it is. That’s what makes me keep riding — knowing that Imight get it again.”

• 11CHAPTER TWO

8/19/2019 Jackson 2004

24/101

These feelings of great enjoyment may come only after a flow performance; during aflow performance, energy is directed fully into the task. Thus, it is generally uponreflection that the autotelic aspect of flow is realized and provides high motivationtoward further involvement.

Considered together, these nine dimensions of flow provide an optimal experience.Considerable consistency of flow experience has been found across many differentdomains (see Csikszentmihalyi, 1990, 1997; Csikszentmihalyi & Csikszentmihalyi,1988). The dimensions of flow provide a conceptually coherent framework for under-standing optimal experience. The next chapter describes the measurement approachdesigned by Jackson and colleagues to tap into these flow dimensions.


8/19/2019 Jackson 2004

25/101

Chapter 3.Description and Use of the

Dispositional Flow Scale-2 (DFS-2)and the Flow State Scale-2 (FSS-2)

In this chapter, we outline the intended use of the two versions of the flow scales, as

well as describe the composition, administration, and scoring of the scales. This chap-ter also discusses the intended audiences for the scales, as well as potential researchsettings.

By designing two versions of the scales, it is possible to assess flow at both a disposi-tional level and a state level. That is, general tendency to experience flow can beassessed, as well as particular incidence (or non-incidence) of flow characteristics dur-ing a particular event. In accordance with other psychological concepts employing astate-trait distinction, it is proposed that flow is a specific psychological state, amenableto state-based assessments, and also that people differ in their propensity to experienceflow on a regular basis (Jackson, Kimiecik, Ford, & Marsh, 1998). Assessment of fre-quency of flow experiences in a chosen activity affords an assessment of a dispositiontowards flow in that activity.

When administering the DFS-2 and FSS-2, the recommended name for each question-naire is Activi ty Experience Scale and Event Experience Scale, respectively. These namesreflect what is being assessed in general, without biasing respondents according to theirunderstanding of the term flow.

The DFS-2was designed as a dispositional assessment of the flow experience. It assess-es the general tendency to experience flow characteristics within a particular settingnominated by the respondent. It is also possible for investigators to nominate the activ-ity, or setting in which they want participants to respond. The respondent is directedto think about the frequency with which he or she generally experiences the flow itemswithin a particular activity.

• 13CHAPTER THREE

8/19/2019 Jackson 2004

26/101

The suggested instructions for answering the DFS-2 are as follows:

“Please answer the following questions in relation to your experience in yourchosen activity. These questions relate to thoughts and feelings you may experi-ence during participation in your activity. You may experience these

characteristics some of the time, all of the time, or none of the time. There areno right or wrong answers. Think about how often your experience each char-acteristic during your activity and circle the number that best matches yourexperience.”

In order to focus the respondent on one selected activity when answering the scale, thefollowing lead-in statement is included with these instructions: “When participating in(name activity) . . .” The rating scale used for the DFS-2 is a 5-point Likert scale, rang-ing from “1” (never) to “5” (always). The premise for using this type of assessment isthat people who report more frequent occurrence of flow characteristics possess agreater predisposition towards experiencing flow.

Respondents are asked to focus on how they experience one particular activity. Thereare several reasons for this. The first is to provide a context for participants’ responsesand to ground their thinking in a particular setting. Second, the DFS-2 was designedin parallel with the FSS-2, where respondents report flow experience within a particu-lar, just-completed event. The contextualizing of the DFS-2 enables researchers tocompare responses to the same activity across the FSS-2 and DFS-2, and thus examinerelationships between state and dispositional factors in experience. Third, it is likely

that most investigations using the DFS-2 will focus on activities in which the respon-dents have invested psychic energy, activities of importance to the respondents, wherethey are likely to encounter challenge and for which they have developed some skills:that is, activities conducive to flow experiences. Through assessing experience in suchactivities, it is possible that more can be learned about the autotelic personality.

An autotelic person is one who is more able to experience flow and is akin to a person-ality type as described by Csikszentmihalyi (1990, 1997). It could be argued thatpeople invest their energies in activities where they are more likely to experience flow.By investigating individual differences in experience within such self-selected activities,it may be possible to learn more about the factors that contribute to the autotelic per-sonality.

While the DFS-2 is designed for grounding in a particular activity (or type of activity),it should be answered at a time separate from immediate involvement in this activity.


8/19/2019 Jackson 2004

27/101

As a dispositional measure, the DFS-2 is designed to elicit typical responses, or how theperson feels in general about his or her participation in a chosen activity. As a disposi-tional measure, the DFS-2 is designed to assess individual differences in the tendencyto experience flow in specific activities. According to Csikszentmihalyi (e.g., 1990)

people differ in their ability to experience flow. This difference is signified in theautotelic personality as described above. The DFS-2 was designed to tap into thisautotelic personality concept; thus it is anticipated that responses to this instrumentwill remain fairly stable over a long time frame. Future research will need to examinethe temporal stability of the DFS-2.

There are no set timeframes in which the respondent is asked to recall his or her expe-rience; however, it is possible to assign a timeframe by adding this to the instructionspreceding the scale. For example, you can include in the instructions a statement suchas ”Think about your experience in (name of activity) over the past year, and answer

the questions about how you have generally felt while participating.” The timeframemost appropriate to specify for respondents when answering the DFS-2 may dependon the particular characteristics of the sample (for example, their age, amount of timein the activity, frequency of participation and so on).

The FSS-2 is designed as a post-event assessment of flow with instructions worded toground the respondent in the just completed activity. The suggested instructions are asfollows:

“Please answer the following questions in relation to your experience in the

event or activity just completed. These questions relate to thoughts and feelingsyou may have experienced while taking part. There are no right or wronganswers. Think about how you felt during the event or activity and answer thequestions using the rating scale below. For each question, circle the number thatbest matches your experience.”

The lead-in statement, “During the event...” follows these instructions in order tofocus respondents on the just completed activity. The rating scale for the FSS-2 is a 5-point Likert scale, ranging from “1” (strongly disagree) to “5” (strongly agree). That is,respondents are asked to indicate their extent of agreement with each of the flowdescriptors in relation to the activity that he or she just completed. The FSS-2 shouldbe administered as close as possible to the completion of the activity being assessed topromote clear recall. It is recommended that responses to the FSS-2 be collected with-in one hour of completion of the activity, with the aim of gathering the data as closeto the finish of an activity as possible, while minimizing intrusion on the participants.

• 15CHAPTER THREE

8/19/2019 Jackson 2004

28/101

By collecting responses close to the conclusion of an activity, a more accurate assess-ment of the state flow experience is likely.

Another possible use of the FSS-2 is to collect data on particular experiences of signif-icance to the participants. Respondents can be asked to think about a particular

experience (for example, a peak experience) and answer the questions in relation to thisevent. There is research support for strength of memory in relation to personally sig-nificant life events (e.g., Wagenaar, 1986). It can be argued that a high-level flowexperience, such as one tied into a peak performance or peak experience, will remain astrong memory for the recipient. Qualitative research conducted by the authors (e.g.,Jackson, 1996; Eklund, 1994) supports this assertion, with such events being remem-bered as highlights of an athlete’s career. Initial flow scale development researchconducted by Jackson and Marsh (1996) used the recalled optimal experiences of respondents. The recommended usage of the FSS-2, as described above, is to ground

participant responses in a particular event that has just occurred; however, there maybe users of this scale who are particularly interested in understanding, for example,best-ever experiences, which may reside further in the past than the most recently com-pleted event of a respondent.

Flow Scale Items

The items used in both versions of the flow scales follow a similar structure but differprimarily in the tense used; specifically, the FSS-2 items use a past tense, whereas the

DFS-2 items use a present tense. The scales differ in this way to fit with the contextthat each version of the scales assesses.

Both the FSS-2 and DFS-2 contain 36 items. There are four items for each of the ninedimensions of flow. Each dimension comprises a subscale of the total scale. Using itemsfrom the DFS-2, an example of each dimension is provided below:

Challenge-Skill Balance: “I am challenged, but I believe my ski lls wi ll allow me to meet the challenge.”

Action-Awareness Merging: “I make the correct movements wi thout thinking about trying to do so.”

Clear Goals: “I know clearly what I want to do.”

Unambiguous Feedback: “It is really clear to me how my performance is going.”


8/19/2019 Jackson 2004

29/101

Concentration on Task at Hand: “My attention is focused entirely on what I am doing.”

Sense of Control: “I have a sense of control over what I am doing.”

Transformation of Time: “The way time passes seems to be di fferent from normal.” Autotelic Experience: “I really enjoy the experience.”

The 36 items are designed to tap into the nine flow dimensions described in ChapterTwo. In formulating the items, the definition of each flow dimension was analyzedacross several of Csikszentmihalyi’s (1975, 1990, 1993) writings. Other sources of information for the initial item development phase included earlier self-report scalesdesigned to measure flow characteristics (Begly, 1979; Csikszentmihalyi &Csikszentmihalyi, 1988; Privette, 1984; Privette & Bundrick, 1991), as well as quali-

tative descriptions of flow from elite athletes (Jackson, 1992, 1995, 1996).

Scoring of the Flow Scales

There are four items for each of the nine flow dimensions represented in the flow scales.The item numbers are consistent across the FSS-2 and DFS-2, and are listed inAppendix C according to the dimension upon which they load. The scoring procedureis simple. The item scores for each dimension are employed to obtain flow dimensionscores. These dimension scores can be expressed either as a summed score or as an itemmean score. Item mean scores are typically easiest to interpret because the value can beevaluated against the anchor descriptors employed in the scale response format. A totalscale score can also be obtained by adding (or averaging) the scores across all the dimen-sions.

Interpretation of Scores

Two types of scores can be obtained from the flow scales: subscale scores, which repre-sent the dimensions of flow, and total scale scores, which represent a global flowconstruct. Jackson and colleagues (e.g., Jackson & Marsh, 1996; Jackson & Eklund,

2002; Marsh & Jackson, 1999) have discussed the relative merits of using a global ver-sus a dimensional approach to scoring the flow scales. Confirmatory factor analyseshave consistently demonstrated the dimensional approach to be stronger psychometri-cally. However, the global approach has received satisfactory psychometric supportoverall, and there may be instances where a single, global assessment of flow is the

• 17CHAPTER THREE

8/19/2019 Jackson 2004

30/101

information required by users of the scale. In other situations, gaining the more fine-grained detail of the flow experience through reliance on the specific dimension scoreswill be more appropriate.

Issues raised by Jackson and colleagues (e.g., Jackson & Marsh, 1996; Jackson &

Eklund, 2002; Marsh & Jackson, 1999) in relation to the use of a global or dimension-al score include the following measurements: a global measure of flow implies that theweighting given to each specific component is the same across situations and individ-uals; an alternative approach is to assign specific weightings to the nine componentsaccording to their past loadings in factor analytic studies.

As Jackson and Marsh (1996) point out, an optimally weighted average of the specificcomponents explains as much or more variance in any criterion measure as a singleglobal score derived from the same responses. Previous research with the flow scales(Jackson & Marsh, 1996; Jackson & Eklund, 2002; Marsh & Jackson, 1999) demon-strated that, based on analysis of the higher order factor loadings, the nine flowdimensions contribute unequally to the global flow factor. The specific findings are dis-cussed in Chapters Four through Six. At this time, based on the very modest empiricalrelationship of the time transformation subscale to the global flow factor, it is recom-mended that this subscale not be included in calculating the global score.Recommended use of the time transformation sub-scale is discussed in detail inChapter Five.

The flow scales were developed as multidimensional instruments, to facilitate assess-

ment of the flow construct at the level of the nine flow dimensions of which theconstruct is comprised. More information about the flow experience can be obtainedvia this approach compared with a single global score. Thus, where it fits with theresearch questions being addressed, a multidimensional approach to scoring is recom-mended.

As mentioned previously, flow scale scores are easiest to interpret when they are con-sidered as item averages for the total scale or the dimension of interest. The lowestpossible item average score is 1 and the highest is 5. FSS-2 item average scores, forexample, can easily be interpreted against the response format anchors employed in thedata collection. A person completing the FSS-2 is asked to indicate the extent of his orher agreement with each item by selecting the most appropriate response category rang-ing from 1 (strongly disagree) to 5 (strongly agree). Therefore, lower item averagevalues indicate a stronger degree of disagreement with statements proposed and higheritem average values indicate a stronger degree of agreement with statements proposed.


8/19/2019 Jackson 2004

31/101

Low agreement with statements indicative of a flow dimension is suggestive that theperson’s experience was not substantively “flow-like” in nature. Conversely, strongendorsement of FSS-2 dimension statements indicates that the individual was under-going (or had undergone) a substantively “flow-like” experience.

When considering dimension-level scores, there is likely to be variation across thescores obtained for each of the nine flow dimensions. This information may provide anindication of relative importance of the various flow dimensions to the activity beingassessed.

The mid-range score of “3” on the FSS-2 scale represents a “neither agree nor disagree”option. This moderate score may indicate some degree of endorsement of the dimen-sions/total flow experience being assessed. It could, however, also indicate someambiguity regarding relevance of some of the items to the person’s experience of theactivity under consideration. It is nonetheless reasonable to interpret moderate-levelscores as being neither strongly indicative that the person has experienced flow (or oneof its attributes) nor strongly indicative that the person’s experience was devoid of flowattributes.

A similar pattern can be interpreted for scores on the DFS-2, although the context of interpreting the scores is one of frequency of experience, rather than extent of endorse-ment of a specific experience as with the FSS-2. A low range score on the DFS-2indicates that the flow characteristics of this questionnaire are “never” (1) or “rarely” (2)experienced. Again, this lack of endorsement may provide very useful information on

the relevance of the different flow dimensions to the setting or individuals beingassessed.

A moderate score (for example, “3” or “sometimes”) on the DFS-2 indicates that theflow characteristics are experienced some of the time in the respondent’s experience.Because what is being assessed is a state that is difficult to achieve, a moderate scorerange demonstrates that the respondent does report experiencing flow during the nom-inated activity at a better than average frequency.

A high score range on the DFS-2 indicates that the respondent “frequently” (4) to

“always” (5) experiences flow (as assessed by the flow scales) in his or her nominatedactivity. Such individuals may be described as having autotelic personalities, asdescribed by Csikszentmihalyi (1990).

• 19CHAPTER THREE

8/19/2019 Jackson 2004

32/101

Appropriate Audiences and Potential Research Settings for theFlow Scales

The flow scales were designed to assess flow in physical activity settings. This type of

setting was kept in mind when devising the items and instructions for answering thequestionnaire. For example, a qualitative database of athletes’ descriptions of being inflow was used when developing the original items for the scales.

Physical activi ty is used here as an inclusive term. The scales are appropriate to use in avariety of physical activity settings, and research has been conducted with the scalesboth in sports (e.g., Jackson et al., 1998) and exercise (e.g., Jackson & Eklund, 2002;Karageorghis, Vlachopoulos, & Terry, 2000). While there is a specific focus on move-ment in a small number of items, there is no reference to structure of the activity orcompetitiveness, aspects that might have otherwise tied the scales to a sports environ-

ment.

Since their development, interest in using the flow scales has come not only fromresearchers interested in studying optimal experiences in physical activity settings, butalso include music (Wrigley, 2001), web-based instructional activity (Chan & Repman,1999), and computer games (BBCWorld, 2002). The first author has communicatedwith researchers from diverse areas (e.g., gifted education, work addiction, yoga, andbusiness) regarding utilization of the flow scales in their research setting. Moreover,there is considerable interest in examining flow in relationship to other psychological

constructs across diverse settings. Relationships with concepts such as hope, cohesion,personality type, intrinsic motivation, burnout, self-efficacy, self-esteem, anxiety, andso on have all captured the interest of optimal experience researchers. Thus, it is clearthat there is considerable interest in examining flow across a range of settings and inrelation to a diverse set of psychological constructs. It should be pointed out thatCsikszentmihalyi’s (1975) initial comprehensive book about the flow concept includeddata from a variety of settings including surgery, music, dance, sports, and chess. Thisseminal publication gave strong support to the idea of a consistent state of conscious-ness (that Csikszentmihalyi labelled “flow”) across a diverse range of settings. Theutility of the flow scales for assessing this experience across different settings awaits fur-ther empirical research.

Future research must also shed light on the issue of an appropriate age range for use of the flow scales. Research conducted by the authors included a broad age span: 16 to 82years. In order to be confident that respondents understand the concepts being assessed


8/19/2019 Jackson 2004

33/101

by the items, a minimum age of 15 may provide a rough guide to suitability of popu-lations for the scales. There is no upper age range limit for use of the flow scales.

The flow scales have potential utility across a range of research settings. Because flowhas been demonstrated to be a consistent experience across many different walks of life

(Csikszentmihalyi, 1975; Csikszentmihalyi & Csikszentmihalyi, 1988), there is poten-tial applicability of the flow scales across a number of settings. Wherever there isinterest in assessing quality of experience and quality of performance, assessment of flow is a useful inclusion. When a dimensional assessment of flow is desired, the flowscales provide an instrument shown to be valid and reliable. Section II addresses thepsychometric properties of the flow scales.

• 21CHAPTER THREE

8/19/2019 Jackson 2004

34/101

8/19/2019 Jackson 2004

35/101

• 23

Section II:The Development and Validation

of the Flow Scales

8/19/2019 Jackson 2004

36/101

8/19/2019 Jackson 2004

37/101

Chapter 4.Development of the Original

Flow Scales

This chapter outlines the background and development of the original versions of theflow scales. Research by Jackson and colleagues (e.g., 1992, 1995, 1996; Jackson &Marsh, 1996; Jackson, Kimiecik, Ford, & Marsh, 1998; Jackson, Thomas, Marsh, &

Smethurst, 2001) focused on understanding and examining the flow state in physicalactivity. Beginning with qualitative approaches, Jackson (1992, 1995, 1996) exploredthe perceptions elite performers held of flow and how they attained this state duringtheir performances. In an effort to understand the relationship of flow to other psycho-logical factors, Jackson and her colleagues developed self-report instruments to assessflow experiences. The Flow State Scale (FSS; Jackson & Marsh, 1996) andDispositional Flow Scale (Jackson et al., 1998) were designed to assess, respectively,flow experiences within a particular event and the dispositional tendency to experienceflow in physical activity.

When developing the flow scales for primarily a physical activity setting, it was consid-ered important to assess flow tied to a specific context, but to avoid interruptingperformance. Interrupting experience in some settings does not pose significant prob-lems. However, to do so in settings involving substantial physical movement, a highdegree of structure, and for some, competition or high levels of risk, is problematic.Thus, the FSS was designed for administration as soon as possible after completion of the activity.

Another consideration in developing self-report instruments to assess flow involved an

interest in being able to tap into the whole flow experience. Nine flow dimensions havebeen described. These were used as the background structure when forming the FSS.In contrast to this approach is another useful method for assessing flow. This methodis called, The Experience Sampling Method (ESM; Csikszentmihalyi & Larson, 1987).The ESM takes an in-situ approach that focuses primarily on the assessment of chal-

• 25CHAPTER FOUR

8/19/2019 Jackson 2004

38/101

lenges and skills in a situation as operational definition of flow. The ESM provides formore immediate assessment of experience that includes flow as well as other experien-tial states. The flow scales were designed with a different purpose in mind—to assessflow from a multidimensional perspective and to include assessments of specific expe-

riences as well as more general tendencies to experience flow. The ESM and the flowscales provide complementary approaches to assessing flow experience.

Taking the aforementioned multidimensional approach to the measurement of stateflow, Jackson and Marsh (1996) developed a 36-item self-report instrument. Itemswere developed from the nine dimensions of flow described by Csikszentmihalyi(1990). In forming an initial pool of items, earlier self-report scales designed withmeasurement of flow in mind (e.g., Begly, 1979; Csikszentmihalyi & Csikszentmihalyi,1988; Privette, 1984; Privette & Bundrick, 1991) were examined as a reference base.Qualitative research examining the flow construct (e.g., Jackson, 1992, 1995, 1996)

was drawn upon for the phrasing of items. Seven experts in sports and exercise psychol-ogy evaluated this initial item pool. These researchers all had good familiarity with theflow concept as attested by their own research publications. Feedback from the panelof experts led to a 54-item (6 items per scale) instrument. This instrument was admin-istered first to 252 physical activity participants. This pilot study led to identificationof some problematic items. Specifically, several negatively or ambiguously wordeditems were found to be less effective in item analyses and were replaced with more clear-ly stated, positively worded items.

The revised Flow State Scale was then administered to 394 physical activity partici-pants, primarily comprised of athletes. Confirmatory factor analyses of the data (N =381) analysed the fit of both a 54-item model and a shorter 36-item model. The fit of the 36-item model was satisfactory, while the fit of the 54-item model was marginal.With better reliability estimates also attained with the 36-item scale, it was selected asthe final version of the FSS.

The 36-item FSS contains four items for each of the nine flow dimensions.Confirmatory factor analyses demonstrated a satisfactory fit of both a nine first-orderfactor model and one higher order model with a global flow factor. Parameter estimates

provided good support for the nine-factor structure with freely estimated factor corre-lations. The factor loadings were all substantial, ranging from .56 to .88, with a medianloading of .74. Correlations between the nine factors supported the separation intonine flow factors. Although the relationships between the factors were all positive, the

SECTION TWO: TH E DEVELOPMENT AND VALIDATION OF THE FLOW SCALES26 •

8/19/2019 Jackson 2004

39/101

size of the correlations ranged from low to moderate, varying from .18 to .72 (medianr = .50), and supporting the multidimensional model.

In addition to examining the fit of a nine first-order factor model, Jackson and Marsh(1996) assessed a higher order model with one global flow factor. Support was obtained

for this higher order model, albeit not as strong as for the first-order model. All of thenine factors loaded on the higher order factor but with considerable variability in thesize of the loadings. These ranged from .39 for time transformation to .91 for sense of control. Two of the flow dimensions, loss of self-consciousness and time transforma-tion, were found to be weaker factors in these initial analyses of the Flow State Scale byJackson and Marsh (1996).

The DFS was developed subsequent to the FSS to assess individual differences inpropensity to experience flow, using instructions that focused upon the frequency of experience of flow characteristics. This DFS variation of the FSS was developed becauseCsikszentmihalyi and other flow researchers (e.g., see Csikszentmihalyi &Csikszentmihalyi, 1988) have proposed that individual differences exist in the abilityto experience flow. Csikszentmihalyi (1990) suggested that certain types of peoplemight be better psychologically equipped, regardless of the situation, to experienceflow. This individual difference factor is termed the “autotelic personality.” The con-cept of the autotelic personality requires greater elucidation. Several important factorshave been identified such as a desire for challenge (Logan, 1988) and superior concen-tration abilities (Hamilton, as cited in Csikszentmihalyi & Csikszentmihalyi, 1988). Aspart of the process of exploring the constituents of the autotelic personality, Jacksonand colleagues (1998; 2001) have identified psychological factors that are expected torelate to dispositional flow in physical activity settings. These factors are described inChapter Six.

Marsh and Jackson (1999) reported a series of sophisticated confirmatory factor analy-ses to individually and simultaneously evaluate the FSS and DFS measurements.Overall, support was presented for the construct validity of both the state and disposi-tional measures. Item loadings on first order factors ranged from .43 to .89 for FSS(mean = .78), and from .29 to .86 for DFS (mean = .74). Simultaneous modelling of

the DFS and FSS scales provided support for the construct validity of the measures.Observed correlations were substantially higher between matching trait factors andstate factors (.38 to .78, median r = .62) than between non-matching factors in allinstances. The correlation between DFS and FSS loss of self-consciousness factors (r =.38) was the only correlation less than .56. In all cases, non-matching factor relation-

• 27CHAPTER FOUR

8/19/2019 Jackson 2004

40/101

ships were lower than those observed between matching factors. In accordance withother investigations employing the flow scales (e.g., Jackson & Marsh, 1996; Jacksonet al., 2001), internal consistency estimates for both the FSS and DFS were reasonable,varying from .72 to .91 (mean alpha = .85) and from .70 to .88 (mean alpha = .82),

respectively.Marsh and Jackson (1999) found that models involving first-order factors only fit mar-ginally better than models with higher order factors. Higher order factor loadingsranged from .00 to .88 for the FSS (mean = .55) and from .04 to .89 for the DFS(mean = .62). Only one factor had a loading factor of less than .40 on the higher orderfactor. While most higher order factor loadings were reasonable (i.e., > .40), the timetransformation factor did not load on the higher order factor at all. In this sample, thisfactor exhibited essentially no relationship with the global factor in either DFS or FSSmeasurement.

Lack of robust support for the time transformation dimension, as well as for the loss of self-consciousness dimension, was found in two other published studies using a physi-cal activity setting. In a study with masters’ level swimmers, Kowal and Fortier (1999)found that, compared to other flow dimensions, these dimensions were not significant-ly associated with their measures of situational motivation. In their factor analyticstudy with aerobic dance participants, Vlachopoulos et al. (2000) found time transfor-mation and loss of self-consciousness to be less associated with global flow than the restof the flow dimensions.

The analyses of data collected with the original flow scales indicate that, while they per-formed reasonably well on the whole, there were areas with improvement potential. Forexample, in the hierarchical factor analytic model (Jackson & Marsh, 1996; Kowal &Fortier, 1999; Marsh & Jackson, 1999; Vlachopoulos et al., 2000), the original flowscales exhibited relatively weak associations between certain flow dimensions (such asloss of self-consciousness and time transformation) and the global flow factor.Parameter estimates (Jackson & Marsh, 1996; Marsh & Jackson, 1999; Vlachopouloset al, 2000) indicated that a small number of particular items warranted some addition-al conceptual and empirical consideration. Thus, revised versions of the flow scales

were developed to address these concerns.


8/19/2019 Jackson 2004

41/101

Chapter 5.Development of the Revised

Flow Scales (DFS-2 & FSS-2)

When evaluating the measurement qualities of the flow scales, conceptual and statisti-cal issues were considered. As part of the conceptual evaluation, feedback on items inthe original scale was obtained from the developer of the flow model, Csikszentmihalyi

(1975, 1990). From this feedback (Csikszentmihalyi, personal communication, April1997), several issues were considered. For example, in relation to the time transforma-tion dimension, Csikszentmihalyi suggested that the more usual occurrence in flow isa perceptual shortening of time, rather than a lengthening of time. Because two of theoriginal four time transformation dimension items focused on time lengthening whilenone focused on time shortening, it was considered possible that the lack of relation-ship between this dimension and the global flow factor could be due to the wording of the original items. To determine whether changing the focus of the sub-scale wouldimprove its performance, additional items were developed as possible alternatives.

Feedback was also given about the sense of control dimension. Csikszentmihalyi sug-gested that the original formulation of the items put too much emphasis on totalcontrol, rather than on a sense of control, or indeed a lack of worry about control dur-ing one’s performance. This was also considered when formulating additional items.

Following consideration of conceptual issues like those exemplified above, empiricalissues related to the psychometric performance of the questionnaire, its subscales, anditems were examined to ensure that statistical weaknesses in items or higher order fac-tor loadings were also considered. For example, in terms of statistical weakness, the

item that had consistently performed most poorly was one from the loss of self-con-sciousness sub-scale. Item loadings ranged from .29 on the DFS and .43 on the FSS(Marsh & Jackson, 1999) to .56 on the FSS (Jackson & Marsh, 1996). This itemwhich read (DFS version) “I am not worried about my performance during the event,”was designed to tap into a lack of concern for oneself during performance, but was

• 29CHAPTER FIVE

8/19/2019 Jackson 2004

42/101

8/19/2019 Jackson 2004

43/101

• 31CHAPTER FIVE

Table 5.1. Activity types identified by participants in Jackson and Eklund (2002) Study 1

Activity DFS-2 FSS-2 Total N

Aerobics 8 1 9

Australian rules football 5 1 6Basketball 10 4 14Climbing 2 0 2Cricket 3 0 3Cycling 2 2 4Dance 6 3 9Duathlon 0 56 56Equestrian 2 0 2Golf 2 0 2Hockey 11 1 12Ice hockey 19 0 19

Kayaking 2 2 4Martial arts 9 6 15Netball 15 4 19Other individual sport 9 6 15Other sport or activity 4 0 4Other team sport 3 0 3Outrigger canoe 2 1 3Rowing 9 4 13Rugby 19 6 25Running 6 59 65

Soccer 16 8 24Softball 3 1 4Squash 5 2 7Surf boat rowing 23 22 45Swimming 18 9 27Tennis 9 3 12Touch football 88 57 145Track & field 26 15 41Triathlon 13 92 105Volleyball 13 10 23Walking 7 7 14

Water polo 3 0 3Weight training 13 8 21Yoga 1 1 2

Total 386 391 777

8/19/2019 Jackson 2004

44/101

running (N = 65), duathlon (N = 56), surfboat rowing (N = 45), track & field (N =41), swimming (N = 27), rugby (N = 25), soccer (N = 24), and volleyball (N = 23). Atable that lists the activities represented in this sample is provided in Table 5.1. Notethat activities with only one respondent were classified into either “Other Individual

Sports” or “Other Team Sports.”Participation levels also varied, ranging from international (10%) to national (15%),state (24%), and club or school (26%) involvement. There were also participants whoindicated they either saw themselves as individual competitors (10%) or who did nothave any sort of competitive involvement (14%). Participants were recruited from uni-versity undergraduate classes, sport teams, and sport events (such as triathlons). Therewas a standardized information sheet given to all participants, outlining the informedconsent procedures and purpose of the study. The dispositional version of the scale wascompleted at a time separate from participation, while the state version of the scale was

given to participants to complete directly after their activity. For the state version, par-ticipants were asked to indicate the length of time between event completion and thecompletion of the questionnaire. The average time was 24.6 minutes (SD = 25.2).

To select an optimal set of indicators from existing items and potential new itemsdescribed earlier, structural equation modelling procedures that used maximum likeli-hood estimation were employed in an iterative process. Items were loaded uniquelyupon factors in all analyses. For identification purposes, factor variances were fixed tounity in all measurement-model analyses. In higher order factor model analyses, anindicator on each factor was fixed to 1.0, along with the variance of the higher orderfactor. In the selection process, a single item was introduced into a 36-item measure-ment model consistent with those previously reported (e.g., Jackson & Marsh, 1996;Marsh & Jackson, 1999). This process allows the performance of an item to be evalu-ated (for example, item loading, pattern of associated residuals, modification indices)within the context of all other construct indicators. This process was repeated until aconceptually and empirically optimal 36-item solution (4 items per factor) was identi-fied. In the few instances of where item selection was statistically ambiguous,conceptual issues and the advantage of having a consistent set of indicators acrossinventory formats were deciding issues.

Goodness-of-fit in these analyses was evaluated through the use of the χ2 test statisticas well as the Non-Normed Fit Index (NNFI), the Comparative Fit Index (CFI), andthe root mean square error of approximation (RMSEA) (Hoyle & Panter, 1995). Theχ2 provides the basis for statistical tests of the lack of fit resulting from over-identifying


8/19/2019 Jackson 2004

45/101

restrictions placed on models. The NNFI estimates the relative improvement perdegree of freedom of the target model over a baseline model. The CFI assesses the rel-ative reduction in lack of fit as estimated by referencing the non-central χ2 of a targetmodel to a baseline model. The RMSEA assesses the fit function of the target model

adjusted by the degrees of freedom.NNFI and CFI values exceeding .90 and .95 are typically taken to indicate acceptableand excellent model fits to the data (Hoyle & Panter, 1995; Hu & Bentler, 1999). Forthe RMSEA, values of less than .05 and .08 are taken to reflect, respectively, a close fitand a reasonable model fit (Browne & Cudeck, 1993), while the relevant 90% confi-dence intervals provide a useful context for interpretation of the observed point values.Finally, evaluation of parameter estimates (i.e., factor loadings), modification indices,and the pattern of standardized residuals were also crucial in making decisions aboutthe utility and statistical appropriateness of potential new items. Items were considered

to be strong indicators of their factor if they had larger factor loadings, modificationindices suggesting the item loaded simply, and residuals indicating a small discrepancybetween observed and model reproduced correlations for the variable.

Five of 13 new items were selected through these analyses to replace existing items inthe measurement of the flow experience scales. One new item replaced the problemat-ic item identified earlier from the loss of self-consciousness scale. The new item selectedwas “I am not concerned with how others may be evaluating me” (DFS wording). Asecond new item, “ It was really clear to me how my performance was going” (FSSwording), replaced the identified problematic unambiguous feedback item. Two newtime transformation items (“It feels like time goes by quickly;” “I lose my normalawareness of time;” DFS wordings) replaced original items that focused on time slow-ing. Finally, a new sense of control item (“I have a sense of control over what I am

• 33CHAPTER FIVE

Table 5.2. Global Fit Indices from Jackson and Eklund (2002) Study 1 Analyses

n χ2 df NNFI CFI RMSEA 90% CI

Measurement Model (9 First Order Factors)

FSS-2 391 1171.026 558 .915 .925 .053 .049 - .057DFS-2 386 956.859 558 .943 .950 .043 .038 - .048

Higher Order Factor Model (9 Fi rst Order Factors, 1 Second Order Factor)

FSS-2 391 1266.189 585 .910 .917 .055 .050 - .059DFS-2 386 1063.348 585 .935 .940 .046 .042 - .050

8/19/2019 Jackson 2004

46/101

doing;” DFS wording) replaced an original item that had focused on the notion of “total control.”

Table 5.2 presents the goodness-of-fit values for the final set of 36 items (5 new, 31original) that are identified in the item identification analyses for both the first order

factor model and higher order model. Significant χ2 values were observed in allinstances. Nonetheless, both the first-order and the higher order models exhibitedNNFI and CFI values well above .9 and RMSEA confidence interval values suggestingthe .05 criterion as tenable in these analyses. The fit values were slightly better for themodel involving exclusively first-order factors, but the difference is largely inconse-quential.

Parameter estimates are presented from the Study 1 evaluation of the higher ordermodel in Table 5.3. The loadings of items on first-order factors are all substantial, rang-ing from .51 to .89 for the FSS-2 (mean = .78). The corresponding DFS-2 loadingsranged from .59 to .86 (mean = .77). The loading of the first-order factors on the glob-al flow factor is also presented in Table 5.4. They range between .23 and .94 (mean =.66) for the FSS-2 and between .44 and .91 (mean = .71) for the DFS-2. Correlationsobserved in Study 1 between the revised FSS-2 and DFS-2 first-order latent factors arepresented in Table 5.4. They ranged from .13 to .76 (median r = .48) for the FSS-2,and from .24 to .78 (median r = .51) for the DFS-2. These values indicate that the nineflow factors, while sharing common variance as expected, measure reasonably uniqueconstructs.

In summary, these results indicate that revised DFS and FSS scales (referred to hence-forth as the DFS-2 and the FSS-2) demonstrated acceptable factorial validity forassessing dispositional and state flow, respectively. We considered it important to cross-validate the FSS-2 and DFS-2 models to ensure that the results observed in the firststudy were not sample specific. Data for this first study was collected with 49 item ver-sions of the scales. Cross-validation with the final 36-item versions of these scales wasconsidered important to ensure that items behaved appropriately in the context of thefinal measurement presentation format. Study 2 was conducted to address these issues.

Study 2. Cross-val idation sample. Just under 900 (N = 897) physical activity partici-pants completed the new versions of the flow scales identified in Study 1. Mostprovided only DFS-2 or FSS-2 data (n = 798), but a minority provided data on bothof the revised scales (n = 99). The DFS-2 followed the format of its predecessor in thestructure of the questionnaire. Respondents were asked to think about how often theyexperienced each characteristic described in the items and to rate their responses on a


8/19/2019 Jackson 2004

47/101

8/19/2019 Jackson 2004

48/101

5-point Likert scale ranging from 1 (never) to 5 (always). The FSS-2 also asked respon-dents to follow the same structure employed in the original FSS. This structureinvolved indicating the extent of agreement with the items potentially characterizingtheir experience in the event they had just completed. Responses were on a 5-pointLikert scale, ranging from 1 (strongly disagree) to 5 (strongly agree).

The age range for this sample was 16 to 82 years (M = 26.3, SD = 11.1). Males (48%)and females (52%) were approximately equally represented. Eligibility for inclusion inthe sample was as for Study 1, a minimum of twice-per-week participation in one’sactivity. There were 27 activity types ranging from highly competitive sports (such asUS college football) to exercise activities (such as aerobics). This sample included a sub-stantial number of dance and yoga participants, providing responses from physicallyoriented activities that were not sports or exercise. The most commonly reported activ-ities included running (N = 255), dance (N = 177), yoga (N = 99), triathlon (N = 56),Australian rules football (N = 51), basketball (N = 47), American football (N = 46),


Table 5.4. Correlations Among First Order Latent Factors for Jackson and Eklund (2002)Study 1 Analyses

F1 F2 F3 F4 F5 F6 F7 F8

FSS-2 F2 .657F3 .584 .608F4 .408 .509 .596F5 .510 .588 .574 .482F6 .762 .760 .645 .528 .745F7 .235 .325 .316 .260 .397 .449F8 .166 .252 .172 .163 .222 .140 .132F9 .584 .493 .475 .366 .502 .608 .336 .303

DFS-2

F2 .725F3 .642 .594F4 .572 .516 .650F5 .594 .570 .670 .483F6 .768 .776 .720 .623 .727F7 .348 .396 .269 .275 .398 .439F8 .436 .309 .350 .237 .361 .376 .243F9 .648 .511 .638 .442 .611 .567 .251 .458

8/19/2019 Jackson 2004

49/101

• 37CHAPTER FIVE

Table 5.5. Activity types identified by participants in Eklund and Jackson (2002) Study 2

Activity DFS-2 FSS-2 Total N

Aerobics 9 6 15

American football 46 0 46Australian rules football 22 29 51Basketball 28 19 47Cricket 7 0 7Cycling 8 0 8Dance 108 69 177Golf 5 0 5Hockey 20 0 20Martial arts 3 0 3Netball 27 0 27Other individual sport 7 0 7

Other team sport 2 0 2Rowing 2 0 2Rugby 33 0 33Running 26 229 255Soccer 31 0 31Softball 6 9 15Squash 13 0 13Surfing 3 0 3Swimming 15 2 17Tennis 16 0 16

Touch football 17 0 17Track & field 33 0 33Triathlon 21 35 56Volleyball 16 0 16Walking 5 0 5Weight training 7 0 7Yoga 48 51 99

Total 584 449 1033

rugby (N = 33), track & field (N = 33), and soccer (N = 31). Involvement ranged frominternational (5%) to national (11%), US College (16%), state (17%), club or school(23%), as well as participants who indicated being individual competitors (13%) orwho did not compete at all (14%). A table that lists the activities represented in thissample is provided in Table 5.5. Note that activities with only one respondent were

8/19/2019 Jackson 2004

50/101

classified as either “Other Individual Sports,” “Other sports or activity,” or “OtherTeam Sports”.

Participants were recruited from a variety of physical activity settings as described

above, as well as from university undergraduate classes of human movement or psy-chology. A standard introduction sheet was given to all participants, outlining theinformed consent procedures and purpose of the study. The dispositional version of thescale was completed at a time separate from participation, while the state version of thescale was given to participants to complete directly after completing their activity. Forthe state version, participants were asked to indicate the length of time between eventcompletion and completion of the questionnaire. The average time was 24.8 minutes(SD = 26.1).

With a single exception, analyses and fit indices employed in Study 2 were consistentwith the procedures described in Study 1. Specifically, analyses in Study 2 were con-ducted using means and covariances that were obtained via Graham and Hofer’s (1995)EMCOV23 program to manage missing data. This program employs the EM algo-rithm (Dempster, Laird & Rubin, 1977). The EM algorithm implements, by repeatedimputation-estimation cycles, the Full Information Maximum Likelihood (FIML)approach for estimating means and covariance matrices from incomplete data. FIMLtreatment of missing data is a theory-based approach considered to be superior to themean-imputation method (Wothke, 2000) that was employed in Study 1.

Goodness-of-fit values for the DFS-2 and FSS-2 in cross-validation analyses for both anine first order factor measurement model and the higher order global flow factormodel are presented in Table 5.6. The observed fit values for both the first-order andthe higher order models in these cross-validations are satisfactory. The DFS-2 and FSS-2 measurement models exhibit NNFI and CFI values all exceeding .9. RMSEA point


Table 5.6. Global Fit Indices for Jackson and Eklund (2002) Study 2 Analyses

n χ2 df NNFI CFI RMSEA 90% CI

Measurement Model (9 First Order Factors)

FSS-2 422 1177.558 558 .931 .939 .051 .047 - .055DFS-2 574 1427.219 588 .901 .912 .052 .049 - .055

Higher Order Factor Model (9 Fi rst Order Factors, 1 Second Order Factor)

FSS-2 422 1305.374 585 .923 .929 .054 .050 - .058DFS-2 574 1606.487 585 .889 .897 .055 .052 - .058

8/19/2019 Jackson 2004

51/101

• 39CHAPTER FIVE

Table 5.7. Loadings from Jackson and Eklund (2002) Study 2 Analyses

Item/Factor Factor FSS-2 DFS-2

01 F1 - Balance .605 .51410 F1 - Balance .812 .70719 F1 - Balance .809 .74528 F1 - Balance .763 .77102 F2 - Merging .743 .71111 F2 - Merging .848 .73320 F2 - Merging .845 .82829 F2 - Merging .864 .83203 F3 - Goals .779 .71912 F3 - Goals .850 .74721 F3 - Goals .795 .77330 F3 - Goals .758 .70904 F4 - Feedback .736 .72813 F4 - Feedback .785 .797

22 F4 - Feedback .853 .82431 F4 - Feedback .832 .78805 F5 - Concentration .775 .61114 F5 - Concentration .697 .64323 F5 - Concentration .866 .80632 F5 - Concentration .892 .78006 F6 - Control .799 .67515 F6 - Control .799 .71824 F6 - Control .842 .77133 F6 - Control .786 .71807 F7 - Consciousness .854 .76016 F7 - Consciousness .912 .812

25 F7 - Consciousness .780 .63834 F7 - Consciousness .903 .82308 F8 - Time .813 .73317 F8 - Time .871 .82626 F8 - Time .433 .60635 F8 - Time .722 .73209 F9 - Autotelic .849 .68318 F9 - Autotelic .771 .55027 F9 - Autotelic .885 .78936 F9 - Autotelic .898 .779F1 F10 -Flow .819 .821F2 F10 -Flow .704 .718

F3 F10 -Flow .739 .768F4 F10 -Flow .597 .707F5 F10 -Flow .669 .725F6 F10 -Flow .895 .908F7 F10 -Flow .471 .432F8 F10 -Flow .208 .300F9 F10 -Flow .649 .605

8/19/2019 Jackson 2004

52/101

estimate values for these models marginally exceed .05. Nonetheless, RMSEA 90%

confidence intervals surrounding the point estimates indicate that it would be intem-perate to conclude that the RMSEA values do not indicate a close fit of models to data.The higher order factor models exhibit NNFI and CFI values approximating orexceeding .9. RMSEA point estimate values for these models marginally exceed .05.RMSEA 90% confidence intervals indicate that the models provide a reasonable if notclose fit for the data. Overall, the fit values suggest a slightly better fit for the first orderfactor models, particularly for the DFS-2.

Parameter estimates presented in Table 5.7 show good support for the nine flow dimen-sions. The loadings of items on the first-order factors are all substantial, ranging from

.43 to .91 for the FSS-2 (mean = .80). The corresponding DFS-2 loadings ranged from

.51 to .83 (mean = .73). Correlations among the first-order factors ranged from .06 to

.74 (median r = .40) for the FSS-2, and between .16 and .73 (median r = .48) for theDFS-2 (see Table 5.8). Again, the magnitude of these relationships indicate that theflow subscales tap into reasonably unique aspects of the flow experience. Table 5.7


Table 5.8. Correlations First Order Latent Factors for Jackson and Eklund (2002) Study 2 Analyses

F1 F2 F3 F4 F5 F6 F7 F8

FSS-2

F2 .642F3 .559 .538F4 .402 .394 .598F5 .491 .371 .541 .488F6 .737 .656 .620 .522 .630F7 .374 .348 .372 .312 .239 .425F8 .206 .186 .155 .125 .063 .128 .078F9 .614 .358 .459 .285 .487 .579 .322 .268

DFS-2

F2 .707

F3 .606 .505F4 .620 .496 .636F5 .510 .380 .624 .520F6 .727 .675 .654 .612 .715F7 .319 .327 .292 .249 .243 .480F8 .248 .318 .167 .159 .208 .224 .196F9 .484 .373 .514 .333 .540 .540 .254 .345

8/19/2019 Jackson 2004

53/101

reveals that the loadings of the first-order factors on the global flow factor rangebetween .21 and .90 (mean = .64) for the FSS-2 and between .30 and .91 (mean = .67)for the DFS-2.

Summary of Findings from Confirmatory Factor Analyses of Revised Scales

The two studies described above demonstrate that the revised flow scales provide satis-factory tools that can be used to assess dispositional and state flow. These new scales,the DFS-2 and FSS-2, contain five replacement items that provide a more conceptual-ly coherent and statistically sound measurement of the flow dimensions. On the whole,the fit values for the new item set are better than those obtained with the original flowscales.

More recent but yet to be published analyses of FSS-2 and DFS-2 data (i.e., Jackson &Eklund, 2004) are revealing in a couple of important ways. These analyses were con-ducted using a newer version of the EQS program employed in the Jackson and Eklund(2002) report. The newer version of the EQS program allows for FIML treatment of data within the program and the correction of fit estimates (via the Yuan-Bentler scaledstatistic) for violations of multivariate normality. These analyses replicate the satisfac-tory findings reported by Jackson and Eklund (2004). Interestingly, these neweranalyses also suggest that the estimates of fit (uncorrected for violations of multivariatenormality) provided by Jackson and Eklund (2004) were likely substantial underesti-

mates of the actual fit of the model to their data.

More specifically, chi-square values in the new analyses were all significant (consistentwith the analyses reported by Jackson and Eklund, 2002) indicating room for signifi-cant fit improvement for the DFS-2 and FSS-2. Nonetheless, CFI, NNFI, andRMSEA values observed in the new analyses all indicated much more satisfactory fitsthan had been reported in earlier analyses. These findings are very encouraging andsuggest that the measurement provided by the DFS-2 and FSS-2 is even better thanpreviously reported. Full reporting of these results will occur in a Jackson and Eklundmanuscript that is currently in preparation for submission for publication.

Returning to the Study 1 and Study 2 analyses presented in this chapter, the item-iden-tification analyses did not reveal any substantial weaknesses statistically with the scales.Nonetheless, the higher-order factor loadings for time transformation remained rela-tively weak. At the item level, one time transformation item had a relatively weak factor

• 41CHAPTER FIVE

8/19/2019 Jackson 2004

54/101

loading in the cross-validation analysis of the FSS-2. The loading on the DFS-2 cross-validation analysis was reasonable and so it is unclear whether the item is problematicor simply dependent on the situational variation that is part of FSS sampling.Interestingly, the item was one of the new items that focused on time passing quickly.

Overall, the higher order factor loadings for loss of self-consciousness and more so,time transformation on the global flow factor, remained relatively low, despite theintroduction of new items. Jackson and Eklund (2002) suggest several possible reasonsfor this pattern of relationships. In relation to loss of self-consciousness, the self andbody awareness necessary for competent physical performance may cloud the distinc-tion between this level of awareness and what is measured in the loss of self-consciousness sub-scale. For example, a figure skater is concerned with how shepresents herself during her performance, since she is judged on the presentation of herroutine. For performers such as this, there is likely to be low endorsement of a loss of

self-consciousness item like “I am not concerned with how I am presenting myself.”

An unintended but interesting development to the new loss of self-consciousness sub-scale described by Jackson and Eklund (2002) is a self-presentational flavor to how thisdimension is measured in the DFS-2 and FSS-2. The items tend toward a focus on lossof concern with evaluation of self by others. This is a central consideration in loss of self-consciousness and may be particularly relevant in the public realm of sports andphysical activity. The item set probably presents a more narrow definition of loss of self-consciousness than intended by Csikszentmihalyi (1990), who refers to a lack of focusupon information we normally use to represent who we are to ourselves when experi-encing this dimension.

Turning to the time transformation dimension, it has previously been discussed howtime awareness may be part of the challenge to some activities (Csikszentmihalyi,1990). Certainly in some sports, the clock is an integral part of the structure of the sit-uation or the performance evaluation (Jackson & Marsh, 1996; Jackson & Eklund,2002). While improvements in the higher-order loading of time transformation withthe new item set of the DFS-2 and FSS-2 were observed, it remained the dimensionwith the lowest higher-order factor loading on the global flow factor. Future research

may be able to determine whether this dimension is dependent on certain situations ortypes of activities. At the present time, we generally recommend excluding the timetransformation scores from the global flow score due to its lack of substantial relation-ship with the higher order flow latent variable. It is still useful to include the timetransformation items in data collection. The factor has demonstrated good internal


8/19/2019 Jackson 2004

55/101

consistency. It may provide a useful and conceptually relevant measure of the extent towhich respondents perceive a difference in the passage of time during flow experiences.With more data collected on this dimension across different types of settings, it shouldbecome clearer whether there are situations, or types of individuals, where time trans-

formation is a significant aspect of the flow experience.The FSS-2 and DFS-2 are presented as valid self-report instruments to assess theexpression of flow in physical activity. The scales may also have relevance for othertypes of activities and settings that are not physically oriented. Future research willdemonstrate the extent of utility of these self-report instruments.

The final chapter in this section summarizes psychometric properties of the flow scales.

• 43CHAPTER FIVE

8/19/2019 Jackson 2004

56/101

8/19/2019 Jackson 2004

57/101

Chapter 6.Psychometric Properties of the

Flow Scales

Reliability

The reliability, or internal consistency, of the flow scales has been consistently shownto be acceptable. The alphas for each subscale in studies by Jackson and her colleaguesare presented in Table 6.1. In summary, the initial study of the original Flow State Scale(Jackson and Marsh, 1996) found alphas ranging from .80 to .86, with a mean alphaof .83. Subsequent data collections also exhibited similar internal consistency values.Jackson et al. (1998) in their study of master athletes, found alphas ranging from .72to .91 (mean alpha = .85) for the FSS, and from .70 to .88 for the DFS (mean alpha= .82). A study with a cohort of competitive athletes by Jackson et al. (2001) obtainedalphas ranging from .76 to .92 (mean alpha = .85) for the FSS, and from .72 to .89(mean alpha = .81) for the DFS.

DFS-2 and FSS-2 scales appear to exhibit internal

Date post:	07-Jul-2018
Category:	Documents
Upload:	cassia-lmt
View:	237 times
Download:	0 times

Jackson 2004

Documents