Full text of DOI:10.1207/S15327841Mpee0402_4

This article was downloaded by: [The University of British Columbia]
On: 20 November 2014, At: 09:11
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,
UK
Measurement in Physical
Education and Exercise Science
Publication details, including instructions for
authors and subscription information:
http://www.tandfonline.com/loi/hmpe20
Revisiting the Measurement
of Exercise-Induced Feeling
States: The Physical Activity
Affect Scale (PAAS)
Curt L. Lox , Shannon Jackson , Stephen W.
Tuholski , David Wasley & Darren C. Treasure
Published online: 18 Nov 2009.
To cite this article: Curt L. Lox , Shannon Jackson , Stephen W. Tuholski , David
Wasley & Darren C. Treasure (2000) Revisiting the Measurement of Exercise-Induced
Feeling States: The Physical Activity Affect Scale (PAAS), Measurement in Physical
Education and Exercise Science, 4:2, 79-95, DOI: 10.1207/S15327841Mpee0402_4
To link to this article: http://dx.doi.org/10.1207/S15327841Mpee0402_4
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the
information (the “Content”) contained in the publications on our platform.
However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness,
or suitability for any purpose of the Content. Any opinions and views
expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the
Content should not be relied upon and should be independently verified with
primary sources of information. Taylor and Francis shall not be liable for any
losses, actions, claims, proceedings, demands, costs, expenses, damages,
and other liabilities whatsoever or howsoever caused arising directly or
indirectly in connection with, in relation to or arising out of the use of the
Content.

This article may be used for research, teaching, and private study purposes.
Any substantial or systematic reproduction, redistribution, reselling, loan,
sub-licensing, systematic supply, or distribution in any form to anyone is
expressly forbidden. Terms & Conditions of access and use can be found at
http://www.tandfonline.com/page/terms-and-conditions

MEASUREMENT IN PHYSICAL EDUCATION AND EXERCISE SCIENCE, 4(2), 79–95
Copyright © 2000, Lawrence Erlbaum Associates, Inc.
Revisiting the Measurement of
Exercise-Induced Feeling States: The
Physical Activity Affect Scale (PAAS)
Curt L. Lox and Shannon Jackson
Department of Kinesiology and Health Education
Southern Illinois University Edwardsville
Stephen W. Tuholski
Department of Psychology
Southern Illinois University Edwardsville
David Wasley
Department of Kinesiology and Health Education
Southern Illinois University Edwardsville
Darren C. Treasure
Department of Exercise Science and Physical Education
Arizona State University
Interest in acute exercise-induced feeling states has intensified in recent years, sig-
naling the need for development of domain-specific measures of these constructs. In
response to this call, 2 exercise-specific measures of feeling states have been intro-
duced—namely, the Exercise-Induced Feeling Inventory (EFI; Gauvin & Rejeski,
1993) and the Subjective Exercise Experiences Scale (SEES; McAuley &
Courneya, 1994). Although certain subscales of the 2 instruments are unique, a
number of similar constructs exist both within and between instruments. The pur-
pose of this study, therefore, was to (a) combine the EFI and SEES scales into a sin-
gle, more encompassing instrument and (b) investigate potential factor redundancy
within and between the 2 instruments. Initial correlational analyses indicated that
the Positive Well-Being subscale of the SEES and the revitalization and Positive
Engagement subscales of the EFI were highly correlated. Similar results were ob-
tained for the Fatigue (SEES) and Physical Exhaustion (EFI) subscales. These find-
Requests for reprints should be sent to Curt L. Lox, Department of Kinesiology and Health Educa-
tion, Box 1126, Southern Illinois University Edwardsville, Edwardsville, IL 62026–1126. E-mail:
clox@siue.edu

80
LOX ET AL.
ings provided the impetus for the construction of the Physical Activity Affect Scale
(PAAS), an instrument consisting of the Psychological Distress subscale of the
SEES and the subscales of the EFI. Exploratory and confirmatory factor analyses
demonstrated support for 4 components of exercise-induced affect that were subse-
quently labeled positive affect, negative affect, fatigue, and tranquility. The PAAS
appears to be an improvement over the EFI and SEES in that it is a more encompass-
ing measure of exercise-induced feeling states with less component redundancy
than its predecessors.
Key words: exercise, affect, feeling states, measurement
Duringthe1980s,popularsocialcognitivemodelsoriginallyproposedbysuchpromi-
nent theorists as Bandura (1977) and Weiner (1979) were reformulated to account for
the important role of emotional well-being (Bandura, 1986; Weiner, 1986). Around
thesametime,Russell(1980),WatsonandTellegen(1985),andThayer(1989)devel-
oped separate theoretical frameworks for better understanding the concepts of emo-
tion, affect, and mood. Based on the early theorizing of Schlosberg (1952), Russell
(1980) introduced a circumplex model of affect in which 28 distinct affective states
could be represented along the perimeter of a circle incorporating two bipolar axes re-
flecting pleasure–displeasure and arousal–sleepiness. This bipolar scheme produced
at least two important tenets that would lay the foundation for future theory and re-
searchfocusingonaffect.First,affectiveexperienceslieonacontinuumfrompositive
to negative hedonic tone. Second, regardless of whether they are positive or negative,
affective states lie on a continuum from low to high activation (intensity). A series of
studiesreportedbyRussell(1980)providedinitialempiricalsupportforthemodeland
indicated that the set of primary affective states could, in fact, be located along the
circumplex according to the hedonic tone and intensity properties of each. For exam-
ple, excitement would fall within the quadrant labeled positive-high activation,
whereas tranquility would be considered a positive-low activation state. Similarly,
frustration would represent negative-high activation affect, whereas depression
would be reflective of a negative-low activation state.
Support for Russell’s bipolar consideration of affect has also been provided by
Thayer (1989), who proposed a model of mood with continuums of high–low en-
ergetic arousal and high–low tense arousal. Most notably, Thayer addressed the
exercise–mood relation by suggesting that exercise is capable of regulating mood
due to the enhancement of energetic arousal and the reduction of tense arousal.
Thayer further espoused the importance of exercise by stating that “the single most
important natural mood modulator is exercise, and the most effective way of en-
hancing mood is through increased exercise” (p. 171).
Taken together, the works of these authors have stimulated considerable re-
search in a number of cognitive and social science domains. In the physical activ-
ity literature, initial investigations focused solely on the benefits of regular
exercise on chronic mood states. More recently, however, researchers have begun

PHYSICAL ACTIVITY AFFECT SCALE
81
to explore acute exercise-induced feeling states due to their potential impact on ex-
ercise behavior and certain aspects of mental health (Dishman, 1982; Gauvin &
Rejeski, 1993; Rejeski, 1992). To this end, exercise-induced psychological and
emotional states have been highlighted as important moderator, mediator, and out-
come variables in the field of exercise psychology (Gauvin & Spence, 1998). In
other words, feeling states may be operationalized as dependent variables follow-
ing an exercise bout or as independent mediator and moderator variables used to
predict behavior (e.g., long-term physical activity habits).
Although the theoretical tenets of the general psychology models were readily
applicable to the exercise setting, concern emerged that the most popular self-re-
port measurement tools such as the State–Trait Anxiety Inventory (Spielberger,
Gorsuch, Luschene, Vagg, & Jacobs, 1983), Profile of Mood States (McNair, Lorr,
& Droppleman, 1981), and Positive Affect Negative Affect Schedule (Watson,
Clark, & Tellegen, 1988) were not capable of accounting for the unique stimulus
properties of exercise (e.g., physical work, bodily movements, and perceptions).
In response to this need for domain-specific measures, Hardy and Rejeski (1989)
developed the Feeling Scale (FS) as a simple, quick, and global measure of acute
exercise-induced feeling states. Unfortunately, the single-item, bipolar design of
the scale is problematic for a number of reasons. First, the design contradicts cur-
rent theoretical positions regarding the multidimensional nature of exercise-in-
duced feeling states. Second, many of the commonly accepted indicators of
validity and reliability (e.g., factor analysis, comparing model fit statistics, internal
consistency) cannot be ascertained from a single-item scale. Finally, it is not possi-
ble to determine the potential mechanism(s) responsible for changes in global
“positive” and “negative” feelings that occur as a function of exercise. In other
words, the FS cannot address the extent to which improvements in feeling states
are generated by physiological (e.g., physical exhaustion) and/or cognitive stimuli
(e.g., positive affect).
With these limitations in mind, Gauvin and Rejeski (1993) introduced the Exer-
cise-Induced Feeling Inventory (EFI), a multidimensional measure designed to as-
sess feeling states that occur in conjunction with acute bouts of physical activity.
The EFI was designed to capture four distinct feeling states—namely, positive en-
gagement, revitalization, physical exhaustion, and tranquility. Subsequent statisti-
cal analyses, however, yielded equivocal support for this conceptualization and
suggested that the Positive Engagement and Revitalization subscales may not be
discrete factors (Gauvin & Rejeski, 1993; Rejeski, Reboussin, Dunn, King, &
Sallis, 1999; Vlachopoulos, Biddle, & Fox, 1996). Support for this proposition is
clearly evident in Gauvin and Rejeski’s (1993) initial validation article. First, a rel-
atively large number of items cross-loaded saliently on the positive engagement
and revitalization components. Second, the subscale intercorrelation between the
two factors was high (r = .68, p < .001). Finally, an exploratory factor analysis (ex-
tracting all components with an eigenvalue greater than 1.0) indicated support for a

82
LOX ET AL.
three-factor structure with a combination of positive engagement and revitaliza-
tion items representing a single component.
The argument for separate positive engagement and revitalization components
was further weakened by the authors’ results of the forced extraction factor analy-
sis. The first two components that emerged were clearly representative of physical
exhaustion and tranquility. In each case, all items loaded strongly on their respec-
tive factor with no salient cross-loadings. The third component consisted of the
positive engagement items Enthusiastic and Upbeat as well as the revitalization
items Energetic, Refreshed, and Revived. In addition, the positive engagement
items Enthusiastic, Upbeat, and Happy, as well as the revitalization item Ener-
getic, loaded saliently on the fourth component extracted. In sum, the results of
various factor analytic procedures conducted by the authors of the EFI suggest that
significant overlap and component redundancy exist and that positive engagement
and revitalization may be best conceptualized as a single component. Unfortu-
nately, researchers employing the EFI (Bozoian, Rejeski, & McAuley, 1994;
Gauvin, Rejeski, & Norris, 1996; Rejeski, Gauvin, Hobson, & Norris, 1995; Trea-
sure & Newbery, 1998) have failed to conduct factor analyses to confirm the pro-
posed component structure.
Shortly after publication of the EFI, McAuley and Courneya (1994) proposed
the Subjective Exercise Experiences Scale (SEES) as a more global measure of ex-
ercise-induced feeling states. Based on the popular two-factor (i.e., Positive and
Negative) structure of affect, the SEES comprises positive well-being and psycho-
logical distress, as well as a third factor (Fatigue), presumed to be salient in the ex-
ercise setting. Subsequent research employing the SEES has provided consistent
and strong support for the proposed structure in a number of populations (Lox,
McAuley, & Tucker, 1996; Lox & Rudolph, 1994; Markland, Emberton, &
Tallon, 1997). Of interest, however, is the authors’ suggestion that the subscales of
the EFI “may well represent underlying structural aspects of the more general psy-
chological responses assessed by the SEES” (p. 173). Although the Psychological
Distress subscale is clearly a unique and highly relevant inclusion, no empirical
evidence exists to support the contention that positive well-being and fatigue mea-
sure different (i.e., more global) constructs than similar subscales found in the EFI
(i.e., positive engagement and physical exhaustion, respectively).
The EFI and SEES are important first steps in the attempt to construct a feeling
state inventory sensitive to the stimulus properties of exercise. However, both in-
struments are, according to Russell’s (1980) conceptualization of affect, incom-
plete. Specifically, the EFI lacks a subscale that assesses negative feeling states,
whereas the SEES fails to include measures of tranquility and revitalization. Thus,
researchers must employ both instruments (each with different response formats)
to assess the full domain of affective states predicted by Russell’s (1980) model.
Accordingly, the purpose of this series of studies was to investigate the utility of
merging the EFI and SEES into a single encompassing instrument. The secondary

PHYSICAL ACTIVITY AFFECT SCALE
83
purpose was to investigate potential factor redundancy within and between the two
instruments. To accomplish these tasks, this investigation is divided into three sep-
arate studies. Study 1 consists of a correlational analysis of the EFI and SEES
subscales. Study 2 consists of an exploratory factor analysis of a hybrid EFI–SEES
instrument. Finally, Study 3 consists of a series of confirmatory factor analyses
comparing the EFI, SEES, and the newly constructed Physical Activity Affect
Scale (PAAS), a shortened version of the hybrid instrument utilized in Study 2.
STUDY 1: CORRELATIONAL ANALYSES OF THE EFI
AND SEES
Participants and Procedure
To address the propositions put forth by McAuley and Courneya (1994) con-
cerning the relation between the EFI and SEES, the two instruments were ad-
ministered to 86 male and 220 female university students (M age = 23.43, SD =
4.72) attending a medium sized public university in the Midwest. Participants
were asked to indicate the degree to which they were currently experiencing
each of the 16 items by circling a number corresponding to the respective re-
sponse format of each scale. For the EFI, responses ranged from 0 (do not feel)
to 4 (feel very strongly). SEES responses ranged from 1 (not at all) to 7 (very
much so). Participants were enrolled in one of eight activity courses (walking,
jogging, aerobic dance, step aerobics, weight training, badminton, soccer, or
cross training), each lasting for a total of 45 min.
Results and Discussion
Descriptive statistics for all items are presented in Table 1. Bivariate correlations
indicated that each of the SEES subscales was significantly correlated with each of
the EFI subscales (see Table 2), with correlation coefficients ranging from –.291 to
.810 (M = .500). These results suggest that the EFI and SEES subscales are signifi-
cantly related, and in some instances, share almost 66% of their variability. Most
revealing, however, are the specific relations among (a) positive engagement, revi-
talization, and positive well-being and (b) physical exhaustion and fatigue. Con-
cerning the former set of factors, results indicate that the EFI subscales (Positive
Engagement and Revitalization) share over 63% and 65% of common variability,
respectively, with positive well-being (SEES). Additionally, the Revitalization and
Positive Engagement subscales share 73% of common variability. These findings
suggest that the Revitalization and Positive Engagement subscales represent the
same entity and, further, that this entity is significantly correlated, and shares sub-
stantial variability, with the Positive Well-Being subscale. In a similar fashion,
physical exhaustion (EFI) and fatigue (SEES) were found to share 63% of common

84

85

86
LOX ET AL.
variability, suggesting that these two subscales may also be measuring a common,
underlying construct. This proposition is further supported by the rather simplistic
observation that both the Physical Exhaustion and Fatigue subscales contain the
items Tired and Fatigued.
In summary, the findings reported earlier refute the proposition that the EFI and
SEES measure different levels of affective response (McAuley & Courneya,
1994). Indeed, tremendous overlap appears to exist between the two instruments.
Given this, a merging of the two instruments was deemed to be an appropriate and
acceptable procedure for constructing a more encompassing measure of exer-
cise-induced feeling states.
STUDY 2: EXPLORATORY FACTOR ANALYSIS OF THE
HYBRID INSTRUMENT
Participants and Procedure
The first step in the development of the proposed instrument was to combine the
four items from the Psychological Distress subscale of the SEES (Awful, Crummy,
Discouraged, and Miserable) with the 12 items (four subscales) of the EFI. The hy-
brid instrument was administered to 68 male and 129 female university students (M
age = 23.16 years, SD = 5.41) enrolled in one of six activity courses (walking, jog-
ging, aerobic dance, step aerobics, weight training, or cross training), each lasting
for a total of 45 min.
Results and Discussion
A maximum likelihood exploratory factor analysis with promax rotation was
performed on the 16-item hybrid instrument with forced extraction of four fac-
tors. Results demonstrated support for such a structure (no salient cross loading
present) with components reflecting positive affect, negative affect, fatigue, and
tranquility (Table 3). This solution accounted for 71.98% of the total variance
and manifested good to excellent internal consistency. Specifically, alpha coeffi-
cients for positive affect, negative affect, fatigue, and tranquility were .94, .86,
.91, and .84, respectively.
The first factor extracted, Positive Affect, contained all items from the Revital-
ization (EFI) and Positive Engagement (EFI) subscales. Negative Affect was ex-
tracted next and comprised all items from the Psychological Distress (SEES)
subscale. Fatigue, the third factor, consisted of all items from the Physical Exhaus-
tion (EFI) subscale. Finally, Tranquility comprised two items from the Tranquility
(EFI) subscale (Calm, Relaxed), with Peaceful falling just below the .40 criteria
for inclusion. In short, the results of the exploratory factor analysis support those
of the correlational analyses in Study 1 and suggest that the Positive Engagement

PHYSICAL ACTIVITY AFFECT SCALE
87
TABLE 3
Maximum Likelihood Exploratory Factor Analysis With Promax Rotation (Hybrid Instrument)
Extraction of Four Factors
1
2
3
4
M
SD
Positive Affect
Refreshed
Enthusiastic
Energetic
Happy
.031
–.046
.046
–.174
.172
–.176
.141
–.066
.155
–.169
.053
.145
–.038
.067
.049
–.017
–.186
2.14
2.11
2.25
2.55
2.09
2.06
1.22
1.15
1.21
1.21
1.21
1.23
.646
.947
.803
.704
.763
.902
Revived
Upbeat
.036
Negative Affect
Awful
Crummy
Discouraged
Miserable
Fatigue
.002
–.048
.115
.068
.058
–.069
–.031
0.415
0.384
0.436
0.378
0.855
0.779
0.812
0.862
.690
.888
.697
.778
.054
.013
–.030
.066
–.039
Fatigued
Tired
Worn-out
Tranquility
Calm
.093
–.083
.012
–.039
.067
.136
.018
.081
–.066
1.23
1.40
1.18
1.12
1.21
1.23
.857
.741
.703
–.071
.266
.335
.033
.018
–.097
.038
–.057
.073
1.13
1.17
1.08
1.00
.512
.381
2.25
2.03
2.51
Relaxed
Peaceful
Eigenvalue
% variance
5.92
37.01
3.22
20.12
1.36
8.51
1.01
6.33
Note. Total variance explained by rotated components = 71.98%; all loadings of .40 and higher are
presented in bold.
and Revitalization subscales are best conceptualized as a single factor. The find-
ings also offer preliminary support for a hybrid instrument that assesses four dis-
crete components of exercise-induced affect.
STUDY 3: CONFIRMATORY FACTOR ANALYSES OF
THE EFI, SEES, AND PAAS
Participants and Procedure
Becauseitwasdesirabletoachieveconsistencyinthenumberofsubscaleitemsofthe
PAAS, further modification of the instrument was required. To reduce the number of
items in each of the positive and negative affect subscales, we retained those items
withthehighestloadingsineachcase. Withthisinmind, theitemsEnthusiastic, Ener-

88
LOX ET AL.
getic, and Upbeat were chosen to make up the Positive Affect subscale; Discouraged,
Crummy, and Miserable were selected for the Negative Affect subscale. In addition,
threeitemseachfromtheTranquility(Calm,Peaceful,Relaxed)andPhysicalExhaus-
tion (Tired, Fatigued, Worn-out) subscales of the EFI were included. Thus, the newly
constructed PAAS consists of four subscales (Positive Affect, Negative Affect, Fa-
tigue, Tranquility), each containing three items (see Appendix).
The instruments were administered to 89 male and 203 female university stu-
dents (M age = 22.56 years, SD = 5.07) immediately following participation in one
of eight activities (walking, jogging, aerobic dance, step aerobics, weight training,
badminton, soccer, or cross training), each lasting for a total of 45 min.
Results and Discussion
Employing the PROC CALIS procedure within the SAS statistical software pro-
gram (Statistical Analysis Systems Institute, 1990), three maximum likelihood
confirmatory factor analyses were performed on the correlation matrix to assess the
fit of the PAAS model and compare the adequacy of the PAAS with the EFI and
SEES (see Figures 1, 2, and 3). Five indexes of fit were examined including the root
mean square residual (RMSR), the root mean square error approximation
(RMSEA), the goodness of fit index (GFI), the comparative fit index (CFI), and the
expected cross-validation index (ECVI). Based on the criteria proposed by Browne
and Cudeck (1993), an RMSR or RMSEA of .05 or less was considered a good fit
with values between .05 and .08 representing an acceptable fit. GFI and CFI calcu-
lations of .90 and above were considered to be a good fit. Finally, ECVI estimates
were computed on each model. The ECVI estimate is an indicator of how well a
model, based on the current sample covariance matrix, would fit in a cross-valida-
tion sample of the same size. To interpret this statistic, the ECVI is computed for a
number of competing models, and the model that produces the lowest ECVI esti-
mate is said to be more desirable (Jöreskog & Sörbom, 1993). It should be noted
that, although this estimate allows the researcher to rank order particular models in
terms of desirability, it does not provide an absolute measure of fit.
An examination of the individual descriptive statistics indicated a moderate
amount of univariate skewness and kurtosis in some items (see Table 1). Because
excessive departures from multivariate normality can result in less valid fit in-
dexes and estimated model parameters (Hoyle & Panter, 1995), analyses on trans-
formed data were performed before proceeding with further statistical procedures.
Several analyses were conducted using log and square root transformations of in-
dividual items. Transformations were applied to all variables, including those that
did not show high levels of skewness and/or kurtosis. Although the transforma-
tions were successful for reducing these levels in all variables, they were particu-
larly helpful for those maintaining the highest degrees of skewness and kurtosis. In
each case, there were no significant departures in fit indexes or model parameters

PHYSICAL ACTIVITY AFFECT SCALE
89
FIGURE1 Confirmatoryfactoranalytic model forthe Exercise-Induced Feeling Inventory.
from the original, nontransformed data analyses reported later. Thus, for the sake
of brevity and interpretation, we report only those analyses obtained from the orig-
inal, nontransformed data.
The results of the confirmatory factor analyses are presented in Table 4. In gen-
eral, the results indicated that (a) the SEES model exhibited slightly better fit in-

90
LOX ET AL.
FIGURE 2 Confirmatory factor analytic model for the Subjective Exercise Experiences
Scale.
dexes than the PAAS and (b) the PAAS and SEES models both exhibited a better
overall fit of the data than the EFI model. Because we were unable to conduct
chi-square difference tests (see Loehlin, 1987, for a discussion), the aforemen-
tioned model comparisons may only be interpreted descriptively.

PHYSICAL ACTIVITY AFFECT SCALE
91
FIGURE 3 Confirmatory factor analytic model for the Physical Activity Affect Scale.
GENERAL DISCUSSION
Based on the pioneering work of Gauvin and Rejeski (1993) and McAuley and
Courneya (1994), the purpose of this study was to (a) merge the two most recent
and promising measures of exercise-induced feeling states into a single, encom-

92
LOX ET AL.
TABLE 4
Confirmatory Factor Analyses for Exercise-Induced Feeling Inventory (EFI), Subjective
Exercise Experiences Scale (SEES), and Physical Activity Affect Scale (PAAS) Models
Model
RMSR
RMSEA
GFI
CFI
ECVI
EFI
SEES
PAAS
.07
.04
.05
.13
.07
.09
.83
.92
.90
.86
.95
.92
1.28
.69
.82
Note. RMSR = root mean square residual; RMSEA = root mean square error of approximation; GFI
= goodness of fit index; CFI = comparative fit index; ECVI = expected cross-validation index.
passing instrument and (b) explore potential factor redundancy in the EFI and
SEES instruments. Correlational and factor analytic results from this study have
produced two important conclusions. First, several factors within and across the
instruments are highly correlated. Indeed, it appears that the Revitalization and
Positive Engagement subscales may be better conceptualized as a single compo-
nent of exercise-induced feeling states. Furthermore, these factors appear to also
be redundant with respect to the Positive Well-Being subscale of the SEES. In a
similar vein, Physical Exhaustion (EFI) and Fatigue (SEES) were found to es-
sentially measure the same construct. In short, the validity of the original
four-factor structure of the EFI, and the proposed conceptual differences be-
tween the SEES and EFI, remain in question.
Second, the results of this study offer preliminary evidence of a more encom-
passing measure of exercise-induced feeling states that (a) corrects for the redun-
dancy in factors described earlier, (b) effectively merges the EFI and SEES
instruments, and (c) boasts fit indexes that are comparable (i.e., SEES), if not supe-
rior (i.e., EFI), to those of its predecessors. The factor structure of the PAAS is also
well supported, theoretically, by the four quadrants of the circumplex model of af-
fect proposed by Russell (1980). Specifically, positive affect items would be ex-
pected to fall within the positive-high activation quadrant, negative affect items
are predicted to fall within the negative-high activation quadrant, fatigue items
should be contained within the negative-low activation quadrant, and tranquility
items are hypothesized to fall within the positive-low activation quadrant. Further
support for domain-specific measures of affect such as the PAAS, EFI, and SEES
may be obtained when the predictive utility of popular theoretical models incorpo-
rating affect has been enhanced. For instance, in a test of Bandura’s (1986) social
cognitive theory, exercise-specific self-efficacy may be a better predictor of affec-
tive states known to be influenced by exercise than more general affective states
assessed by instruments such as the PANAS, POMS, or STAI.
From a utility standpoint, the brevity of the EFI and SEES has been maintained
in the PAAS, thereby satisfying the researcher’s need for a multidimensional mea-
sure of exercise-induced affect that requires very little time to complete. This issue

PHYSICAL ACTIVITY AFFECT SCALE
93
becomes particularly salient when multiple assessments of affect are required in
the research design. For example, the scale would be well-suited for assessments
of affect taken at various times before (baseline), during, and after exercise. Such a
design would allow researchers to go beyond simply describing affective experi-
ences following exercise by providing a measure of affective change. Addi-
tionally, because the response format of the EFI has also been maintained in the
PAAS, researchers should find the PAAS just as easy to administer and interpret as
its predecessors. Finally, on a more practical note, the use of the term physical ac-
tivity in the naming of the PAAS was somewhat strategic in that we believe future
research will not be limited to traditional definitions of exercise and physical edu-
cation (i.e., sport and fitness activities). Instead, activities such as yoga, tai chi,
dancing, and hiking may comprise additional modalities from which assessments
of affect may be obtained.
Although we believe that the PAAS is an advancement in exercise-induced
feeling state measurement, considerable validation work still remains. Spe-
cifically, additional investigations should be performed to examine the stability of
the PAAS component structure as a function of factors such as mode and intensity
of exercise and participant characteristics including age, gender, fitness level, and
stage of exercise behavior (Gauvin & Spence, 1998). It may also be a worthwhile
endeavor to modify the instructions to assess more chronic feeling states. For ex-
ample, it would be of interest to examine whether or not relatively permanent
changes have occurred in the PAAS feeling states following a prolonged regimen
(weeks, months) of regular exercise.
REFERENCES
Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Re-
view, 84, 191–215.
Bandura, A. (1986). Social foundations of thought and action. Englewood Cliffs, NJ: Prentice Hall.
Bozoian, S., Rejeski, W. J., & McAuley, E. (1994). Self-efficacy influences feeling states associated
with acute exercise. Journal of Sport and Exercise Psychology, 16, 326–333.
Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Bollen & J. S.
Long (Eds.), Testing structural equation models (pp. 136–162). Newbury Park, CA: Sage.
Dishman, R. K. (1982). Compliance/adherence in health-related exercise. Health Psychology, 2,
237–267.
Gauvin, L., & Rejeski, W. J. (1993). The exercise-induced feeling inventory: Development and initial
validation. Journal of Sport and Exercise Psychology, 15, 403–423.
Gauvin, L., Rejeski, W. J., & Norris, J. L. (1996). A naturalistic study of the impact of acute physical ac-
tivity on feeling states and affect in women. Health Psychology, 15, 391–397.
Gauvin, L., & Spence, J. C. (1998). Measurement of exercise-induced changes in feeling states, affect,
mood, and emotions. In J. L. Duda (Ed.), Advances in sport and exercise psychology measurement
(pp. 325–336). Morgantown, WV: Fitness Information Technology.
Hardy, C. J., & Rejeski, W. J. (1989). Not what, but how one feels: The measurement of affect during ex-
ercise. Journal of Sport & Exercise Psychology, 11, 304–317.

94
LOX ET AL.
Hoyle, R. H., & Panter, A. T. (1995). Writing about structural equation models. In R. H. Hoyle (Ed.),
Structural equation modeling: Concepts, issues, and applications (pp. 158–176). Thousand Oaks,
CA: Sage.
Jöreskog, K., & Sörbom, D. (1993). LISREL 8: Structural equation modeling with the SIMPLIS com-
mand language. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Loehlin, J. C. (1987). Latent variable models. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Lox, C. L., McAuley, E., & Tucker, R. S. (1996). Physical training effects on acute exercise-induced
feeling states in HIV–1–positive individuals. Journal of Health Psychology, 1, 235–240.
Lox, C. L., & Rudolph, D. L. (1994). The Subjective Exercise Experiences Scale (SEES): Factorial va-
lidity and effects of acute exercise. Journal of Social Behavior and Personality, 9, 837–844.
Markland, D., Emberton, M., & Tallon, R. (1997). Confirmatory factor analysis of the Subjective Exer-
cise Experiences Scale among children. Journal of Sport and Exercise Psychology, 19, 418–433.
McAuley, E., & Courneya, K. S. (1994). The subjective exercise experiences scale (SEES): Develop-
ment and preliminary validation. Journal of Sport and Exercise Psychology, 16, 163–177.
McNair, D. M., Lorr, M., & Droppleman, L. F. (1981). Profile of Mood States. San Diego, CA: Educa-
tional and Industrial Testing Service.
Rejeski, W. J. (1992). Motivation for exercise behavior: A critique of theoretical directions. In G. C.
Roberts (Ed.), Motivation in sport and exercise (pp. 129–158). Champaign, IL: Human Kinetics.
Rejeski, W. J., Gauvin, L., Hobson, M. L., & Norris, J. L. (1995). Effects of baseline responses, in-task
feelings, and duration of activity on exercise-induced feeling states in women. Health Psychology,
14, 291–300.
Rejeski, W. J., Reboussin, B. A., Dunn, A. L., King, A. C., & Sallis, J. F. (1999). A modified exercise-in-
duced feeling inventory for chronic training and baseline profiles of participants in the Activity
Counseling Trial. Journal of Health Psychology, 4, 97–108.
Russell, J. A. (1980). A circumplex model of affect. Journal of Personality and Social Psychology, 39,
1161–1178.
Schlosberg, H. (1952). The description of facial expressions in terms of two dimensions. Journal of Ex-
perimental Psychology, 44, 229–237.
Spielberger, C. D., Gorsuch, R. L., Luschene, R., Vagg, P. R., & Jacobs, G. A. (1983). Manual for the
State–Trait Anxiety Inventory. Palo Alto, CA: Consulting Psychologists.
Statistical Analysis Systems Institute. (1990). SAS/STAT user’s guide (Vol. 2). Cary, NC: Author.
Thayer, R. E. (1989). The biopsychology of mood and arousal. New York: Oxford University Press.
Treasure, D. C., & Newbery, D. M. (1998). Relationship between self-efficacy, exercise intensity, and
feeling states in a sedentary population during and following an acute bout of exercise. Journal of
Sport and Exercise Psychology, 20, 1–11.
Vlachopoulos, S., Biddle, S., & Fox, K. (1996). A social-cognitive investigation into the mechanisms of
affect generation in children’s physical activity. Journal of Sport and Exercise Psychology, 18,
174–193.
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of posi-
tive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54,
1063–1070.
Watson, D. A., & Tellegen, A. (1985). Toward a consensual structure of mood. Psychological Bulletin,
98, 219–235.
Weiner, B. (1979). A theory of motivation for some classroom experience. Journal of Educational Psy-
chology, 71, 3–25.
Weiner, B. (1986). An attributional theory of motivation and emotion. New York: Springer-Verlag.

PHYSICAL ACTIVITY AFFECT SCALE
95
APPENDIX
Instructions: Please use the following scale to indicate the extent to which each word below
describes how you feel at this moment in time. Record your responses by circling the appropriate
number.
Do Not
Feel
Feel
Slightly
Feel
Moderately
Feel
Strongly
Feel Very
Strongly
1. Upbeat
2. Calm
3. Energetic
4. Tired
5. Peaceful
6. Miserable
7. Worn-out
8. Relaxed
9. Fatigued
10. Discouraged
11. Enthusiastic
12. Crummy
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
4
4
Note. Subscales and corresponding items are as follows: Positive Affect (1, 3, 11); Negative Affect
(6, 10, 12); Fatigue (4, 7, 9); Tranquility (2, 5, 8).