Psychology homework on Social Learning Theory
Journal of Physical Activity and Health, 2013, 10, 544-555 © 2013 Human Kinetics, Inc. Official Journal of ISPAH
www.JPAH-Journal.com ORIGINAL RESEARCH
544
Dewar, Morgan, and Plotnikoff are with the School of Educa- tion, University of Newcastle, Newcastle, New South Wales, Australia. Lubans is with the School of Education and Arts, Uni- versity of Newcastle, Newcastle, New South Wales, Australia.
Development and Evaluation of Social Cognitive Measures Related to Adolescent Physical Activity
Deborah L. Dewar, David Revalds Lubans, Philip James Morgan, and Ronald C. Plotnikoff
Background: This study aimed to develop and evaluate the construct validity and reliability of modernized social cognitive measures relating to physical activity behaviors in adolescents. Methods: An instrument was developed based on constructs from Bandura’s Social Cognitive Theory and included the following scales: self-efficacy, situation (perceived physical environment), social support, behavioral strategies, and outcome expectations and expectancies. The questionnaire was administered in a sample of 171 adolescents (age = 13.6 ± 1.2 years, females = 61%). Confirmatory factor analysis was employed to examine model-fit for each scale using multiple indices, including chi-square index, comparative-fit index (CFI), goodness-of-fit index (GFI), and the root mean square error of approximation (RMSEA). Reliability properties were also examined (ICC and Cronbach’s alpha). Results: Each scale represented a statistically sound measure: fit indices indicated each model to be an adequate-to-exact fit to the data; internal consistency was acceptable to good (α = 0.63–0.79); rank order repeatability was strong (ICC = 0.82–0.91). Conclusions: Results support the validity and reli- ability of social cognitive scales relating to physical activity among adolescents. As such, the developed scales have utility for the identification of potential social cognitive correlates of youth physical activity, mediators of physical activity behavior changes and the testing of theoretical models based on Social Cognitive Theory.
Keywords: psychosocial measures, scales, youth, physical activity behavior
The health benefits of physical activity (PA) for chil- dren and adolescents are well documented.1 Participation in PA contributes to improved musculoskeletal health and the potential for reducing the risk for chronic disease such as Type 2 diabetes, cardiovascular disease, obesity, and certain cancers.2–5 Furthermore there is evidence for related psychosocial benefits including improvements in self esteem and self concept6,7 and a reduction in anxiety and depressive symptoms.8,9 Yet, unfortunately there is evidence indicating many adolescents are not sufficiently active and fail to meet PA guidelines.10,11
Childhood and adolescence have been identified as critical periods for the establishment of health behaviors that are likely to track into adulthood.12 Hence, interven- tions targeting the PA behaviors of youth are important. However, many interventions directed at this population have been ineffectual or produced only modest outcomes for PA behavior change.13,14 Explanations for these findings may be a result of a number of methodological limitations in these studies, including atheoretical inter- ventions, underpowered studies, weak assessment mea- sures, inadequate intervention duration or intensity, poor
program compliance or exposure, and a lack of tailored interventions that exclusively target priority groups.14–16
The design and development of ineffective programs may also be due to inadequate understanding of the causal mechanisms of PA behavior change specific to children and adolescents.17 Theories of behavior change (eg, Social Cognitive Theory, Theory of Planned Behavior) postulate that there are underlying “mechanisms” through which intervention effects are achieved.18 By examining the role hypothesized variables (eg, intentions or social support) may have in mediating the pathway between an intervention and observed behavioral outcomes,19 researchers can establish which components of an intervention contributed wholly or partially to behavior change. Despite this knowledge, few studies have exam- ined hypothesized mediators of PA intervention effects among children and adolescents.20
Furthermore poor quality measures used to assess potential mediators has compounded the limited research and evidence for mediators of youth PA. In their recent review, Brown and Colleagues21 revealed the psychomet- ric properties reported for reliability and validity by many measures for children and adolescents was substandard or limited implicating the potential for inaccurate conclu- sions regarding effective mediators and the efficacy of PA interventions. Moreover, it was suggested that instrument precision is problematic when modified versions of adult measures are employed in child and adolescent studies without prior testing.
Social Cognitive Measures Physical Activity 545
It is clear a stronger evidence base is needed for mediators of PA behavior in developing an improved understanding of “what works” among youth to inform the design of more efficacious interventions. The use of valid and reliable measures that can lead to accurate conclusions regarding effective mediators is an essential component of this process. Hence, the aim of this study is to develop a novel, valid, and reliable questionnaire that assesses social cognitive measures relating to adolescent PA behaviors; has utility for population-based research in adolescents; and is suitable for use in adolescent obser- vational and intervention studies to examine potential correlates and mediators of PA behavior. Population spe- cific measures that are current and contextually suitable for the intended audience have important implications for researchers. This is because instrument sensitivity is likely to be improved if measures and their items are deemed appropriate for the particular demographic being investigated.22
Methods
Development of Scales and Items
During the initial development of the scales a series of qualitative processes were employed.23 First, a review of the literature reporting the properties of existing social cognitive measures relating to adolescent PA was con- ducted. A preliminary instrument was then developed, comprising of 5 scales designed to be unidimensional measures of the following constructs from Bandura’s Social Cognitive Theory (SCT):24 perceived self-efficacy, situation (perceptions of the physical environment), behavioral strategies (self-control), social support and outcome expectations (perceived benefits) and expectan- cies (value placed on benefits) relating to PA. The SCT purports that behavior change is influenced by a complex interaction between personal and environmental factors, and attributes of the behavior itself.18 While many other models of health behavior are limited to predicting health behaviors, a strength of SCT lies within the provision of predictors and principles that lead to informing, guid- ing, enabling and motivating individuals to modify their behaviors in promoting good health.25 This may help to explain why SCT has emerged as a prominent health behavior model in guiding the development of interven- tions and examination of mechanisms of health behavior change in children and adolescents.20,26,27
In developing the measures, an objective was to include items that address the significant role that tech- nology plays in the lives of contemporary adolescents.28 As such, several items refer to various modern technolo- gies (eg, personal music devices, mobile phones, and pedometers), that may be used to support participation in PA and that research has shown are often accessible to and routinely used by many adolescents today.28,29 The relevance of such technology to youth PA has become more apparent in PA research in the past decade. Not only has there been increase in the number of PA interventions that have adopted the use of modern technological devices
such as pedometers and mobile phones to encourage or motivate personal monitoring of PA, but there is grow- ing evidence to support the success of these strategies in promoting PA.30–32
Three experts in PA, SCT, and scale development were consulted to review the measures and determine content validity. Specifically the experts were asked to a) examine how well each item contributed to the theo- retical conceptualization of each construct, b) examine how well response options supported respective items, and c) evaluate item comprehension and the potential for participant burden.33
To further review and refine the scales, a focus group was conducted with 12 secondary school students (age 14.1 ± 0.6 years; females = 58%) in Spring 2009. Partici- pants were randomly selected from a group of consenting students in Grades 8 and 9 from an independent (non- government) school. A semistructured interview setting was adopted where probing facilitated the examination of thought processes used in arriving at an answer and interpreting instruction sets and response options. The focus group was digitally recorded and transcribed.
To increase scale sensitivity, the number of Likert- type response options employed by each scale used no fewer than 4 response options.34 No neutral/uncertain response category was provided for any scale on the basis that this may lower questionnaire reliability through reducing variability.34 As such, scales ensured the provi- sion of weak response categories (eg, slightly disagree/ agree) in attracting students who would otherwise prefer a neutral option.
Self-Efficacy. Self-efficacy was operationalized as an individual’s confidence in personal ability to adopt and maintain PA behaviors and overcome barriers to PA. Eight items were measured on a 6-point Likert-type scale of 1 (strongly disagree) to 6 (strongly agree); for example, “I find it difficult to be physically active when I have no one to be active with.” The scale combined original and modified items from previous scales devel- oped for older children35 and adolescents.36 For example, Motl’s36 earlier self efficacy measure included the item “I can be physically active during my free time on most days no matter how busy my day is,” which was abridged to read “I can still be physically active even when I’ve had a busy day.”
Situation. Eight items assessed an individual’s mental representation of their physical home/neighborhood and school environments which may influence their PA behaviors. Specifically, items examined how neighbor- hood safety and accessibility to facilities and equip- ment at home and school impact PA; for example, “It is difficult to be physically active in my neighborhood because of lots of traffic.” A 6-point Likert-type scale again examined the respondents’ level of agreement/dis- agreement with each item. Original items were merged with modified items from an earlier measure intended for older children.37
Social Support. Social support was operationalized as various supportive behaviors received from friends
546 Dewar et al
and family in the previous 3 months that encouraged participation in PA; for example, “. . . did members of your family take you to places where you could be physi- cally active (for example, to the beach, sports training, or weekend sport?)”. Twelve items examined the frequency of supportive behaviors received using a 5-point Likert- type scale (1 = never, to 5 = always). Some items were modified versions from a previous scale that examined parent-reported correlates of child and adolescent PA.38
Behavioral Strategies. Eight items examined self- regulation strategies used to reinforce participation in PA, including methods used to enhance enjoyment, set goals and self-monitor PA behaviors. Two modified items from Dishman’s39 earlier measure intended for older children and adolescents were included. A 5-point Likert-type scale assessed the frequency (1 = never, to 5 = always) at which various self-regulation strategies were employed during the previous 3 months; for example, “Did you keep track of how much physical activity you did—for example, using a pedometer, timer on your phone or by keeping a log book?”
Outcome Expectations and Expectancies. Outcome expectations were operationalized as anticipated physi- cal, social, and emotional benefits of being physically active. Eight items were rated on a 6-point Likert-type scale (1 = strongly disagree, to 6 = strongly agree); for example, “Participation in regular physical activity can help me to manage stress better.” Some expectation items were modified versions sourced from previous PA enjoy- ment and attitude scales developed for children40 and adolescents.41 Five outcome expectancy items provided a corresponding personal evaluation of the benefit identi- fied by each outcome expectation item. Items were rated on a 4-point Likert-type scale (1 = not at all important, to 4 = very important); for example, “How important is managing stress to you”?
Questionnaire Administration
Following approval from the University Research Ethics Committee, consent was obtained from the Principals of 3 nongovernment schools from the Newcastle/Central Coast region of New South Wales for their school’s involvement in the questionnaire’s testing. Consenting secondary school-aged students from predominantly middle-class backgrounds were recruited from these schools to complete a 2-week test-retest in Autumn 2010.
Data Analyses
Descriptive statistics were obtained for all variables [Means (M), standard deviations (SD) and frequencies (f)] using SPSS 17.0. Since the percentage of missing data were very small (0.02%), mean substitution was the preferred imputation method used to manage incomplete data rather than exclusion methods.42
Reliability. Reliability analyses were conducted using SPSS 17.0. To provide a coefficient of individual repeat- ability, the 95% limits of agreement were calculated.43
Scores for the intertrial difference (T2 – T1) were plot- ted against the intertrial mean [(T1 + T2)/2] for each individual, after which the range of differences falling within the mean of the intertrial differences ± 1.96 stan- dard deviations was calculated43,44. Bivariate correlations between the intertrial difference and the intertrial mean were also assessed to establish if the limits of agreement were consistent throughout the range of measurements. Cronbach’s alpha coefficient was calculated from T1 (baseline) data to estimate the internal consistency of each scale. Values > 0.6 are considered reliable.45 Finally, intraclass correlation coefficients (ICCs) were calculated to provide a measure of rank order repeatability. For each scale, an ICC score ≥ 0.75 indicates excellent reliability.46
Factor Analysis. Confirmatory factor analysis (CFA) using AMOS 17.0 was used to directly test model fit for each of the scales. The chi-square tests for statistically significant difference between the covariance matrix of the hypothesized model and the observed population variables.47 While a nonsignificant chi-square result (P > .05) indicates the model being examined is a good fit, it is often too sensitive to sample size and a rejection of the hypothesized model likely results.48 For this reason, additional measures should be used to examine model fit. Hence the following model-fit indices were calcu- lated from baseline (T1) data: chi-square index, the root mean error of approximation (RMSEA), goodness-of-fit index (GFI), adjusted goodness-of-fit index (AGFI) and the comparative fit index (CFI). In interpreting GFI, AGFI and CFI scores; values ≥ 0.9, ≥ 0.95, or equal to 1 indicate adequate, good, and exact fit of the model, respectively.49 The RMSEA is widely regarded as a principal index in examining model fit,50 where scores ≤ 0.08, ≤ 0.06, and 0.0, signify acceptable, close, and exact fits, respectively.49 If data showed multivariate nonnormality (multivariate kurtosis represented by a Mardia’s coefficient > 3),51 the Bollen-Stine bootstrap procedure was employed to examine model fit and bias corrected regression coefficients are reported.52 CFA was also used to examine factor loadings for each item on its latent construct in determining scale homogeneity for each measure. Coefficients ≥ 0.45 are considered fair, while values ≥ 0.55 and ≥ 0.71 indicate a factor loading to be good and excellent respectively.53
Results
Descriptive Statistics
The study sample included 171 secondary school students (age = 13.6 ± 1.2; 61% female), comprising 80.1% Australian, 9.9% European, 3.5% Asian, 1.8% Middle Eastern, 1.2% African, and 3.5% other. A 1-way analysis of variance (ANOVA) revealed there were no statistically significant intertrial gender differences for any of the scales. Hence separate analyses by gender were not carried out. Scale means and standard devia- tions are presented in Table 1. Means ranged from 2.7 (±1.3) to 5.5 (±0.7) on the 6-point Likert-type scale for
Social Cognitive Measures Physical Activity 547
the self-efficacy, situation and outcome expectations measures; 2.2 (±1.2) to 4.0 (±1.0) on the 5-point Likert- type scale for the social support and behavioral strategies measures; and 3.1 (±0.9) to 3.4 (±0.7) on the 4-point Likert-type scale for the outcome expectancy measure.
Confirmatory Factor Analysis Self-Efficacy. Preliminary analyses revealed the original 8-item self-efficacy scale to show inadequate model-fit and requiring further refinement. An iterative process involving the removal of 1 item at a time found 3 items represented unacceptable factor loadings on the model, contributed poorly to model-fit indices and were considered redundant by other similarly worded items seeking the same information. Subsequently, Table 2 shows the final composite resulted in a 5-item 1-factor model where fit indices demonstrated good-to-exact fit, and factor loadings for items ranged from fair (0.45) to excellent (0.70).
Situation. Initially a 1-factor model resulted in a poor fit to the 8-item situation questionnaire. Further confirmatory analyses revealed a more robust scale was established when treated as a 2-factor model compris- ing of a home/neighborhood environment factor and school environment factor. Following the removal of 2 items which loaded poorly on the home/neighborhood structure, fit indices significantly improved for the final 2-factor model which demonstrated good fit and com- prised item loadings that ranged from fair to excellent for the home/neighborhood (0.49–0.72) and school (0.52–0.73) factors respectively (Table 2).
Social Support. The original 12-item social sup- port scale demonstrated poor model fit when treated as a 1-factor model. Analyses supported a 2-factor structure as items were categorized to either a friend or family factor, indicating from whom social support for being physically active was received. Following the removal of 2 items from each of the family and friend support subscales, the final 2-factor model showed an improved and parsimonious fit represented by adequate- to-good fit indices and item loadings that ranged from good to excellent for the friend (0.57–0.71) and family (0.62–0.73) support factors, respectively (Table 2). The final measures comprised 4 friend support items and 4 family support items.
Behavioral Strategies. The original 8 items loaded adequately on the 1-factor model, however, some fit indices proved less than satisfactory. Two items which contributed poorly to the scale’s psychometric properties and showed extreme kurtosis were removed. The result- ing scale showed improved model fit, demonstrating adequate-to-good fit indices (Table 2) and good factor loadings which ranged from 0.55–0.70.
Outcome Expectations and Expectancies . Pre- liminary analyses showed model-fit for a paired 8-item expectations and expectancies measure did not satisfy all criteria. The removal of 3 expectations items, which loaded poorly on the expectations structure, resulted in
a refined 5-item scale which satisfied most model-fit criteria and comprised good (0.50) to excellent (0.79) factor loadings (Table 2). The removal of the 3 cor- responding expectancy items also improved model fit indices and factor loadings (ranging from 0.29–0.79) for the expectancy scale. Yet, poor values for the RMSEA and some factor loadings persisted, suggesting fur- ther refinement was needed. Additional confirmatory analyses revealed the removal of 1 pair of items, which provided a noninterpretable factor loading (< 0.30) on the expectancy scale, did improve and satisfy all model- fit indices and factor loadings. However a decision was made to retain the corresponding items because of the content representativeness value.
Reliability Analysis
Reliability results for the final questionnaires follow- ing item reduction are shown in Table 3. Bland-Altman analyses revealed narrow limits of agreement for each of the scales. Nonsignificant bivariate correlations between the intertrial difference and intertrial mean indicated the limits of agreement were consistent throughout the range of measures for all scales, except one (home/neighbor- hood situation scale). Intraclass correlation coefficients (ICCs) indicated very good rank-order repeatability, ranging from 0.82 for outcome expectations to 0.91 for the self-efficacy, family social support and behavioral strategies scales. Meanwhile, internal consistency coef- ficients were at least acceptable and ranged from 0.63 for the home/neighborhood situation subscale to 0.79 for the behavioral strategies scale.
Discussion There is strong support for the influence of social cog- nitive factors on the PA behavior of children and ado- lescents.20,38,54 However, quality measures with strong psychometric properties are needed to improve our understanding of PA in these populations. The current study describes the development and evaluation of new scales for assessing social cognitive measures related to adolescent PA. While all scales demonstrated acceptable reliability, CFA was able to establish acceptable construct validity in supporting the scales utility for identifying potential correlates and mediators of adolescent PA.
Few comprehensive questionnaires that include several social cognitive measures have been devel- oped and evaluated in adolescent populations. While a multitude of instruments for assessing social cognitive influences of PA exist, many scales comprise a large number of items.35,37,55,56 This may be problematic for the researcher(s) when more than 1 measure is of interest. For instance, lengthy questionnaires can be burdensome on respondents, which in turn may negatively impact instrument sensitivity and hence data accuracy.34 As such, improved and more parsimonious social cognitive measures of youth PA are needed.
A unique feature of these scales is the inclusion of novel items that assess modern technology’s potential
548
Ta b
le 1
S
c a le
M e a n
s , S
ta n
d a rd
D e v ia
ti o
n s (
S D
), a
n d
I te
m K
u rt
o s is
V a lu
e s
C o
n st
ru ct
s an
d d
es cr
ip ti
o n
s
S o
u rc
e o
f m
o d
ifi ed
it
em s
R an
g e
(N o
. o f
it em
s)
T 1
(b as
el in
e) T
2 (2
-w ee
k re
te st
)
M ea
n ±
S D
It em
K u
rt o
si s
M ea
n ±
S D
It em
K u
rt o
si s
S el
f ef
fi ca
cy :
P ar
ti ci
pa nt
s w
er e
as ke
d to
r at
e th
ei r
co nfi
de nc
e in
t he
ir a
bi li
ty t
o ad
op t
an d
ov er
co m
e ba
rr ie
rs t
o PA
b eh
av io
rs
S ca
le :
1 =
d is
ag re
e a
lo t:
6 =
a gr
ee a
l ot
E xa
m p
le :
“W he
n I’
m p
hy si
ca ll
y ac
ti ve
( eg
, d ur
in g
P E
o r
sc ho
ol s
po rt
) I
ge t
em ba
rr as
se d
ab ou
t m
y fi
tn es
s or
s ki
ll l
ev el
.”
[3 4,
3 5]
1– 6
(5 )
4. 2
± 1
.0 –1
.0 9
to 0
.1 1
4. 1
± 1
.0 –1
.0 8
to 0
.1 2
S it
u at
io n
a : P
ar ti
ci pa
nt s
w er
e as
ke d
to r
es po
nd t
o st
at em
en ts
a bo
ut t
he ir
m en
ta l
re pr
es en
ta -
ti on
o f
th e
ho m
e/ ne
ig hb
or ho
od a
nd s
ch oo
l ph
ys ic
al e
nv ir
on m
en t
th at
m ay
i nfl
ue nc
e th
ei r
PA
be ha
vi or
.
S ca
le :
1 =
d is
ag re
e a
lo t:
6 =
a gr
ee a
l ot
a)
H om
e/ n
ei gh
b or
h oo
d e
xa m
p le
: “A
t ho
m e
I ha
ve a
cc es
s to
e qu
ip m
en t
th at
h el
ps m
e
to
b e
ph ys
ic al
ly a
ct iv
e (e
g, b
ik es
, b al
ls , s
ka te
bo ar
ds ).
” [3
6] 1–
6 (3
) 5.
0 ±
0 .9
–0 .2
0 to
6 .1
9 5.
0 ±
0 .8
–0 .5
7 to
1 .7
6
b
) S
ch oo
l ex
am p
le :
“A t
sc ho
ol , f
ac il
it ie
s ar
e av
ai la
bl e
du ri
ng r
ec es
s/ lu
nc h
fo r
m e
to b
e
ph
ys ic
al ly
a ct
iv e
(e g,
t he
g ym
, d an
ce s
tu di
o, s
po rt
s eq
ui pm
en t)
.” 1–
6 (3
) 3.
8 ±
0 .9
3. 7
± 1
.0
B eh
av io
ra l
st ra
te gi
es :
P ar
ti ci
pa nt
s w
er e
as ke
d to
r at
e th
e fr
eq ue
nc y
w it
h w
hi ch
t he
y re
in -
fo rc
ed t
he ir
o w
n PA
b eh
av io
rs t
hr ou
gh s
et ti
ng g
oa ls
, s el
f- m
on it
or in
g an
d st
ra te
gi es
f or
en
ha nc
in g
en jo
ym en
t, s
ta rt
in g
w it
h th
e co
m m
on s
te m
“ In
t he
p as
t 3
m on
th s
ho w
o ft
en .
. . .”
S ca
le :
1 =
n ev
er :
5 =
a lw
ay s
E xa
m p
le :
“. .
. d id
y ou
p ar
ti ci
pa te
i n
a va
ri et
y of
p hy
si ca
l ac
ti vi
ti es
t o
av oi
d bo
re do
m ?”
[3 8]
1– 5
(6 )
3. 2
± 0
.8 –0
.7 8
to –
0. 19
3. 2
± 0
.8 –0
.6 3
to 0
.5 5
S oc
ia l
su p
p or
ta :
P ar
ti ci
pa nt
s w
er e
as ke
d to
r at
e th
e fr
eq ue
nc y
w it
h w
hi ch
f ri
en ds
a nd
f am
il y
re in
fo rc
ed P
A t
hr ou
gh e
nc ou
ra ge
m en
t, r
ol e
m od
el in
g, a
nd t
he p
ro vi
si on
o f
PA o
pp or
tu ni
ti es
, st
ar ti
ng w
it h
th e
co m
m on
s te
m “
In t
he p
as t
3 m
on th
s ho
w o
ft en
. . .
.”
S ca
le :
1 =
n ev
er :
5 =
a lw
ay s
a)
F ri
en d
s u
p p
or t
ex am
p le
: “.
. . d
id y
ou m
ak e
pl an
s w
it h
yo ur
f ri
en ds
t o
be p
hy si
ca ll
y
ac
ti ve
t og
et he
r? ”
[3 7]
1– 5
(4 )
3. 1
± 0
.9 –0
.8 9
to 0
.8 1
3. 2
± 0
.8 –0
.8 0
to 0
.7 3
b
) F
am il
y su
p p
or t
ex am
p le
: “.
. . d
id y
ou r
pa re
nt s
bu y
yo u
eq ui
pm en
t th
at e
nc ou
ra ge
d
yo
u to
b e
ph ys
ic al
ly a
ct iv
e (e
g, s
po rt
s cl
ot he
s, j
og ge
rs , b
ik e
. . .
.” 1–
5 (4
) 3.
8 ±
0 .9
3. 9
± 0
.8
(c on
ti nu
ed )
549
Ta b
le 1
( c o
n ti
n u
e d
)
C o
n st
ru ct
s an
d d
es cr
ip ti
o n
s
S o
u rc
e o
f m
o d
ifi ed
it
em s
R an
g e
(N o
. o f
it em
s)
T 1
(b as
el in
e) T
2 (2
-w ee
k re
te st
)
M ea
n ±
S D
It em
K u
rt o
si s
M ea
n ±
S D
It em
K u
rt o
si s
O u
tc om
e ex
p ec
ta ti
on s:
P ar
ti ci
pa nt
s w
er e
as ke
d to
r es
po nd
t o
st at
em en
ts a
bo ut
v ar
io us
b en
- efi
ts o
f PA
S ca
le :
1 =
d is
ag re
e a
lo t:
6 =
a gr
ee a
l ot
E xa
m p
le :
“P ar
ti ci
pa ti
on i
n re
gu la
r ph
ys ic
al a
ct iv
it y
he lp
s to
i m
pr ov
e m
y fi
tn es
s. ”
[3 9,
4 0]
1– 6
(5 )
5. 2
± 0
.6 0.
10 t
o 5.
85 5.
3 ±
0 .5
–0 .6
7 to
8 .1
8
O u
tc om
e ex
p ec
ta n
ci es
: P
ar ti
ci pa
nt s
w er
e as
ke d
to r
at e
pe rs
on al
v al
ue p
la ce
d on
e ac
h co
r- re
sp on
di ng
o ut
co m
e ex
pe ct
at io
n it
em f
or P
A
S ca
le :
1= n
ot a
t al
l im
po rt
an t:
4 =
e xt
re m
el y
im po
rt an
t
E xa
m p
le :
“H ow
i m
po rt
an t
is i
m pr
ov in
g yo
ur fi
tn es
s to
y ou
?”
1– 4
(5 )
3. 2
± 0
.5 –0
.7 4
to 0
.8 1
3. 3
± 0
.5 –0
.6 5
to 0
.5 4
a S
ca le
i s
pr es
en te
d as
a 2
-f ac
to r
m od
el .
A bb
re vi
at io
ns :
P E
, p hy
si ca
l ed
uc at
io n;
P A
, p hy
si ca
l ac
ti vi
ty .
550 Dewar et al
influence on the PA experiences of adolescents today. For example, in the social support scale, respondents were asked to consider equipment provided by parents that may encourage participation in PA and a reference to an ‘i-pod’ (iPod; personal music device) is provided in the prompts that follow the question: “. . . did your parents buy you equipment that encouraged you to be physically active? (eg, sport clothes, joggers, a bike, an i-pod for listening to music while being physically active)”. In a second example, the behavioral strategies scale includes an item that prompts respondents to consider if modern
technological devices may assist personal monitoring of PA: “. . . did you keep track of how much physical activity you did (eg, by using a pedometer, timer on your mobile phone . . .)?”. To the authors’ knowledge, no similar measures intended for adolescents have included prompts addressing the potential for modern technology to support participation in PA. While such a feature sup- ports a more contemporary set of measures, instrument sensitivity may also be improved especially when modern technological devices have become commonplace for many adolescents today.28,29
Table 2 Validity Results of the Confirmatory Factor Analysis Showing Model Fit and Factor Loadings From Baseline Data
Constructs χ2 P RMSEA GFI AGFI CFI Factor
loadings
Self efficacy 3.82 0.58 0.00 0.99 0.97 1.00 0.45–0.70
Situationa
Home/neighborhood 11.22 0.19 0.05 0.98 0.95 0.98 0.49–0.72
School 0.52–0.73
Behavioral strategies 15.45 0.16 0.07 0.97 0.93 0.97 0.55–0.70
Social supporta
Friend support 27.40 0.10 0.05 0.97 0.93 0.98 0.57–0.71
Family support 0.62–0.73
Outcome expectations 11.26 0.05 0.09 0.97 0.92 0.97 0.50–0.79
Outcome expectancies 15.74 0.01 0.11 0.97 0.90 0.91 0.29–0.79
a Scale is presented as a 2-factor model.
Abbreviations: χ2, chi-square; RMSEA, root mean square error of approximation; GFI, goodness of fit index; AGFI, adjusted goodness of fit index; CFI, comparative fit index.
Table 3 Reliability Results: Bivariate Correlations, 95% Limits of Agreement, Test-Retest Stability, and Internal Consistency
Constructs Rb 95% limits
of agreement ICC
(95% CI) Cronbach’s alphac
Self efficacy 0.03 –1.14 to 1.02 0.91 (0.88–0.93) 0.69
Situationa
Home/neighborhood –.18* –1.12 to 1.07 0.88 (0.83–0.91) 0.63
School 0.08 –1.45 to 1.29 0.85 (0.79–0.89) 0.65
Behavioral strategies –0.03 –0.81 to 0.97 0.91 (0.88–0.93) 0.79
Social supporta
Friend support –0.12 –1.12 to 1.13 0.86 (0.81–0.90) 0.74
Family support –0.12 –0.79 to 1.04 0.91 (0.88–0.94) 0.78
Outcome expectations –0.15 –0.77 to 0.95 0.82 (0.75–0.86) 0.75
Outcome expectancies –0.05 –0.54 to 0.72 0.88 (0.83–0.91) 0.66
a Scale is presented as a 2-factor model; b Bivariate correlations between the difference (T2 – T1) and the mean [(T1 + T2)/2]; c Cronbach’s alpha cal- culated from baseline (T1) data; *P < .05; 95% limits of agreement calculated as the intertrial mean difference ± 1.96 SDs (of the intertrial difference).
Abbreviations: ICC, intra class correlation; CIs, confidence intervals.
Social Cognitive Measures Physical Activity 551
In comparing psychometric properties of the pre- sented scales with those reported by similar measures, some challenges were observed. Firstly, while CFA was found to be a popular approach for examining the construct validity of other social cognitive measures, considerable variation in the model-fit indices reported have made comparisons between studies problematic. Clearly a set of universally agreed upon standards for examining and reporting model-fit need to be estab- lished. Regardless, CFA demonstrated each of the cur- rent measures to represent good-to-excellent construct validity as shown by acceptable model-fit indices and factor loadings.
Secondly, the author’s found the internal consistency reliability (Cronbach’s α) of earlier measures to be fre- quently provided, yet few to report ICC values to indicate scale stability. Rather, Pearson correlation coefficients are more commonly reported to inform consistency between test and retest scoress.21 Yet, their use to report stability of data has been regarded inappropriate and flawed because a relation between test and retest scores is merely pro- vided.57 In contrast, ICCs examine agreement, between scores within individuals, and so are considered a more suitable assessment of instrument stability.58
Self Efficacy Scale
Bandura25 has proposed that self-efficacy is the central determinant of SCT because it influences health behavior both directly and indirectly through its effect on the other behavioral determinants. Beyond this premise, there is strong empirical support for self-efficacy as a correlate of child and adolescent PA59,60 and more recently evidence for self efficacy as a potential mediator of PA behavior change in adolescent intervention studies has developed.61 Our findings support a valid and reliable measure of self- efficacy. The 5-item single factor structure represents a more parsimonious measure than previous self-efficacy scales which have comprised up to 17 items and 3 factors.35,36,40,55
With regards to CFA, where direct comparisons can be made through reporting of common fit-indices (RMSEA and CFI), our findings have improved upon results of several earlier self-efficacy measures.36,39,55 A further strength of the present self efficacy scale has seen all items load adequately on the 1-factor model. While the authors found few previous self-efficacy measures to report item factor loadings, earlier measures36 have reported items that have loaded inadequately (< 0.45)53 on the hypothesized model without providing argumen- tative support for doing so. It has been suggested that retaining items that load poorly on a latent construct can compromise questionnaire homogeneity.53
Situation Scale
While individual correlates of PA (eg, self efficacy, enjoy- ment, and intentions) have been reported to account for up to 40% of the predicted variance in PA behavior only,62 there has been increasing interest in the role various
aspects of the environment (eg, physical, social, and cul- tural) may play in facilitating or impeding PA behavior.37 While the environment is a key construct within SCT hypothesized to influence individual behavior change, there is strong support for various aspects of the environ- ment, including the physical (eg, access to facilities and opportunities that promote PA) to correlate with the PA behaviors of children and adolescents.63,64
While few previous measures assessing barriers and facilitators of adolescent PA have been place-specific,65 the present situation measure examined the perceived physical features of specific environments (home/neigh- borhood and school) that may promote or impede oppor- tunities for PA. While local neighborhoods and parks have been highlighted in the literature as key locations used by adolescents for PA,66 important associations between the school environment and student PA levels have also been established.67
Results indicated the 2-factor situation scale to dem- onstrate favorable construct validity and reliability. While several earlier measures of the physical environment and its relation to PA do exist, variability in content does make comparisons of psychometric results perhaps futile. For example, where earlier questionnaires examine PA facili- tators or impediments of the neighborhood environment only68,69 the current single-factor home/neighborhood situation scale assesses both the home and neighborhood environments. Similarly, different information is sourced from Robertson-Wilson’s37 earlier measure of the school environment which extends the current single-factor sub- scale assessing the school environment by investigating physical features of physical education classes, school and intramural sport opportunities. Even so, the present measure may offer researchers a more parsimonious scale that still is capable of examining 3 physical environments within a concise 2-factor structure.
Social Support Scale Social support for PA is another environmental variable that has received widespread attention for its potential influence on PA behaviors. While there is good evidence for parent and peer support to correlate with the PA behaviors of children and adolescents,63,64 the impor- tant influence family support may provide has begun to materialize with recent reviews revealing the most efficacious school-based PA programs have integrated a familial component.13
The current social support scale comprised a 2-factor model assessing friend and family support for PA. Where common fit indices are reported (RMSEA and CFI), present model-fit results are analogous with previous results from a similar social support measure.39 Although stability reliability for the family (ICC = 0.91) and friend support (ICC = 0.86) subscales were higher than coefficients reported by Norman and colleagues68 for an earlier family support (ICC = 0.74) and peer sup- port (ICC = 0.68) measure, a comparison of internal consistency revealed marginal difference between the respective measures.
552 Dewar et al
Behavioral Strategies Scale
The behavioral strategies scale was found to be reli- able measure with good construct validity. Although personal regulation of behavior through strategy use is hypothesized to be a primary mechanism for behavior change in several theories of health behavior, there is little empirical evidence available to support such an assump- tion. Specifically, very few studies have examined the role self-management strategies may play as a potential mediating variable of PA behavior change in youth inter- ventions20,21 and so strong conclusions cannot be formed. More research is needed in this area, and valid and reliable measures examining self-management strategies for PA are necessary to facilitate this research.
Although the present measure demonstrated sound construct validity, common model fit indices (RMSEA and CFI) were marginally inferior to earlier reports for self-management scales that assessed behavioral and cognitive strategies among adolescent girls.39,70 Both prior studies evaluated scales that were adaptations of a measure initially developed for adults.56 The internal consistency reliabilities of our scale compared both favor- ably39 and less favorably68 to earlier measures. Although the authors found very few comparable measures to report ICC values, the present measure (ICC = 0.91) did represent stronger instrument stability than Norman’s68 scale (ICC = 0.75). However, while a comparison of psychometric properties have been made, it should be noted that content varies between established measures of self-management strategies for PA. For example Norman’s68 single-factor scale examined cognitive and behavioral strategies, and was specifically developed to reflect content of an intended intervention.
Outcome Expectations and Expectancies Scale
Refinement following preliminary analyses resulted in a reduced 5-item outcome expectations questionnaire which assessed perceived physical, social and psy- chological benefits of PA, and included corresponding expectancy items examining personal evaluations of each benefit. A primary consideration was to develop a questionnaire that addressed PA benefits relevant to adolescents. Contento and colleagues71 suggest knowl- edge of long-term health outcomes do little to motivate adolescent food choices because the ramifications may be perceived as remote and inconsequential. While the same may be true for adolescent motivations that drive participation in PA, the current scale focused more on potential immediate or short-term benefits such as fit- ness, enjoyment, and socialization rather than potential long-term health implications.
Although most fit indices for the expectations/expec- tancy subscales were adequate to good, weak RMSEA values (> 0.08) suggest further scale refinement may contribute to a more robust model. In particular, 1 pair of items proved problematic. Although the expectation
statement (“Participation in regular physical activity can help me to control my weight better”) loaded adequately on its factor structure (0.55), this was not true for the corresponding expectancy statement (“How important is controlling your weight to you?”), which loaded poorly (0.29) on its respective structure. However, a decision was made to retain the paired items, arguing that the content was particularly relevant to the construct being measured.
Although reliability results demonstrated adequate- to-good internal consistency for the expectancy and expectations subscales respectively, values were poorer than those reported by Dishman and colleagues39,70 for earlier expectancy measures, yet improved upon values reported in a validation sample40 for a similar scale. Further, the current questionnaire is a more parsimoni- ous and coherent single factor measure than Saunder’s40 2-factor 16-item belief scale. Although ICC values are not reported by these previous studies to allow a comparison of scale stability, the present expectation and expectancy measures demonstrated a high degree of test-retest reli- ability (ICC = 0.82 and 0.88, respectively).
Implications
A developing body of literature has highlighted the potential influence of social cognitive factors such as self-efficacy, social support, and behavioral strategies on adolescent PA.20,38,54 However, it has been suggested that the measurement of such influences has been problem- atic and may have contributed to inaccurate conclusions regarding correlates or the most effective mediators of PA.21 More specifically, it’s important that measures used by researchers for this purpose exhibit acceptable reliability and validity.
The current study has demonstrated evidence for the construct validity and reliability of 5 scales designed to measure SCT constructs related to PA in adolescents. The authors believe this questionnaire provides a contempo- rary and parsimonious solution for researchers interested in more than 1 social cognitive measure relating to PA behavior in this population.
Despite the strengths of this study, there are some limitations that should be noted. The tests of validity used in the current study were not extensive. Future researchers are encouraged to test the concurrent and convergent validity of these scales by comparing them with similar validated measures and actual PA behavior. Furthermore, although the study’s sample size is com- parable to many other validation studies, it may present as a limitation for additional factor analytical techniques that could be carried out; for example, 1) cross validation of the measurement models by employing a multigroup analysis (eg, between different races and populations) of factorial invariance, and 2) testing for longitudinal factorial invariance of the measurement models across time. Focus group participants were adolescents from 1 low-fee paying independent secondary school and therefore might not be representative of a diverse popu- lation of adolescents. Finally, although the racial/ethnic
Social Cognitive Measures Physical Activity 553
demographics of the study sample were fairly well rep- resentative of Australia,72,73 the sample nevertheless was relatively homogenous; additional testing of the measures in multiethnic populations is advised.
Conclusions The results of this study provide support for the construct validity and reliability of modernized social cognitive measures assessing: perceived self-efficacy, situation (including a home/neighborhood factor and school factor), behavioral strategies, social support (including a friend factor and family factor), and outcome expecta- tions and expectancies related to PA for use among an adolescent population. As such, these scales are suitable for the identification of potential social cognitive corre- lates of youth PA, mediators of PA behavior changes and the testing of theoretical models based on SCT.
Acknowledgments
The authors would like to thank each of the schools, teachers, and students who participated in this study. This work was sup- ported by an Australian Postgraduate Award (APA) scholarship to Deborah L. Dewar.
References 1. Strong WB, Malina RM, Blimkie CJ, et al. Evidence
based physical activity for school-age youth. J Pedi- atr. 2005;146(6):732–737. PubMed doi:10.1016/j. jpeds.2005.01.055
2. Lee IM. Physical activity and cancer prevention—data from epidemiologic studies. Med Sci Sports Exerc. 2003;35(11):1823–1827. PubMed doi:10.1249/01. MSS.0000093620.27893.23
3. Biddle S, Gorely T, Stensel D. Health-enhancing physical activity and sedentary behaviour in children and adoles- cents. J Sports Sci. 2004;22(8):679–701. PubMed doi:10 .1080/02640410410001712412
4. Eisenmann J, Welk G, Ihmels M, Dollman J. Fatness, fit- ness, and cardiovascular disease risk factors in children and adolescents. Med Sci Sports Exerc. 2007;39(8):1251– 1256. PubMed doi:10.1249/MSS.0b013e318064c8b0
5. Wang G, Dietz WH. Economic burden of obesity in youths aged 6 to 17 years: 1979-1999. Pediat- rics. 2002;109(5):E81–E1. PubMed doi:10.1542/ peds.109.5.e81
6. Walters S, Martin J. Does aerobic exercise really enhance self-esteem in children? A prospective evaluation in 3rd- 5th graders. J Sport Behav. 2000;23:51–60.
7. Asci FH, Kosar SN, Isler AK. The relationship of self- concept and perceived athletic competence to physical activity level and gender among Turkish early adolescents. Adolescence. 2001;36(143):499–507. PubMed
8. Motl R, Birnbaum A, Kubik M, Dishman R. Naturally occurring changes in physical activity are inversely related to depressive symptoms during early adolescence. Psycho- som Med. 2004;66(3):336–342. PubMed doi:10.1097/01. psy.0000126205.35683.0a
9. Strauss R, Rodzilsky D, Burack G, Colin M. Psychosocial correlates of physical activity in healthy children. Arch Pediatr Adolesc Med. 2001;155(8):897–902. PubMed
10. Australian Institute for Health and Welfare. Young Aus- tralians—their health and wellbeing. Canberra: AIHW; 2007.
11. Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, McDowell M. Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc. 2008;40(1):181–188. PubMed
12. Malina RM. Tracking of physical activity and physical fitness across the lifespan. Res Q Exerc Sport. 1996;67(3, Suppl):S48–S57. PubMed doi:10.1080/02701367.1996.1 0608853
13. van Sluijs E, McMinn A, Griffin S. Effectiveness of interventions to promote physical activity in children and adolescents: systematic review of controlled trials. Br J Sports Med. 2008;42(8):653–657. PubMed
14. Brown T, Summerbell C. Systematic review of school- based interventions that focus on changing dietary intake and physical activity levels to prevent childhood obesity: an update to the obesity guidance produced by the National Institute for Health and Clinical Excellence. Obes Rev. 2009;10(1):110–141. PubMed doi:10.1111/j.1467- 789X.2008.00515.x
15. Stone E, McKenzie T, Welk G, Booth M. Effects of physical activity interventions in youth. Review and synthesis. Am J Prev Med. 1998;15(4):298–315. PubMed doi:10.1016/S0749-3797(98)00082-8
16. Salmon J, Booth M, Phongsavan P, Murphy N, Timperio A. Promoting physical activity participation among chil- dren and adolescents. Epidemiol Rev. 2007;29:144–159. PubMed doi:10.1093/epirev/mxm010
17. Baranowski T, Jago R. Understanding the mechanisms of change in children’s physical activity programs. Exerc Sport Sci Rev. 2005;33(4):163–168. PubMed doi:10.1097/00003677-200510000-00003
18. Glanz K, Rimer B, Lewis F. Health behaviour and health education: theory, research and practice. 3rd ed. San Francisco, CA: Jossey-Bass; 2002.
19. Bauman A, Sallis J, Dzewaltowski D, Owen N. Toward a better understanding of the influences on physical activ- ity: the role of determinants, correlates, causal variables, mediators, moderators, and confounders. Am J Prev Med. 2002;23(2, Suppl):S5–S14. PubMed doi:10.1016/S0749- 3797(02)00469-5
20. Lubans D, Foster C, Biddle S. A review of media- tors of behavior in interventions to promote physi- cal activity among children and adolescents. Prev Med. 2008;47(5):463–470. PubMed doi:10.1016/j. ypmed.2008.07.011
21. Brown H, Hume C, Chin A, Paw M. Validity and reli- ability of instruments to assess potential mediators of children’s physical activity: a systematic review. J Sci Med Sport. 2009;12(5):539–548. PubMed doi:10.1016/j. jsams.2009.01.002
22. Fitzpatrick R, Davey C, Buxton M, Jones D. Evaluating patient-based outcome measures for use in clinical trials. Health Technol Assess. 1998;2:14. PubMed
23. Streiner D, Norman G. Health measurement scales: a practical guide to their development and use. 3rd ed. New York: Oxford Univeristy Press; 2003.
24. Bandura A. Social foundations of thought and action: a social cognitive theory. Engelwood Cliffs, N.J: Prentice- Hall; 1986.
554 Dewar et al
25. Bandura A. Health promotion by social cognitive means. Health Educ Behav. 2004;31(2):143–164. PubMed doi:10.1177/1090198104263660
26. Cerin E, Barnett A, Baranowski T. Testing theories of dietary behavior change in youth using the mediating variable model with intervention programs. J Nutr Educ Behav. 2009;41(5):309–318. PubMed doi:10.1016/j. jneb.2009.03.129
27. Sharma M. International school-based interventions for preventing obesity in children. Obes Rev. 2006;8:155–167. PubMed doi:10.1111/j.1467-789X.2006.00268.x
28. Roy Morgan Research. Young Australians Survey (14+). Melbourne: Roy Morgan Research; 2010.
29. Australian Communications & Media Authority. Media and communications in Australian homes. Canberra: ACMA; 2007.
30. Lubans D, Morgan P, Callister R, Collins C. Effects of integrating pedometers, parental materials and e-mail sup- port within an extracurricular school sport intervention. J Adolesc Health. 2009;44:176–183. PubMed doi:10.1016/j. jadohealth.2008.06.020
31. Lubans D, Morgan P, Tudor-Locke C. A systematic review of studies using pedometers to promote physical activity among youth. Prev Med. 2009;48(4):307–315. PubMed doi:10.1016/j.ypmed.2009.02.014
32. Newton KH, Wiltshire EJ, Elley CR. Pedometers and text messaging to increase physical activity: randomized controlled trial of adolescents with type 1 diabetes. Dia- betes Care. 2009;32(5):813–815. PubMed doi:10.2337/ dc08-1974
33. Grant J, Davis L. Selection and use of content experts f o r i n s t r u m e n t d eve l o p m e n t . R e s N u rs H e a l t h . 1997;20(3):269–274. PubMed doi:10.1002/(SICI)1098- 240X(199706)20:3<269::AID-NUR9>3.0.CO;2-G
34. Anderson L, Bourke S. Assessing affective characterisitcs in schools. 2nd ed. Mahwah, NJ: Lawrence Erlbaum Associates; 2000.
35. Jago R, Baranowski T, Watson K, et al. Development of new physical activity and sedentary behavior change self-efficacy questionnaires using item response model- ing. Int J Behav Nutr Phys Act. 2009;6:20. PubMed doi:10.1186/1479-5868-6-20
36. Motl R, Dishman R, Trost S, et al. Factorial validity and invariance of questionnaires measuring social-cognitive determinants of physical activity among adolescent girls. Prev Med. 2000;31(5):584–594. PubMed doi:10.1006/ pmed.2000.0735
37. Robertson-Wilson J, Levesque L, Holden R. Development of a questionnaire assessing school physical activity envi- ronment. Meas Phys Educ Exerc Sci. 2007;11(2):93–107. doi:10.1080/10913670701294088
38. Sallis J, Taylor W, Dowda M, Freedson P, Pate R. Correlates of vigorous physical activity for children in grades 1 through 12: comparing parent-reported and objectively measured physical activity. Pediatr Exerc Sci. 2002;14:30–44.
39. Dishman R, Hales D, Sallis J, et al. Validity of social- cognitive measures for physical activity in middle-school girls. J Pediatr Psychol. 2010;35(1):72–88. PubMed doi:10.1093/jpepsy/jsp031
40. Saunders R, Pate R, Felton G, et al. Development of ques- tionnaires to measure psychosocial influences on children’s
physical activity. Prev Med. 1997;26(2):241–247. PubMed doi:10.1006/pmed.1996.0134
41. Motl R, Dishman R, Saunders R, Dowda M, Felton G, Pate R. Measuring enjoyment of physical activity in adolescent girls. Am J Prev Med. 2001;21(2):110–117. PubMed doi:10.1016/S0749-3797(01)00326-9
42. Olinsky A, Chen S, Harlow L. The comparative efficacy of imputation methods for missing data in structural equation modeling. Eur J Oper Res. 2003;151:53–79. doi:10.1016/ S0377-2217(02)00578-7
43. Bland J, Altman D. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1(8476):307–310. PubMed doi:10.1016/ S0140-6736(86)90837-8
44. B l a n d J , A l t m a n D . M e a s u r e m e n t e r r o r. B M J . 1 9 9 6 ; 3 1 2 ( 7 0 4 7 ) : 1 6 5 4 . P u b M e d d o i : 1 0 . 1 1 3 6 / bmj.312.7047.1654
45. Sim J, Wright C. Research in health care: concepts, designs and methods. Cheltenham: Stanley Thornes Ltd.; 2000.
46. McDowell I. Measuring health: a guide to rating scales and questionnaires. New York: Oxford University Press; 2006.
47. Bollen K, Long J. Testing structural equation models. Newbury Park, CA: Sage; 1993.
48. Bollen K. Structural equations with latent variables. New York: John Wiley & Sons, Inc.; 1989.
49. Hu L, Bentler P. Cutoff criteria for fit indices in covari- ance structure analyses: conventional criteria versus new alternatives. Struct Equ Modeling. 1999;6:1–55. doi:10.1080/10705519909540118
50. Vanderberg R, Lance C. A review and synthesis of the mea- surement invariance literature: suggestions, practices, and recommendations for organizational research. Organ Res Methods. 2000;3:4–69. doi:10.1177/109442810031002
51. Mardia K. Measures of multivariate skewness and kur- tosis with application. Biometrika. 1970;57:519–530. doi:10.1093/biomet/57.3.519
52. Bollen K, Stine R. Bootstrapping goodness-of-fit mea- sures in structural equation models. Sociol Methods Res. 1992;21:205–229. doi:10.1177/0049124192021002004
53. Comrey A. A first course on factor analysis. London: Academic Press; 1973.
54. Taylor W, Sallis J, Dowda M, Freedson P, Eason K, Pate R. Activity patterns and correlates among youth: differences by weight status. Pediatr Exerc Sci. 2002;14:418–431.
55. Perry C, De Ayala R, Lebow R, Hayden E. A valida- tion and reliability study of the Physical Activity and Healthy Food Efficacy scale for children (PAHFE). Health Educ Behav. 2008;35(3):346–360. PubMed doi:10.1177/1090198106294892
56. Saelens B, Gehrman C, Sallis J, Calfas K, Sarkin J, Caparosa S. Use of self-management strategies in a 2-year cognitive-behavioral intervention to promote physical activity. Behav Ther. 2000;31:365–379. doi:10.1016/ S0005-7894(00)80020-9
57. Baumgartner T. Estimating the stability reliability of a score. Meas Phys Educ Exerc Sci. 2000;4(3):175–178. doi:10.1207/S15327841Mpee0403_3
58. Thomas J, Nelson J, Silverman S. Research methods in physical activity. 5th ed. USA: Human Kinetics; 2005.
Social Cognitive Measures Physical Activity 555
59. Biddle S, Whitehead S, O’Donovan T, Nevill M. Correlates of participation in physical activity for adolescent girls: a systematic review of recent literature. J Phys Act Health. 2005;2:423–424.
60. Van Der Horst K, Paw MJ, Twisk JW, Van Mechelen W. A brief review on correlates of physical activity and seden- tariness in youth. Med Sci Sports Exerc. 2007;39(8):1241– 1250. PubMed doi:10.1249/mss.0b013e318059bf35
61. Lubans D, Sylva K. Mediators of change following a senior school physical activity intervention. J Sci Med Sport. 2009;12(1):134–140. PubMed doi:10.1016/j. jsams.2007.08.013
62. Culos-Reed S, Gyuresik N, Brawley L. Using theories of motivated behaviour to understand physical activity. In: Singer R, Hausenblas H, Jannelle C, eds. Handbook of sports psychology. 2nd ed. New York: Wiley; 2001.
63. Sallis J, Prochaska J, Taylor W. A review of correlates of physical activity of children and adolescents. Med Sci Sports Exerc. 2000;32(5):963–975. PubMed
64. Ferreira I, van der Horst K, Wendel-Vos W, Kremers S, van Lenthe FJ, Brug J. Environmental correlates of physical activity in youth—a review and update. Obes Rev. 2007;8(2):129–154. PubMed doi:10.1111/j.1467- 789X.2006.00264.x
65. Sallis J, Prochaska J, Taylor W, Hill J, Geraci J. Correlates of physical activity in a national sample of girls and boys in grades 4 through 12. Health Psychol. 1999;18(4):410–415. PubMed doi:10.1037/0278-6133.18.4.410
66. Hoefer WR, McKenzie TL, Sallis JF, Marshall SJ, Conway TL. Parental provision of transportation for adolescent
physical activity. Am J Prev Med. 2001;21(1):48–51. PubMed doi:10.1016/S0749-3797(01)00314-2
67. Sallis J, Conway T, Prochaska J, McKenzie T, Mar- shall S, Brown M. The association of school envi- ronments with youth physical activity. Am J Public Health. 2001;91(4):618–620. PubMed doi:10.2105/ AJPH.91.4.618
68. Norman G, Sallis J, Gaskins R. Comparability and reliabil- ity of paper- and computer-based measures of psychosocial constructs for adolescent physical activity and sedentary behaviors. Res Q Exerc Sport. 2005;76(3):315–323. PubMed
69. Durant N, Kerr J, Harris SK, Saelens BE, Norman GJ, Sallis JF. Environmental and safety barriers to youth physi- cal activity in neighborhood parks and streets: reliability and validity. Pediatr Exerc Sci. 2009;21(1):86–99. PubMed
70. Dishman R, Motl R, Sallis J, et al. Self-management strategies mediate self-efficacy and physical activity. Am J Prev Med. 2005;29(1):10–18. PubMed doi:10.1016/j. amepre.2005.03.012
71. Contento I, Manning A, Shannon B. Research perspec- tive on school-aged nutrition education. J Nutr Educ. 1992;24:247–260. doi:10.1016/S0022-3182(12)81240-4
72. ABS. 1301.0—Year Book Australia, population distribu- tion: Aboriginal and Torres Strait Islander Australians, 2008. 2008 [cited 2/12/2010]; Available from: http://www. abs.gov.au/ausstats
73. ABS. 1301.0—Year Book Australia—cultural diversity, 2008. 2008 [cited 2/12/2010]; Available from: http://www. abs.gov.au/Ausstats/abs
Copyright of Journal of Physical Activity & Health is the property of Human Kinetics Publishers, Inc. and its
content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for individual use.