A

iiEgSERseam©

K

1

doBfvtatwaf

1d

Available online at www.sciencedirect.com

Journal of Science and Medicine in Sport 12 (2009) 573–578

Original paper

Evidence for the validity of the Children’s Attraction to PhysicalActivity questionnaire (CAPA) with young children

Elizabeth Rose a, Dawne Larkin b, Beth Hands c,∗, Barbara Howard a, Helen Parker c

a Edith Cowan University, School of Exercise, Biomedical and Health Sciences, Australiab The University of Western Australia, School of Sport Science, Exercise & Health, Australia

c University of Notre Dame Australia, School of Health Sciences, Australia

Received 28 August 2008; received in revised form 14 May 2009; accepted 18 May 2009

bstract

Attraction to physical activity is important to an individual’s intrinsic motivation to engage in play, games and sports. While there arenstruments designed to measure attraction to physical activity in middle childhood years, the lack of authentic measures in young children hasmpeded research in this area. In this study we sought to address the validity of a scale to tap young children’s attraction to physical activity.vidence for validity was based on internal consistency, content analysis, and factor structure. Australian school children (180 boys and 154irls) from school year two, aged 6–8 years, were individually administered a modified version of the Children’s Attraction to Physical Activitycale (CAPA) [Brustad RJ. Who will go out to play? Parental and psychological influences on children’s attraction to physical activity. Pediatrxerc Sci 1993;5:210–23; Brustad RJ. Attraction to physical activity in urban school children: parental socialization and gender influences.es Q Exerc Sport 1996;67:316–23]. The results indicated that internal consistency was acceptable for most of the subscales when negative

tatements were excluded from the analyses. Factor analysis revealed that the liking of games and sports, liking of physical exertion and

xercise, and the importance of exercise subscales were more robust. Second order factor analysis indicated that the overall construct ofttraction to physical activity was viable in this age group. With some modifications, the scale appears to provide a valid approach to theeasurement of attraction to physical activity in young children.2009 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved.

foialdravfc

eywords: Children; Measurement of attraction to physical activity

. Introduction

Attraction to physical activity refers to an individual’sesire to participate in a task involving physical exertionr movement, through play, games or sport. According torustad1 both cognitive and emotive states contribute to the

ormation of a child’s attraction to physical activity. Moti-ational theorists3,4 propose that children who are attractedo physical activity by challenge, mastery, and enjoyment,s well as support from significant others, are more likelyo engage in physical play, games and sports than children

ho do not experience these cognitions and emotions. Thus

ttraction to physical activity in childhood provides a basisor enjoyment and participation in games, play and sports

∗ Corresponding author.E-mail address: bhands@nd.edu.au (B. Hands).

aoAmsp

440-2440/$ – see front matter © 2009 Sports Medicine Australia. Published by Eloi:10.1016/j.jsams.2009.05.009

or later in life. For example O’Brien Cousins5 found thatlder women’s enjoyment and engagement in physical activ-ty later in life were reflective of both level of engagementnd enjoyment of physical activity level as children. In theirongitudinal study Thompson et al.6 revealed that the ten-ency to be physically active (or inactive) as an adult waseflected by the quality of childhood and adolescent physicalctivity experiences. Their findings provide support for theiew that specific relationships, circumstances, and attitudesormed in childhood and adolescence influence adult physi-al activity attitudes and behaviours. It is also evident fromn early age that enjoyment is a very salient feature of thisverall attraction. While Brustad1 developed the Children’s

ttraction to Physical Activity scale (CAPA) for upper pri-ary school children (10–12 years old) there is considerable

upport for the importance of exploring what attracts lowerrimary school children (6–8-year olds) to physical activity.

sevier Ltd. All rights reserved.

5 nd Med

Hacptty

diidtattclpaaaTrbiaefCt

iPyoaitw

atcrYsd

scy

2

aarogDp

cse(oAHrqNs

FC

74 E. Rose et al. / Journal of Science a

owever little is known of the authentic means by whichttraction to physical activity can be measured in younghildren. In spite of advances in the measurement of self-erceptions7 and self-descriptions8 many questions remainhat highlight the need for a more thorough investigation ofhe validity of measures when adapting scales for use withounger children.

When developing the CAPA scale, Brustad1 identifiedifferent aspects of attraction to physical activity, stress-ng the importance of a multi-dimensional approach thatncorporated cognitive and affective dimensions. The CAPAiffers considerably from earlier measures tapping attractiono physical activity. The items focus on “children’s feelingsbout their physical activity involvement along dimensionshat children themselves have identified as being salient fea-ures of the physical activity experience”.1(p221) The CAPAomprises five subscales: (a) liking of games and sport, (b)iking of physical exertion and exercise, (c) liking of vigoroushysical activity, (d) peer acceptance in games and sports,nd (e) importance of exercise. The three liking subscalesnd the peer acceptance subscale relate to the enjoymentnd fun element of children’s attraction to physical activity.he importance of exercise subscale focuses on the thoughts

elated to children’s attraction to physical activity for healthenefits. The CAPA scale was validated with a North Amer-can sample from school grades four to seven, ranging inge from 9 to 12 years.1,2 Although preliminary studies werencouraging, Brustad1,2 noted that there was clear need forurther examination of the psychometric properties of theAPA, particularly with different populations, to ensure that

he CAPA was suitable for more widespread use.There also are administration considerations when adapt-

ng scales to suit younger children. For example Harter andike9 and Marsh et al.8 reported the difficulties faced byoung children in answering questionnaires formatted forlder children. Young children are less able to differenti-

te among some domains such as physical appearance andnterpersonal relationships.3,8,10 Consequently it is impor-ant to consider the developmental level of children asell as the manner in which questionnaires are formatted

aTsp

ig. 1. The top item is a sample from the original CAPA scale (Brustad1) using a stAPA scale using a Likert format. In each case, from left to right, the scores range

icine in Sport 12 (2009) 573–578

nd administered.3,8–10 Reports with previous questionnaireshat used a structured alternative format8,10,11 indicate thathildren younger than 8 years are not yet able to answer accu-ately and do not understand how to respond in this format.oung children are better able to respond to questions pre-ented verbally and on an individual basis. They also haveifficulty with negatively worded items.8

Our purpose in this study was to examine: (a) internal con-istency reliability, (b) subscale structure and (c) the overallonstruct validity of a modified version of the CAPA withounger Australian children aged 6–8 years.

. Method

Participants were Australian school children (180 boysnd 154 girls) from year two classes in 28 schools with a meange of (M = 6.7, SD = .74) years. The selected schools wereepresentative of high to low socio-economic status and weref equivalent size. Approval to carry out the research wasranted by the ethics committee of the University of Notreame, Australia. Data were collected on a school basis, withassive parent consent.

The CAPA1,2 includes 25 items to measure the extent ofhildren’s interest in physical activity. There are five sub-cales: (a) liking of games and sports, (b) liking of physicalxertion and exercise, (c) liking of vigorous physical activity,d) peer acceptance in sports and games, and (e) importancef exercise. Each of the CAPA subscales included five items.ll items were scored from one (low) to four (high) usingarter’s7 structured alternative format that is designed to

educe socially desirable responses (Fig. 1). The originaluestionnaire was designed for group administration withorth American children with a mean age of 10 years and

tructured in an alternative question format.For this study, the CAPA was modified. The structured

lternative format was replaced with a Likert format (Fig. 1).he responses to statements from each of the items werecored from one to four. A sample item from the liking ofhysical exertion and exercise subscale from the modified

ructured alternative format. The bottom item is a sample from the modifiedfrom one (low) to four (high).

nd Medicine in Sport 12 (2009) 573–578 575

vaJtabt

etwtacpiatetwctt

ww“(tcs

o>ntfttwC0a

3

t1ltvii

Table 1A comparison of internal consistency reliability for Brustad2 (age range:10–12 years) and the present study (age range: 6–8 years).

Subscales Present study, α Brustad2, α

Peer acceptance in sports and games .55 (.30) .72Importance of exercise .72 .44Liking of games and sport .75 .70Liking of physical exertion and exercise .67 .74Liking of vigorous physical activity .74 (.64) .74

Nt

verT

oBntettCdavcctwa‘fatytgclp

awoov2ti

E. Rose et al. / Journal of Science a

ersion of the CAPA is presented in Fig. 1. This change waspproved by the author (Brustad, personal communication,anuary 2005). The CAPA is generally administered in writ-en form but in this study, because the children were younger,dministration was conducted on an individual basis with ver-al questioning by the researcher and verbal responses fromhe child.

Six assistants participated in 2 days training by an experi-nced research officer on strategies to ensure strict adherenceo protocol regarding the behaviour, language and techniqueshen interviewing to ensure the reliability of data collec-

ion. The research officer had extensive experience in thedministration of the questionnaire and maintained protocolonsistency through regular team meetings. Each child com-leted the 20 min questionnaire in their regular school timen a quiet area, away from teachers and peers. Prior to thedministration of the CAPA scale, the researcher explainedhat the questions were related only to themselves and no onelse would know these answers. Each child also was informedhat the procedure was not a test and that there were no right orrong answers. Each child completed a practice question to

larify any difficulties in understanding. Following the prac-ice questions the children responded to the questions fromhe main questionnaire.

In the first instance, each child answered the statementith a yes or no answer. The child was then asked to respondith “no always” or “no sometimes” or “yes sometimes” or

yes always” (see Fig. 1), yielding a score ranging from onelow) to four (high). The researcher circled the response onhe questionnaire once the participant gave an answer. If thehild had difficulty responding, the researcher repeated thetatement and checked for understanding.

Internal consistency reliabilities of the subscales werebtained using Cronbach’s alpha with a level of acceptability.60.12 First order factor analyses, using principal compo-ents analysis with promax rotation, provided evidence ofhe subscale structure. Second order factor analyses obtainedrom promax rotation and regression coefficients were used toest for a common underlying structure representing ‘attrac-ion to physical activity’. The loading of items onto factorsas examined with respect to the subscales in the originalAPA scale. We chose to retain all items with loadings above.3 and eigenvalues above 1.13 SPSS version 15 was used forll analyses.

. Results

Internal consistency reliabilities for the subscales used inhis study were acceptable for three of the dimensions [>.60,2]; importance of exercise, liking of games and sports, andiking physical exertion and exercise. The internal consis-

ency reliabilities of the remaining two subscales, liking ofigorous activity and peer acceptance were low to moderaten their initial form, however removal of negatively wordedtems resulted in higher reliability coefficients for liking of

fsrp

ote: Number in the bracket is the internal consistency reliability prior tohe removal of the negatively worded questions.

igorous physical activity but peer acceptance remained mod-rate (.55). Comparisons of internal consistency reliabilitieseported by Brustad2 and for this study are presented inable 1.

The initial exploratory factor analysis provided evidencef sampling adequacy (Kaiser–Meyer–Olkin measure = .907;artlett’s Test of Sphericity = p < .001). Principal Compo-ents Analysis of the 25 items of the CAPA scale revealedhe presence of six factors with eigenvalues exceeding 1,13

xplaining 29.1%, 6.5%, 6.2%, 5.4%, 4.5% and 4.3% ofhe variance, respectively (Table 2). While two of these fac-ors were fairly consistent with the subscales in the originalAPA, the remaining factors were not. The 1st factor wasominated by items from liking of games and sports butlso included two items from the subscale for the liking ofigorous activity. The 2nd factor was dominated by itemsoncerning importance of exercise. The remaining factorsontained items from different subscales with the excep-ion of the 5th factor which consisted of just two negativelyorded items from the ‘peer acceptance’ subscale (Items 7

nd 8, Table 2). Items from the ‘peer acceptance’ and theliking of vigorous activity’ subscales loaded on many of theactors. Another factor model (Table 3), eliminating the ‘peercceptance’ subscale and the negatively worded item fromhe ‘liking of vigorous activity’ subscale (Item 6, Table 2)ielded four components which together explained 58.8% ofhe variance. Two of the four original subscales, liking ofames and sports and the importance of exercise, were moreohesive using this model. While the remaining subscales,iking of physical exertion and exercise and liking of vigoroushysical activity tended to load together on the third factor.

Two second order factor models explored the overall char-cteristics of the scale (Tables 4 and 5). When all 25 itemsere included, the second order analysis yielded two factors,ne based on the three negatively worded items and the othern the remaining 22 items which explained 57.9% of theariance. Component 1 contributed 39.03% and Componentcontributed 18.89%. The second analysis which excluded

he peer acceptance items and the negative item from the lik-ng of vigorous activity subscale yielded just one common

actor which we called ‘Attraction to Physical Activity’. Thisolution explained 52.7% of the variance and supports theemoval of the negatively worded items and those relating toeer acceptance with this age group.

576 E. Rose et al. / Journal of Science and Medicine in Sport 12 (2009) 573–578

Table 2Six factor solution using a principal components analysis based on the five subscales (25 Items) of the CAPA.

CAPA items Factors

1 2 3 4 5 6

1. I like playing outdoor games and sport .90520. I look forward to playing games and sports .6905. I am told that I’m good at games and sports .580 −.3273. I have more fun playing games and sports than anything else .510 .45311. I enjoy exercise a lot .463 .30813. I wish I could play more games and sports .362 .3064. I like to exercise lots .35319. I think exercise is very important for my health .88122. I think that exercise is the most important thing for good health .6969. I think that the more exercise you get, the better. .62223. I really like to exercise .329 .56721. I like to burn lots of energy by playing hard .325 .62124. I feel good when I run hard .59515. I don’t mind getting out of breath after I play hard .58210. I make a lot of friends when I play games and sports .49314. I think that I will feel really good after I play hard .37718. I really like to run a lot .75025. I am popular when I play games and sports .6522. I like getting sweaty when I exercise or play hard .329 .59617. Playing games and sports is my favourite thing .335 .58816. I think it is very important to always be in good shape .538 .3088. Reversed. I get teased by other kids when I play games and sports .793761 .402

4

Csff

ito

TF

C

1231112921422121111

. Reversed. I get nervous and tired about playing games and sports

. Reversed. I feel really tired after I play games and sports2. I try to stay in good shape

. Discussion

The results present some evidence for the validity of the

APA with this young age group based on internal con-

istency within the subscales, subscale evidence based onactor analyses and construct evidence based on second orderactor analyses. Four of the five subscales had acceptable

tsst

able 3our factor solution using a principal components analysis based on four subscales

APA items

. I like playing outdoor games and sport0. I look forward to playing games and sports. I have more fun playing games and sports than anything else3. I wish I could play more games and sports1. I enjoy exercise a lot9. I think exercise is very important for my health2. I think that exercise is the most important thing for good health. I think that the more exercise you get, the better.3. I really like to exercise2. I try to stay in good shape. I like to exercise lots4. I feel good when I run hard. I like getting sweaty when I exercise or play hard5. I don’t mind getting out of breath after I play hard1. I like to burn lots of energy by playing hard4. I think that I will feel really good after I play hard8. I really like to run a lot7. Playing games and sports is my favourite thing6. I think it is very important to always be in good shape

.763 -.360-.850.419

nternal consistency reliability. The factor analyses revealedhat there were two clear subscales. The evidence for usef the subscales was less compelling with this age group

han found with the older children using the original CAPAcale.1 The second order factor analysis, based on four sub-cales provided better support for the construct of ‘Attractiono Physical Activity’.

(19 Items) of the CAPA.

Factors

1 2 3 4

.867

.789

.715

.504

.504.860.736.628.527.391 −.361 .331

.636

.610 .472

.595.317 .540.316 .329

.727.391 .579

.362 .567

E. Rose et al. / Journal of Science and Med

Table 4Second order factor solution based on total CAPA scale.

Component

1 2

REGR factor score 1 .772REGR factor score 4 .747REGR factor score 2 .730RRR

rg‘WtppCvawyw

cswbmtsdrd

abswop

siaa

TS

RRRR

ygtauaotsSeldmaccd

tgftfiiau

tcOattps2smacstd

EGR factor score 3 .715 −.318EGR factor score 5 .833EGR factor score 6 .590

With the young children in this study, internal consistencyeliability was acceptable for four of the subscales, ‘liking ofames and sports’, ‘liking of physical exertion and exercise’,liking of vigorous activity’ and ‘importance of exercise’.

ith the age group in this study, negatively worded ques-ions reduced the internal consistency of two of the subscales,eer acceptance in sports and games, and liking of vigoroushysical exercise. When the negative items were removed theronbach’s alpha increased from .64 to .74 for the liking ofigorous physical exercise, and from .30 to .55 for the peercceptance subscale. It is not surprising that the negativelyorded items did not work well as Marsh14 has reported thatoung children are unable to respond reliably to negativelyorded items.It is interesting to note that Brustad2 reported low internal

onsistency reliability for the importance of exercise sub-cale with his older sample. This scale performed relativelyell with the younger children in this study with a Cron-ach’s alpha of .72. A possible explanation for the differencesight relate to the heightened community awareness about

he importance of physical activity for health since the earlytudy by Brustad around 15 years ago. In Australia, chil-ren in this age bracket are also exposed to physical activityelated health messages at school (for example http://www.et.wa.edu.au/education/physicalactivity/advocacy.html).

In contrast to our study, the peer acceptance in sportsnd games subscale had higher internal consistency relia-ility for the older sample involved in Brustad’s2 validationtudy. This was due in part to the inclusion of two negativelyorded items. We also need to consider the greater influencef parental feedback on younger children’s participation inhysical activity.15

Further support for the removal of the peer acceptance

ubscale with the younger age group was provided by thenitial factor analysis. Peer acceptance items loaded acrossll six factors. With the removal of the items measuring ‘peercceptance in games and sport’ in the second factor anal-

able 5econd order factor component based on four CAPA subscales.

Component 1

EGR factor score 1 .818EGR factor score 2 .783EGR factor score 4 .665EGR factor score 3 .621

ttrm

P

rpd

icine in Sport 12 (2009) 573–578 577

sis there was good evidence for two subscales (liking ofames and sport and importance of exercise). Children inhis study were able to differentiate between liking of gamesnd sport and importance of exercise. However they werenable to distinguish between physical exertion and exercisend vigorous activity as these items generally loaded togetherr across several factors. This is not surprising consideringhat the items on each scale are very similar, using wordsuch as sweaty, out of breath, play hard and burning energy.imilarly, it was of interest to note that the item ‘I enjoyxercise a lot’ from the ‘liking of vigorous activity’ subscale,oaded with ‘liking of games’ suggesting that young childreno not distinguish between the concepts of liking and enjoy-ent. This too could be explained by similar words such

s ‘like playing’ and ‘look forward to’. Given that younghildren have difficulty differentiating between some of theoncepts, caution must be used in interpreting the subscaleata.

The second order factor analysis provided evidence forhe construct of ‘attraction to physical activity’ for this ageroup. Although the initial second order analysis yielded twoactors, one was based on factors one to four which con-ained most of the items, while the other was based on factorsve and six which were dominated by negatively worded

tems. Removing the ‘peer acceptance’ subscale and the neg-tively worded items resulted in better evidence for a singlenderlying construct.

Our study has several strengths. Firstly, few studies chooseo investigate self-perceptions in young children as it is diffi-ult to gather authentic and valid data from children this age.ur data were collected individually by a trained researcher inquiet corner of the child’s classroom, a safe and familiar set-

ing. We were able to check for understanding and minimisehe tendency for young children to respond in the way theyerceive the researcher would wish. Our sample was large fortudies of this type. We interviewed 334 children attending8 schools representing a broad spectrum of socio-economictatus. A limitation of the study relates to the challenge ofodifying a scale designed for 10–12-year-old children toyounger age group. Young children are less able to dis-

riminate subtle differences in the meaning of some wordsuch as ‘liking’ and ‘enjoyment’ and therefore are less ableo differentiate among a range of domains than older chil-ren. This may necessitate even further modification of bothhe questionnaire structure and wording, as well as reducinghe number of subscales for use with young children. Furtheresearch involving qualitative methods is likely to provide aore in depth understanding of children’s perceptions.

ractical implications

In keeping with other studies with older children,16 weecommend that an overall score representing attraction tohysical activity can be reported with these 2nd grade chil-ren. However we further recommend that measurement of

5 nd Med

tntCp

R

of physical competence information; a cluster analysis. J Sport Exerc

78 E. Rose et al. / Journal of Science a

he overall construct with young children does not includeegatively worded items. Despite the difficulty of interpretinghe subscales with this age group, we suggest that the overallAPA score is a useful measure to represent attractiveness tohysical activity among 6–8-year-old children.

eferences

1. Brustad RJ. Who will go out to play? Parental and psychological influ-ences on children’s attraction to physical activity. Pediatr Exerc Sci1993;5:210–23.

2. Brustad RJ. Attraction to physical activity in urban school chil-dren: parental socialization and gender influences. Res Q Exerc Sport1996;67:316–23.

3. Harter S. The construction of self: a developmental perspective. NewYork: Guilford Press; 1999.

4. Nicholls JG. What is ability and why are we mindful of it? In: SternbergRJ, Kolligian J, editors. Competence considered. New Haven, CT: YaleUniversity Press; 1990. p. 11–40.

5. O’Brien Cousins S. Exercise, aging and health. Philadelphia, PA: Tay-

lor & Francis; 1998.

6. Thompson AM, Humbert ML, MirwaldIt RL. A longitudinal study ofthe impact of childhood and adolescent physical activity experienceson adult physical activity perceptions and behaviours. Qual Health Res2003;13. pp. 358–357.

icine in Sport 12 (2009) 573–578

7. Harter S. The perceived competence scale for children. Child Dev1982;53:87–97.

8. Marsh HW, Craven RG, Debus RL. Self-concepts of young childrenaged 5 to 8: their measurement and multidimensional structure. J EdPsychol 1991;83:377–92.

9. Harter S, Pike R. The pictorial scale of perceived competence and socialacceptance for young children. Child Dev 1984;55:1969–82.

10. Marsh HW, Ellis LA, Craven RG. How do preschool children feelabout themselves? Unraveling measurement and multidimensional self-concept structure. Dev Psychol 2002;30:376–93.

11. Marsh HW, Craven RG, Debus RL. Structure, stability, and develop-ment of young children’s self-concepts: a multi-cohort–multi-occasionstudy. Child Dev 1998;69:1030–53.

12. Wasserman JD, Bracken BA. Psychometric characteristics of assess-ment procedures. In: Graham JR, Naglieri JA, Weiner IB, editors.Handbook of psychology, assessment psychology, vol. 10. Hoboken,NF: John Wiley & Sons; 2003.

13. Tabachnick BG, Fidell LS. Using multivariate statistics. 5th editionBoston, MA: Allyn and Bacon; 2007.

14. Marsh HW. Negative item for rating scales for preadolescent children:a cognitive-developmental phenomenon. Dev Psychol 1986;22:37–49.

15. Weiss MR, Ebbeck V, Horn TS. Children’s self-perceptions and sources

Psychol 1997;19:52–70.16. Welk GJ, Wood K, Morss G. Parental influences on physical activity

in children: an exploration of potential mechanisms. Pediatr Exerc Sci2003;15:19–33.