A icle
Sel -iden i y as a
componen o he Theo y
o Planned Beha iou in
p edic ing physical ac i i y
F ancis Ries
Uni e si y o Se ille, Spain
Vello Hein and Ma e Pihu
Uni e si y o Ta u, Es onia
Jose
´Manuel Se illano A men a
Uni e si y o Se ille, Spain.
Abs ac
This s udy aimed o in es iga e he ole o sel -iden i y, de ined as salien and endu ing aspec s o
one’s sel -pe cep ion (Spa ks, 2000), in ela ion o adolescen physical ac i i y (PA) in en ions
wi hin he Theo y o Planned Beha iou (TPB). School s uden s aged 12 o 18 om wo cul u al
g oups (Es onia and Spain) comple ed measu es o a i udes, subjec i e no ms, pe cei ed beha-
iou al con ol, sel -iden i y, and in en ions o PA. Fou weeks la e pa icipan s comple ed
sel - epo ed measu es o PA. A s uc u al equa ion model showed ha sel -iden i y had a di ec
e ec on PA in en ion (b¼.33, p< .01) and PA beha iou (b¼.31, p< .01). The model accoun ed
o 45% o he a iance o PA beha iou om which 4% can be a ibu ed o sel -iden i y. The e ec
o sel -iden i y on PA was also pa ially media ed ia in en ion. The mul i-g oup compa ison
be ween wo cul u al g oups indica ed ha no in a iances exis ed be ween he models o he
obse ed samples.
Keywo ds
Sel -iden i y, Theo y o Planned Beha iou , physical ac i i y, mo i a ion
Co esponding au ho :
F ancis Ries, Depa men o Physical Educa ion and Spo s, Uni e si y o Se ille, Calle Pi o ecnia, s/n., 41013 Se ille, Spain
Email: [email p o ec ed]
Eu opean Physical Educa ion Re iew
18(3) 322–334
ªThe Au ho (s) 2012
Rep in s and pe mission:
sagepub.co.uk/jou nalsPe missions.na
DOI: 10.1177/1356336X12450792
epe.sagepub.com
In oduc ion
Theo y o Planned Beha iou (TPB) is a popula and aluable social-psychological model ha has
been employed o inc ease ou unde s anding o physical ac i i y (PA) beha iou a he le el o
decision-making (Ajzen, 1985, 1991). I s a es ha in en ions o engage in beha iou a e he p i-
ma y de e minan s o ac ual beha iou . B ie ly, in en ions a e concep ualized as he summa y
mo i a ions o pe o m beha iou and media e he in luence o he h ee main TPB cons uc s
on beha iou (Ajzen, 1991). A i ude is he i s de e minan which e lec s an indi idual’s pe -
sonal belie s abou enac ing a a ge beha iou , such ha an indi idual who has a posi i e a i ude
owa d PA will be much mo e likely o plan o exe cise han a pe son who belie es ha exe cise
has no alue. The second de e minan is subjec i e no ms, which e lec he pe cei ed expec a ions
o speci ic indi iduals o g oups ega ding he adop ion o a gi en beha iou and which may ha e
a ious sou ces: whe he he indi idual’s cul u e alues and p omo es PA; whe he signi ican o he s
p ac ise PA; and how hey eac o his beha iou . The inal de e minan , pe cei ed beha iou al con-
ol (PBC), e lec s pe cep ion o he ease o di icul y o pe o ming he beha iou (Ajzen, 1991).
Fu he , Ajzen (1991) hypo hesized ha PBC p ima ily in luences beha iou h ough in en ions bu ,
o he ex en ha beha iou al con ol e lec s ac ual con ol, i can also in luence beha iou di ec ly.
Sys ema ic e iews (A mi age and Conne , 2000) and me a-analyses (Conne and Spa ks, 2005;
Hagge e al., 2002) ha e demons a ed ha he TPB is use ul o he p edic ion o heal h beha-
iou s. In gene al, he e is s ong con i med suppo o TPB p edic ing exe cise in en ions and
beha iou (Downs and Hausenblas, 2005; Hagge e al., 2002; Hagge e al., 2007). Recen ly in
Spain se e al s udies (Espı´, 2004; Gil e al., 2004; Mon il, 2004; Ries e al., 2009) employed he
TPB in a PA domain, whe eas in Es onia o he esea che s (Hagge e al., 2007; Hagge e al.,
2009; Pihu e al., 2008) ha e used he cons uc s o TPB o p edic PA beha iou among school-
child en. Resul s wi h Es onian schoolchild en indica ed ha a i udes and pe cei ed beha iou al
con ol had a main ole in p edic ing hei in en ion o be physically ac i e in hei leisu e- ime
(Hagge e al., 2007). Fu he , in en ion p edic ed sel - epo ed PA beha iou , and media ed he
impac o a i udes and PBC on beha iou as sugges ed by Ajzen (1985). The e was no e ec o
subjec i e no ms on in en ion among Es onian schoolchild en. The esul s o his s udy indica ed
ha he explained a iance o he exe cise beha iou was 52%. I has been ound (Su on, 1998)
ha he TPB ypically explains he a iance o in en ion be ween 40%and 50%, and he a iance
in beha iou be ween 19%and 38%. Howe e , a la ge amoun o he a iance emains unex-
plained; hence leading esea che s sugges he addi ion o o he a iables o imp o e he p edic i e
capaci y o he TPB. The op ion o including mo e p edic o s was unequi ocally le open by he
‘ a he s’ o he heo y as long as he e would be a s ong heo e ical jus i ica ion o hei inclusion
and hey would cap u e a signi ican po ion o unique a iance in in en ions o beha iou (Ajzen,
1991; Fishbein, 2000; Fishbein and Ajzen, 2010). A signi ican numbe o s udies ha e iden i ied
addi ional a iables ha migh imp o e he p edic i e powe o he TPB, e.g. planning and pos -
in en ional oli ional p ocess (Da ke e al., 2010), sel -iden i y (Hagge and Cha zisa an is, 2006),
pas beha iou , sel -iden i y, g oup no ms, amily social suppo , iends’ social suppo , and social
p o isions (Hamil on and Whi e, 2008), desc ip i e no ms, mo al no ms, an icipa ed eg e ,
sel -concep and pas beha iou (Jackson e al., 2003), and pe cei ed pa en al suppo and pa en s’
physical ac i i y (Ries e al., 2009).
Also, in many s udies he TPB and Sel -De e mina ion Theo y (SDT) ha e been in eg a ed wi h
he pu pose o p o iding mo e explana ions o he p ocesses ha ela ed o PA. SDT ep esen s a
b oad amewo k o he s udy o human mo i a ion and pe sonali y (Deci and Ryan, 1985a, 2000).
Ries e al 323
Cen al o he heo y is he dis inc ion be ween au onomous and con olling o ms o mo i a ion.
Deci and Ryan (2000) ha e sugges ed ha au onomous mo i a ion should p o ide a basis o he
o ma ion o social cogni i e judgmen s owa d speci ic beha iou s. Measu es o he mo i a ional
o ien a ions om SDT e lec an indi idual’s ‘cu en ’ pe cei ed mo i a ional s a us, while he
cons uc s o he TPB a e measu es o expec ancies ega ding ‘ u u e’ beha iou al engagemen .
Resea ch has ound ha gene alized au onomous mo i es ac as an in o ma ion p ecu so in ol ed
in he o ma ion o judgmen s and expec a ions ega ding u u e beha iou . Se e al esea che s
(Ajzen, 1991, Conne and Ab aham, 2001; Hagge e al., 2006) ha e p oposed ha he o ma ion
o he social cogni i e cons uc s om he TPB d aws om disposi ional cons uc s like pe sonal-
i y, as well as belie s ega ding he beha iou . The esul s o he in eg a ed model o TPB and SDT
ha e shown ha au onomous mo i a ion p edic s in en ions ia he media ion o a i udes and PBC
(Cha zisa an is e al., 2002; Hagge e al., 2006; Hagge e al., 2009; Pihu e al., 2008).
Many heo ies o social cogni ion and mo i a ion p opose ha hei p edic ions ep esen
uni e sal mo i a ional and decision-making p ocesses, and should be consis en ac oss samples
(Ajzen, 1985; Deci and Ryan, 1985b). C oss-cul u al esea ch is he e o e impo an o es ablish
whe he e ec s in models a e uni e sal, as p oposed by social cogni i e and mo i a ional heo is s
(Ajzen, 1985; Chi ko and Ryan, 2001; Deci and Ryan, 1985b), o a e a ec ed by indi idual di -
e ences in cul u al o ien a ions, as p oposed by he cul u e-cen ic app oach (Ma kus and
Ki ayama, 1991). The p esen s udy also aims o add o exis ing knowledge by analysing he esul s
in samples ha ei he ha e p edominan ly collec i is (Spain) o p edominan ly indi idualis
(Es onian) cul u al no ms. This will p o ide e idence as o whe he he model p ocesses a e uni-
e sal and can be applied ac oss na ional g oups.
The p esen s udy aims o es an in eg a ed model, whe e mo i a ional cons uc om SDT will
p edic in en ions ia he media ion o TPB cons uc s and he ole o sel -iden i y in ela ion o
adolescen PA in en ions wi hin he TPB.
Sel -iden i y as a componen o he TPB
The in e es o adding sel -iden i y as a p edic o in he TPB de i es a leas in pa o he conclu-
sion ha measu es o subjec i e no ms o en accoun o less a iance in in en ions ha migh be
es ima ed (Ajzen, 1991; Hagge e al., 2006). Sel -iden i y is concep ualized as he salien and
endu ing aspec s o one’s sel -pe cep ion (Spa ks, 2000) and i e lec s he ex en o which a pe son
sees him- o he sel as ul illing he c i e ia o any socie al ole (Conne and A mi age, 1998).
Sel -iden i y is conside ed o ap in o a wide social con ex han he TPB (Cha ng e al., 1988).
The mo e salien he sel -iden i y is, he g ea e he p obabili y ha he pe son will consequen ly
beha e wi h ha iden i y. In o he wo ds, someone who s ongly sees himsel as a spo y pe son is
mo e likely o ake e e y oppo uni y o p ac ise PA (Jackson e al., 2003). Sel -iden i y is one o
he c ucial aspec s ha mos in luence he pe o mance o ce ain beha iou s; i exe s a simila in lu-
ence independen o social e e ences and a i udes, ha ing no eason o be consis en wi h hem. On
he o he hand, he epe i ion o beha iou in luences sel -concep and sel -iden i y, so ha beha-
iou al p ac ice ans o ms he indi idual iden i y – ha is, he e is a wo-way dialogical ela ionship
o de elopmen and cons uc ion be ween hem (Spa ks, 2000; Spa ks and Shephe d, 1992).
Based on hypo heses om he Role Theo y (Biddle, 1979) and he Social Iden i y Theo y
(Cha ng e al., 1988), sel -iden i y has been ound o be an independen p edic o o in en ions in
se e al s udies (Campbell and Shee an, 2001; Cha ng e al., 1988; Conne and A mi age, 1998;
Spa ks and Shephe d, 1992), whe eas o he esea che s ha e no ound his e ec on beha iou al
324 Eu opean Physical Educa ion Re iew 18(3)
in en ions (e.g. Fekadu and K a , 2001). In a ecen me a-analysis o 40 da a se s om s udies
conce ning he p edic ion o in en ions (Rise e al., 2010) sel -iden i y had a medium-sized a e age
co ela ion wi h beha iou al in en ions. A wo-s ep hie a chical eg ession analysis in ended o cla -
i y whe he sel -iden i y imp o es he p edic ion o beha iou al in en ions beyond ha p oduced by
he TPB on i s own. The inclusion o sel -iden i y cap u ed 6%(p< .001) addi ional a iance in
in en ions abo e and beyond ha p esen ed by he main TPB cons uc s (a i ude, subjec i e no ms,
and PBC) and aised i by 9%(p< .001) when pas beha iou was also con olled. The same s udy
sugges ed ha he in luence o sel -iden i y on beha iou al in en ions is mos ly, and maybe en i ely,
media ed by he s eng h o he a ge beha iou . The indings concluded ha sel -iden i y is a i al
p edic o o in en ions and beha iou and should he e o e be inco po a ed in o he TPB.
Al hough he e is impo an suppo o he in luence o sel -iden i y on beha iou al in en ions,
he e is a lack o esea ch examining he ole o sel -iden i y in ela ion o adolescen PA in en ions
wi hin he TPB. Lau e al. (2005) ound ha , when adolescen s iden i y hemsel es as being spo y
pe sons, his p esence o an impo an spo iden i y s eng hens spo beha iou . Ou s udy, he e-
o e, wan ed o e alua e he in luence o sel -iden i y on adolescen PA in en ions and o es he
alidi y o an ex ended TPB model, inco po a ing sel -iden i y as an addi ional a iable, o
p edic ing and unde s anding adolescen PA.
Me hod
Pa icipan s and design
Pa icipan s we e 397 school s uden s (175 boys and 222 gi ls, mean age ¼15.67, SD ¼1.4, ange:
12 o 18) s udying in i e di e en s a e high schools in Es onia (boys ¼55, gi ls ¼91; Mage
¼16.47, SD ¼.08) and in h ee di e en go e nmen - un seconda y schools in Spain (boys ¼120,
gi ls ¼131; Mage ¼15.69, SD ¼.07). The school headmas e s ga e in o ma ion on he socio-
economic s a us o he pa icipan s. The schools d aw hei s uden s om an a ea cha ac e ized
as ‘middle-class’ and ma ched he dis ibu ion o socioeconomic s a us le els among own-
dwelling school child en in bo h Es onia and Spain.
S uden s we e aking Physical Educa ion (PE) as a equi ed cou se ( wo imes a week, 45–50
minu es pe lesson). Consen o school pupils’ pa icipa ion in he s udy was ob ained om
pa en s and he school p incipals p io o da a collec ion. S uden s we e in o med ha hey would
be asked o comple e a sho ques ionnai e o e he coming ou weeks as pa o a su ey on young
people. They we e old ha pa icipa ion was olun a y and hey could choose o op ou i hey
desi ed. S uden s comple ed he ques ionnai es in lesson ime; o p ese e con iden iali y, pa i-
cipan s we e asked no o epo hei names. P ospec i e esponses we e ma ched wi h baseline
esponses using da es o bi h, gende and class iden i ica ion as ma ching indexes.
In he i s wa e o da a collec ion, he PBC, a i udes, subjec i e no ms, sel -iden i y and in en-
ion componen s om he TPB and mo i a ion in a leisu e- ime PA con ex we e measu ed. A e
ou weeks, in he second wa e o da a collec ion, PA beha iou was assessed. A ou -week in e -
es pe iod was employed, because i ep esen s a long- ange p edic ion o beha iou ela i e o he
compa a i ely sho - ange e ec s p e iously s udied using he TPB (Hagge e al., 2002).
Measu es
Mo i a ion in leisu e- ime con ex s. Mo i a ion o he leisu e- ime PA was assessed h ough he sub-
scales om Mullan and Ma kland (1997). Responses o each i em we e measu ed on se en-poin
Ries e al 325
scales anging om ‘no ue a all’ (1) o ‘ e y ue’ (7). The ou mo i a ional cons uc s mea-
su ed by i ems om he Beha iou al Regula ion in Exe cise Ques ionnai e (BREQ) we e in e-
g a ed in o a single index o mo i a ion by calcula ing a ela i e mo i a ion index (Valle and
and Ra elle, 2002).
The heo y o planned beha iou . A i ude, pe cei ed beha iou al con ol, subjec i e no ms and
in en ion subscales om he TPB ques ionnai e we e used. Th ee i ems d awn om Cou neya and
McAuley (1994) and Ajzen and Madden (1986) we e used o measu e beha iou al in en ions.
Th ee i ems measu ed beha iou al in en ions (e.g., ‘I in end o do ac i e spo s and/o igo ous
physical ac i i ies in he nex ou weeks ...’) on se en-poin Like - ype scales ancho ed by
‘s ongly disag ee’ (1) o ‘s ongly ag ee’ (7). A i udes we e assessed in esponse o he ollowing
s a emen : ‘Pa icipa ing in ac i e spo s and/o igo ous physical ac i i ies du ing my leisu e ime
in he nex ou weeks is ...’ Responses we e measu ed on ou se en-poin seman ic di e en ial
i ems wi h he ollowing end poin s: bad-good, ha m ul-bene icial, unenjoyable-enjoyable. PBC
was assessed h ough h ee i ems (e.g. ‘I eel in comple e con ol o e whe he I do ac i e spo s
and/o igo ous physical ac i i ies in my leisu e ime in he nex i e weeks’) measu ed on se en-
poin Like - ype scales anging om ‘no con ol’ (1) o ‘comple e con ol’ (7). Subjec i e no ms
we e assessed by h ee i ems (e.g., ‘People impo an o me hink ha I should do ac i e spo s and/
o igo ous physical ac i i ies du ing my leisu e ime in he nex ou weeks’) on se en-poin
scales wi h 1 (‘s ongly disag ee’) o 7 (‘s ongly ag ee’) endpoin s.
Sel -iden i y. Sel -iden i y was measu ed by he h ee i ems p esen ed by Jackson e al. (2003) The
i ems we e: ‘I see mysel as spo y’, ‘I see mysel as i and heal hy’, ‘I see mysel as a physically
ac i e pe son’. The i ems we e assessed on se en-poin Like - ype scales ancho ed by ‘s ongly
disag ee’ (1) o ‘s ongly ag ee (7)’.
Sel - epo ed physical ac i i y beha iou . PA beha iou was assessed a he second wa e o da a
collec ion h ough an adap a ion o he second i em om Godin and Shepha d’s (1985) Leisu e-
Time Exe cise Ques ionnai e (LTEQ). The pe iod o he ac i i y was adap ed om ‘ he las week’
o ‘ he las ou weeks’ and he scale speci ied he equency o he ac i i y du ing his pe iod. The
ques ionnai e asked wo ques ions ega ding how many imes and how o en he indi idual had
engaged in igo ous exe cise o ac i e spo s o a leas 20 minu es in he pas ou weeks. The
s a emen o he i s beha iou i em was: ‘in he cou se o he pas 4 weeks, how o en ha e you
pa icipa ed in ac i e spo s and/o igo ous physical ac i i ies o 20 minu es a a ime?’ wi h scale
poin s labelled almos ne e , a ew imes, a ew imes bu less han hal , on abou hal he days,
mos days, almos e e y day, e e y day.
The second i em ead: ‘I engaged in ac i e spo s and/o igo ous physical ac i i y o 20
minu es a a ime wi h he ollowing egula i y ... ’ wi h he ollowing scale labels: e e y day,
mos days, some days, occasionally, e y seldom, ha dly e e , ne e . In keeping wi h de ini ions o
leisu e- ime ac i i ies (Godin e al., 1986), he ques ionnai e did no include PA ha was pe -
o med du ing no mal school ime since PE is compulso y. Independen e alua ions o he LTEQ
ha e ound i an easy o adminis e sel - epo measu es wi h sa is ac o y alidi y and eliabili y
s a is ics (e.g., Sallis e al., 1993). In addi ion, his wo-i em e sion o he LTEQ has demon-
s a ed sa is ac o y cons uc alidi y and eliabili y in con i ma o y ac o analyses (Hagge and
Cha zisa an is, 2005; Hagge e al., 2006).
326 Eu opean Physical Educa ion Re iew 18(3)

T ansla ion p ocedu es
S anda dized back- ansla ion echniques (B islin, 1986) we e used o ansla e he ques ionnai es
in o Es onian and Spanish. The i s s ep consis ed o ge ing he i ems ansla ed by a bilingual
ansla o in o Es onian and Spanish, and hen ansla ed back in o he sou ce language by indepen-
den bilingual ansla o s who we e blinded o he o iginal ques ionnai es. These e sions we e
hen compa ed wi h he o iginal English e sion and any inconsis encies and e o s we e high-
ligh ed. The back- ansla ion p ocedu e was epea ed i e a i ely un il he o iginal and back-
ansla ed English e sions o he ques ionnai es we e i ually iden ical, as ecommended by
B acken and Ba ona (1991).
Da a analysis
The LISREL 8.8 s a is ical so wa e was employed o conduc he con i ma o y ac o analysis
(CFA) and s uc u al equa ion modelling (SEM). Fi s ly, mul iple impu a ions we e used o eplace
missing obse a ions wi h obse a ions om cases wi h a simila p o ile o sco es. Mul iple
impu a ions we e chosen o gene a e es ima es ha be e e lec ue a iabili y and unce ain y in
he da a han do eg ession me hods. In his me hod he esul s a e combined and he a e age is
epo ed as he es ima e. Adequacy o he CFA and SEM models we e es ima ed by using ec-
ommended inc emen al goodness-o - i indexes: he compa a i e i index (CFI), he non-no med
i index (NNFI), and he oo mean squa e e o o app oxima ion (RMSEA). Fo he CFI and
NNFI cu -o alues o .90 a e gene ally conside ed o ep esen accep able i (Ben le , 1990).
B owne and Cudeck (1989) sugges ha a RMSEA alue o 0.05 o less indica es good i , and ha
alues up o 0.08 ep esen e o s ha app oxima e o hose expec ed in he popula ion.
Resul s
As shown in Table 1 all he subscales eached adequa e in e nal consis ency (a> .80). The mean
sco es we e highe han he conside ed mid- ange alue in all o he s udy a iables, wi h he Mo i-
a ion index being he one wi h he highes a e age (6.53), whe eas In en ion sco ed lowes (4.42),
anging measu es o dispe sion a ound 1.36. As o he co ela ions be ween s udy a iables, all
had a la ge-sized e ec acco ding o Cohen’s (1992) c i e ia, emaining signi ican a p< .05, wi h
he excep ion ha he ela ionship be ween Mo i a ion index and Subjec i e no ms, al hough sig-
ni ican (p< .05) only had a mode a e co ela ion ( ¼.36). On he o he hand, highe co ela ions
Table 1. Desc ip i e s a is ics and co ela ions among s udy a iables.
aMeanSD12345
1. A i ude .88 5.59 1.34 –
2. Pe cei ed beha iou al con ol .84 5.00 1.34 .74*–
3. Subjec i e no ms .82 5.19 1.35 .50*.57*–
4. In en ion .91 4.42 1.08 .70*.79*.41*–
5. Sel -iden i y .90 4.77 1.67 .66*.69*.40*.71*–
6. Mo i a ion index 6.53 5.08 .61*.58*.36*.57*.56*
*p<.05
Ries e al 327
we e e ealed be ween in en ion and PBC ( ¼. 79, p<.05) and he lowes be ween sel -iden i y
and subjec i e no ms ( ¼. 40, p<.05).
Dis ibu ional p ope ies o he esponses o all i ems we e examined. Resul s e ealed skew-
ness alues g ea e han uni y, which indica ed ha no all a iables we e no mally dis ibu ed.
The e o e, no mal sco es we e compu ed o o dinal a iables be o e he es ima ion o he
measu emen model and he s uc u al equa ion model wi h a maximum likelihood me hod
(Jo¨ eskog e al., 2001). The CFA wi h all Spanish and Es onian s uden s p oduced a well- i ing
measu emen model (Table 2, Model 1) whe e each ac o was adequa ely explained by he se o
indica o i ems.
A mul i-sample measu emen model was conduc ed o explo e he deg ee o which he model
was equi alen o Spanish and Es onian samples. An ini ial baseline model was es ima ed o es
whe he he ac o s a e easible ac oss he samples. As shown in Table 2 (Model 2A), he baseline
model p oduced a good i o he da a. Subsequen ly, we es ima ed a nes ed model in which he
ac o loadings and ac o a iance we e cons ained o be in a ian ac oss he wo samples.
Goodness-o - i indexes o his model a e gi en in Table 2 (Model 2B). These models demon-
s a ed a good i wi h he da a and he chi-squa e di e ences be ween he Spanish and Es onian
samples we e no signi ican . Thus mul i-g oup compa ison indica ed ha no in a iances exis ed
be ween he measu emen models o he obse ed samples.
S uc u al equa ion model
The pu pose o he p esen s udy was o es he in luence o mo i a ion and sel -iden i y on ado-
lescen PA in en ions and leisu e- ime PA. Addi ionally, he aim was o es he alidi y o an
ex ended TPB model, inco po a ing sel -iden i y as an addi ional a iable, o p edic ing and
unde s anding adolescen PA. I was expec ed ha sel -iden i y has a signi ican ole in p edic ing
PA in en ion and beha iou . To es he hypo hesis and he di e ences be ween cul u al g oups
(Spain, Es onia) se e al s uc u al equa ion models (SEM) (Figu e 1) we e cons uc ed.
Table 2. Fi indexes o measu emen model and s uc u al equa ion modelling.
Models d CFI RMSEA CI
90
RMSEA NNFI
Model 1 375.22 137 0.98 0.067 0.059 - 0.075 0.98
Model 2A 716.60 327 0.97 0.078 0.071 - 0.086 0.097
Model 2B 689.73 307 0.97 0.078 0.072 - 0.088 0.097
Model 3 427.87 175 0.93 0.077 0.067 - 0.086 0.91
Model 4 322.79 175 0.93 0.076 0.063 - 0.088 0.92
Model 5 513.02 175 0.94 0.070 0.063 - 0.077 0.93
Model 6 532.87 176 0.94 0.073 0.065 - 0.079 0.92
Model 7 558.67 177 0.93 0.074 0.067 - 0.081 0.91
No e: Model 1 ¼Measu emen model wi h all s udy a iables o bo h samples oge he . Model 2A ¼Baseline model;
Model 2B ¼Fac o loading coe icien s and ac o a iance cons ained; Model 3 ¼S uc u al model o Spanish sample
(mo i a ion in leisu e ime, pe cei ed sel -iden i y, a i udes, pe cei ed beha iou al con ol, subjec i e no ms, in en ion,
and PA); Model 4 ¼S uc u al model o Es onian sample; Model 5 ¼S uc u al model o Es onian and Spanish samples
oge he ; Model 6 ¼S uc u al model in which he pa h om sel -iden i y o PA is ixed o ze o. Model 7 ¼S uc u al
model in which he pa hs om sel -iden i y o in en ions and PA a e ixed o ze o.
CFI: compa a i e i index; RMSEA: oo mean squa e e o o app oxima ion; CI
90
RMSEA: 90% con idence in e al; NNFI:
non-no med i index.
328 Eu opean Physical Educa ion Re iew 18(3)
SEM o Spanish and Es onian samples speci ied s uc u al ela ionships be ween cons uc s in
acco dance wi h he hypo hesis ha sel -iden i y has a signi ican ole beside TPB cons uc s in
p edic ing es ima ed PA in en ion and beha iou . Goodness-o - i s a is ics o bo h samples we e
accep able (Table 2, Model 3; 4).
The hypo hesized SEM ha shows he ole o sel -iden i y in p edic ing PA in en ion and beha-
iou in acco dance wi h TPB cons uc s is p esen ed in Figu e 1. Goodness-o - i s a is ics a e p e-
sen ed in Table 2, Model 5. Sel -iden i y had a di ec e ec on PA in en ion (b¼.33, p< .01) and
PA beha iou (b¼.31, p< .01). The model accoun ed o 67%o he a iance o in en ion and 45%
a iance o PA beha iou . To con i m he hypo hesis ha in en ion media ed he e ec o
sel -iden i y on PA beha iou , he di ec pa h om sel -iden i y o in en ion was ixed o ze o
(Table 2, Model 6). A e his he e ec o in en ion on PA beha iou inc eased (b¼.61,
p< .01) indica ing ha pa ial media ion occu ed. The signi ican di e ences in chi-squa e
(Dw
2
¼19.85, Dd ¼1, p< .01) was ollowed be ween Model 5, which included his pa h as a ee
pa ame e , and Model 6, which did no (Table 2). This es ic ed model accoun ed o 4%o a iance
in PA beha iou , which is he amoun o a iance ha can be a ibu ed o sel -iden i y. In addi ion, he
Sobel es (1982) indica ed ha he indi ec e ec o sel -iden i y on PA ia in en ion was signi ican .
To es he ole o sel -iden i y in he ex ended model o he TPB, he in luence o sel -iden i y
o PA in en ion and leisu e- ime PA was ixed o ze o (Table 2, Model 7). A e his he di ec
Mo i a ion
A i ude
Subjec i e
no ms
Pe cei ed
beha iou al
con ol
Sel -iden i y
In en ion Physical
ac i i y
.31*
.55
.65*
.38*
.62* .54*
.33*
.43*
.17*
−.09
.01
.33
(.34)
(.61)
(.59)
Figu e 1. Hypo hesized s uc u al equa ion model o leisu e- ime PA mo i a ion and sel -iden i y p edic ing
PA in en ion and beha iou oge he wi h TPB cons uc s (p< .05). The s anda dized es ima es p esen he
esul s o he s uc u al equa ion model we e Es onian and Spanish samples a e oge he . The pa h coe i-
cien s in pa en heses ep esen he e ec o sel -iden i y in he al e na i e models in which he pa h om sel -
iden i y o PA beha iou was ixed o ze o (Table 2, Model 6).
No e: Es onian and Spanish samples oge he ; Model 6 ¼S uc u al model in which he pa h om sel -iden i y
o PA is ixed o ze o. Model 7 ¼S uc u al model in which he pa hs om sel -iden i y o in en ions and PA
a e ixed o ze o.
Ries e al 329
e ec o mo i a ion on PA beha iou changed signi ican ly (b¼.12, p< .01), indica ing ha
sel -iden i y explains signi ican a iance o he in eg a ed model.
Discussion
The p esen s udy aimed o es he alidi y o an ex ended TPB model, including addi ional sel
and mo i a ional in luences, o p edic ing and unde s anding adolescen PA. The esul s o e
suppo o he TPB in he sense ha i s cons uc s p edic ed in en ions o engage in egula PA a a
ou -week ollow-up. Fo he addi ional sel -in luence a iable, his s udy suppo s he inclusion o
sel -iden i y as a signi ican p edic o o in en ion and beha iou (PA). The esul ha sel -iden i y
appea s as a signi ican p edic o o beha iou al in en ions and pe o mance sugges s ha adoles-
cen s who iden i y hemsel es as physically ac i e pe sons a e mo e likely o p ac ise egula PA
han hose who do no ha e a PA sel -iden i y (Lau e al., 2005). These esul s a e consis en wi h
hose o o he s udies, like Jackson e al. (2003), who con i med he signi ican ela ionship
be ween sel -iden i y and beha iou al in en ion ( ¼.50). Theodo akis (1994) e ealed a lowe
co ela ion ( ¼.31), whe eas Thompson and Rise (2002) ound he highes co ela ion be ween
bo h a iables in college s uden s ( ¼.61). The p esen esea ch highligh s how he ex ension
o he model o he TPB by indica o s o sel -iden i y and mo i a ion de i ed om he SDT pe mi s
he cap u e o a g ea e pe cen age o he o al a iance and how i inc eases he p edic i e powe o
he exis ing p oposals in ela ion o PA in adolescen s o Spain and Es onia. Thus, he p esen s udy
e eals ha he inclusion o his pe spec i e has led o inco po a e cu en indica o s ha p edic ,
unde he media ion o he cons uc s o he TPB, he adolescen s’ u u e in ol emen in PA. The
ex ended model o he TPB, al eady sponso ed by he au ho s o he heo y (Ajzen, 1991; Fishbein
and Ajzen, 2010), explains 67%o he a iance in in en ion and 45%o he a iance in PA beha-
iou , amoun s ha exceed signi ican ly he esul s epo ed p e iously in he li e a u e (Rise e al.,
2010). In his case, i should be no ed ha he con ibu ion o sel -iden i y is no me ely anecdo al,
bu i is esponsible o a subs an ial inc emen in a iance explained in beha iou al in en ion, and
i s in luence does no occu in isola ion bu in in e ac ion wi h he basic cons uc s o he TPB
(media ed by in en ion). Consis en ly, Rise e al. (2010) p o ided he s onges e idence o da e
ha he concep o sel -iden i y is dis inc om he TPB cons uc s and showed ha he ela ion
be ween sel -iden i y and in en ion is simila o o e en su passes he s eng h o he a i ude–in en-
ion ela ion (as i happened in he p esen s udy). This leads us o he conclusion ha sel -iden i y
is an impo an p edic o o in en ions and beha iou and should he e o e be inco po a ed in o he
TPB.
Howe e , as shown in Hagge e al. (2007), he ela ion o one o he TPB cons uc s, namely
he subjec i e no ms on in en ion, was no signi ican in bo h samples o Spain and Es onia
(al hough i s ela ionship wi h mo i a ion e ealed a la ge e ec ), ques ioning o some ex en he
in eg a ion o his cons uc when conside ing addi ional explana o y a iables. The ole o sel -
iden i y in he TPB p o ides an explana ion o he ailu e o subjec i e no ms (e.g. he social
in luence componen o he TPB) as a p edic o o in en ions, as compa ed wi h a i udes and
PBC (Ajzen, 1991).
Finally, as we can see om he analysis o s uc u al in a iance, he ex ension o he o iginal
model o TPB wi h sel -iden i y o ms a common ac o s uc u e o bo h coun ies (Spain and
Es onia) as a esul o he obus ness o he s udied model (no signi ican di e ences in e ms
o chi-squa e analysis o in a iance we e ound among he wo samples) and p o ides e idence
o c oss-cul u al alidi y o i s applica ion in bo h coun ies.
330 Eu opean Physical Educa ion Re iew 18(3)