supporttf内存卡无法格式化

Runninghead:MeasurementinvarianceoftheYBRSQViladrich,C.
andAppleton,P.
andQuested,E.
andDuda,J.
andAlcaraz,S.
andHeuzé,J.
andFabra,P.
etal.
2013.
MeasurementinvarianceoftheBehaviouralRegulationinSportQuestionnairewhencompletedbyyoungathletesacrossfiveEuropeancountries.
InternationalJournalofSportandExercisePsychology.
11(4):pp.
384-394.
MeasurementInvarianceoftheYouthBehaviouralRegulationinSportQuestionnaireacrossFiveEuropeanCountriesandLanguages.
CarmeViladrich1,PaulR.
Appleton2,EleanorQuested2,JoanL.
Duda2,NikosNtoumanis2,SaülAlcaraz1,Jean-PhilippeHeuzé3,PriscilaFabra4,OddrunSamdal5,YngvarOmmundsen6,AndrewP.
Hill7,NikosZourbanos81UniversitatAutònomadeBarcelona,Barcelona,Spain2UniversityofBirmingham,Birmingham,UK3UniversitéGrenobleAlpes,Grenoble,France4UniversitatdeValencia,Valencia,Spain5UniversityofBergen,DepartmentofHealthpromotionandDevelopment,Bergen,Norway6NorwegianSchoolofSportSciences,Oslo,Norway7UniversityofLeeds,Leeds,UK8UniversityofThessaly,Thessaly,GreeceCorrespondingauthor:Prof.
NikosNtoumanisSchoolofSportandExerciseSciencesUniversityofBirmingham,Birmingham,B301LZUKTel:00441214147981Email:N.
Ntoumanis@bham.
ac.
ukRunninghead:MeasurementinvarianceoftheYBRSQMeasurementInvarianceoftheYouthBehaviouralRegulationinSportQuestionnaire(YBRSQ)acrossFiveEuropeanCountriesandLanguages.
AbstractThepurposeofthisstudywastocontributetothefactorialvalidityoftheBehaviouralRegulationSportQuestionnaire(BRSQ)byproviding(1)anadaptationoftheBRSQtotheyoungcompetitiveathletes(YBRSQ)targetedinthePAPAproject(i.
e.
,9-15yearsold)infiveEuropeancountries(France:n=1248,Greece:n=1507,Norway:n=1397,Spain:n=2245,UnitedKingdom:n=1372),and(2)evidenceofmeasurementinvarianceofitslatentfactorsacrossthesecountries.
MpluswasusedtotestforinvarianceoftheYBRSQ.
First,wetestedtheexploratorystructuralequationmodel(ESEM)factoranalyses,allowingcross-loadingsbetweenfactors,againstthetraditionalindependentclustersconfirmatoryfactoranalysismodel(ICM-CFA),withallcross-loadingsconstrainedtozero.
TheESEMshowedverygoodfitindices,whereastheICM-CFAwasnottenableacrosscountries.
Second,theESEMmodelwasusedasthebaselinemodelforthetestsoffactorloading(metric)invarianceandfactorloadingplusthresholds(scalar)invariance.
Thefivefactorsobtainedfromtheanalysiswerescalarinvariantandinterpretableacrossthefivelanguagesasintrinsicmotivation,identified,introjectedandexternalregulations,andamotivation,inlinewiththetenetsofself-determinationtheory.
Thisstudycontributestomethodologicaladvancesinsportpsychology,asitisthefirsttimeanadaptationoftheBRSQforyoungparticipantshasbeenfactoranalyzedcomparingthemoreflexibleESEMtotheusualindependentclustersfactoranalysismodelICM-CFA.
Ourdataclearlyfavourusingthemoreflexibleweakdimensionalitymodel(ESEM)andsuggestafreshinterpretationofpreviousresultsmayberequired.
Keywords:motivation,self-determinationtheory,youthsport,exploratorystructuralequationmodelling,categoricaldataThisresearchwasfundedbytheEuropeanCommissionundertheSeventhFrameworkProgram–Health-223600-aspartofthePAPAProjectwww.
projectpapa.
orgRunninghead:MeasurementinvarianceoftheYBRSQRunninghead:MeasurementinvarianceoftheYBRSQForseveraldecades,motivationhasbeenoneofthemoststudiedtopicswithinsportpsychology.
Self-determinationtheory(SDT;Deci&Ryan,2000)hasbeenprominentinprogressingconceptualisationofsportmotivationasaqualitativecontinuum.
Thatis,animportantconsiderationwhenconsideringtheroleofmotivationinsportendeavoursisthedegreetowhichmotivationregulationsforatargetedbehaviourareself-determined,controlled,orlackmotivationaltogether.
Severalstudiesattesttotheroleofself-determinedformsofmotivationforyoungpeoples'cognitions,affectandbehavioursinthesportdomain(e.
g.
,Alvarez,Balaguer,Castillo&Duda,2009;Ntoumanis,2012;Ommundsen,Lemyre,Abrahamsen,&Roberts,2010;Ommundsen,Lndal,&Loland,inpress;Sarrazin,Vallerand,Guillet,Pelletier,&Cury,2002).
Indeed,thereisevidencethatyoungathletes'qualityofmotivationaffectstheirpsychologicalwell-being(e.
g.
,Blanchard,Amiot,Perreault,Vallerand,&Provencher,2009;Gagné,Ryan,&Bargmann,2003;Smith,Ntoumanis,Duda,&Vansteenkiste,2011)aswellastheirintentiontodropoutandactualdropout(Sarrazinetal.
,2002).
Recentadvancesinunderstandingself-determinedmotivationinsporthavebeenpossibleduetothedevelopmentoftheBehaviouralRegulationinSportQuestionnaire(BRSQ)(Lonsdale,Hodge&Rose,2008).
TheBRSQisaself-reportmeasureofthreebroadtypesofmotivationadvancedbySDT(Deci&Ryan,2002).
AccordingtoDeciandRyan(1985,2000)motivationregulationsthatunderpinbehaviouralengagementinactivitiessuchassportareconsideredtolieonacontinuum.
Thiscontinuumreflectsvariationsinthedegreetowhichthepurposeofbehaviouralengagementisinternalised.
Intrinsicmotivationdescribesthemotivationthatliesatthemostautonomousendofthecontinuum.
Intrinsicallymotivatedbehavioursareundertakenforreasonssuchastheinherentenjoyment,interestandsatisfactionderivedfromengagementinthebehaviour,andotherreasonsthatmightbeconsideredauthentictotheindividual.
Incontrast,whenthemotivationtoengageinsportisfullyorpartiallyderivedfromsomethingorsomeoneseparablefromtaskparticipation,theregulationisdescribedasextrinsic.
However,thedegreetowhichthebehaviour'svalueandpurposehasbeeninternalisedmayvary(Ryan&Deci,2000).
AccordingtoSDT,thisvariabilityininternalisationcanbecategorisedintofourtypesofextrinsicmotivation,whicharelabelledasintegrated,identified,introjectedandextrinsic.
DeciandRyan(1985,Runninghead:MeasurementinvarianceoftheYBRSQ2000)conceptualiseintegratedregulationasthemostself-determinedoftheextrinsicregulations,lyingclosesttointrinsicmotivationonthecontinuum.
Integratedregulationdescribeswhentheathlete'sbehaviouralengagementisbroughtintocongruencewithpersonallyendorsedneeds,valuesandgoals.
Nextonthecontinuumisidentifiedregulation.
Whenanathletefullyaccepts,identifieswith,andpersonallyendorsesthereasonstoperformthebehaviour,engagementisdescribedasidentified(Ryan&Deci,2000).
Incontrasttointegratedandidentifiedregulationswhichareconsideredautonomousformsofextrinsicmotivation,introjectedbehavioursareregulatedbypsychologicalcontingenciesandthusrepresentacontrolledformofextrinsicmotivation.
Introjectedbehavioursareundertakeninordertoavoidundesirablepsychologicalresponsessuchasguiltorshame,orbecauseexperiencingdesirablepsychologicalresponses(suchasself-worth)aredependentonperformingthebehaviour.
Whensportingbehavioursaremotivatedbyintrojectedregulationsoneislikelytofeelthatone"should"playortrain.
Atthefarendofthecontinuumfromintrinsicmotivationliestheleastautonomousoftheextrinsicregulations,labelledasexternalmotivation.
Externallyregulatedbehavioursaremotivatedbyreasonsthathavenotbeeninternalisedatall.
Thismaybetoavoidpunishmentortogainpraiseorrewards.
Lastly,amotivationessentiallydescribesalackofmotivation.
Amotivatedactionsarepassiveandlackanyintentionalaim(Deci&Ryan,1991).
Amotivatedplayerscannotidentifyanygoodreasontocontinuetheirtrainingandarelikelytostopengagingintheirsport.
ThesixmotivationsproposedwithinSDTareconsideredtoexistalongacontinuumrepresentingincreasinglevelsofautonomyandself-determination.
QuestionnairessuchastheBRSQaredesignedtomeasurethedegreetowhichthesebehaviouralregulationsunderpinsportparticipation.
Individualsscoringhigheronthesubscalesmeasuringtheregulationsattherelativelyself-determinedendofthiscontinuum(i.
e.
,intrinsic,identified,integrated)areexpectedtoscorelowerattheotherendofthecontinuum(i.
e.
,introjected,externalandamotivated)andviceversa.
Moreover,theinter-correlationsbetweenadjacentmotivationregulationsalongthecontinuum(e.
g.
,introjectedandexternal)areexpectedtobehigherthantheassociationswiththeregulationsthataremoredistal(e.
g.
,introjectedandintrinsic),thuscontributingtoaquasi-simplexpattern(Ryan&Connell,1989).
DeciandRyan(1985)andRyanandRunninghead:MeasurementinvarianceoftheYBRSQDeci(2007)proposedthatmoreself-determinedmotivationregulationsarerelatedtopositiveoutcomes(e.
g.
,enjoyment,vitality),whereasmotivationslowerinself-determinationrelatetomaladaptiveoutcomes(e.
g.
,anxiety,intentiontodrop-out)(forasummary,seeNtoumanis,2012).
InitialstudiesontheBRSQshowevidenceoffactorvalidity,internalreliability,andtest-retestreliabilityineliteandnon-elitesamplesofathletes(Lonsdaleetal.
,2008;Lonsdale,Hodge,&Rose,2009).
Lonsdaleandcolleagues(2008)alsoreportedthatfactorvalidityandinternalconsistencyscoresoftheBRSQwereatleastequal,orinsomecasessuperior,tootherself-determination-relatedmeasuresofmotivationinsport(i.
e.
,SMS,SMS-6).
ThereisalsoevidenceforthenomologicalvalidityoftheBRSQrelatedtoself-reportedmeasuresofflowandathleteburnout(Lonsdaleetal.
,2008,2009).
Sinceitspublication,theBRSQhasbeentranslatedtoDutch(Assor,Vansteenkinste,&Kaplan,2009),Greek(Mouratidis,Lens,&Vansteenkiste,2010),Spanish(Viladrich,Torregrosa,&Cruz,2011)andChinese(Chan,Hagger,&Spray,2011)amongotherlanguages(Iranian,JapaneseandSwedish,C.
Lonsdale,personalcommunication,May30,2012),sometimeswithamendmentstothescale.
Despitetheseadvancesintheuseofthescale,somepsychometricissuesremain.
Indatacollectedwithyoungandadultathletes(i.
e.
,from12to58yearsofage,meanrangingfrom14.
04to25.
9years)supportforthediscriminantvaliditybetweentheexternalandintrojectedregulationsandbetweentheidentifiedandintegratedregulationswasinconsistent.
Someresultsshowedlackofdiscriminationorpoorfitofthemodeltothedata(Londsdaleetal.
,2008,2009;Holland,Woodcock,Cumming&Duda,2010;Mouratidisetal.
,2010),whereassomeothersshowedgooddiscriminantvalidity(Assoretal.
,2009;Viladrichetal.
,2011).
Moreover,theBRSQ'sdevelopmentwasorientedtowardyoungadultsportparticipants(i.
e.
,itemsmeaningfulforcompetitiveathletesof20yearsold)and,consequently,itisnotwelldocumentedwhethertheBRSQisanage-appropriatemeasureofyoungerathletes'motivation.
Thus,thedegreetowhichyoungerparticipantsareabletounderstandtheBRSQitemsneedsfurtherconsideration.
Therefore,themainpurposeofthispresentstudywastofurthercontributetotheinstrument'sfactorialvaliditybyproviding(1)anadaptationoftheBRSQtotheyoungcompetitiveathletestargetedinRunninghead:MeasurementinvarianceoftheYBRSQthePAPAproject(i.
e.
,9-15yearsold)infivedifferentEuropeancountries(i.
e.
,France,Greece,Norway,Spain,andUnitedKingdom),and(2)evidenceofmeasurementinvarianceofitslatentfactorsacrossthesecountries.
MethodParticipantsParticipated7769footballplayers(13%female)agedbetween9and15years(mean=11.
76,SD=1.
42)fromfiveEuropeancountries(France=1248,Greece=1507,Norway=1397,Spain=2245,UnitedKingdom=1372).
Theparticipantsbelongedto19regionsinthesecountries,playedfor619teamsatgrassrootslevel,reportedtohavebeenplayingfortheirteamsbetweenthepresentseasonandthelast10seasons(median=3seasons,interquartilerange=4),andtotrainbetween0.
5and10hoursaweek(median=4,interquartilerange=1.
5).
ThegreatestproportionofgirlswasincludedintheNorwegiansample(42%),modestproportionsofgirlscamefromtheUnitedKingdom(14%)andSpanish(9%)samples,whereasveryfewgirlswerepresentinFrench(3%)andGreek(1.
6%)samples.
MeasuresTheYouthBehaviouralRegulationinSportQuestionnaire(YBRSQ)wasinitiallydevelopedfromtheBRSQ-6(Lonsdaleetal.
,2008)inEnglish.
Basedon(1)theresultsfromtheliteratureindicatingthattheidentifiedandintegrationregulationsubscalesdidnotseparateclearly(Lonsdaleetal.
,2008;Mouratidisetal.
,2010)and(2)Lonsdaleetal's(2008)suggestionthataquestionnaireformatwasnotwellsuitedforassessingintegratedregulation,and(3)Vallerand's(1997)assertionthatthistypeofmotivewasnotprevalentuntiladulthood,theintegratedregulationsubscalewasexcludedfromtheYBRSQ.
The20remainingitemsfromtheBRSQ-6werescrutinizedbyacommitteeofresearchersfamiliarwithboththeSDTframeworkandthedevelopmentofquestionnairesforyoungparticipants.
Someitemswererewordedtomakethereadinglevelmoreappropriateforchildrenandadolescents(e.
g.
,theitem"butthereasonswhyarenotcleartomeanymore"tappingamotivation,wasrewordedas"butIreallydon'tknowwhyanymore";theitem"becausethebenefitsofsportareimportanttome"tappingidentifiedregulationwaspresentedwiththeadditionofexamples(e.
g.
,"developingasaplayer,gettingRunninghead:MeasurementinvarianceoftheYBRSQfit,playingwithmyteammates").
Thefinalresultwasa20-itemquestionnairemeasuringamotivation(4items),external(4items),introjected(4items)andidentified(4items)intrinsic(4items)regulations.
Theresponsescalewasmodifiedfromseventofivepoints(1=stronglydisagree,5=stronglyagree)andtheorderofitemswerealsoadapted.
Theitemswereprecededwiththeinstructions:"Indicatehowwelleachofthereasonsbelowreflectswhyyouplayfootballforthisteam",andthestem:"Iplayfootballforthisteam…".
TheYBRSQwastranslatedintoFrench,Greek,Norwegian,andSpanishtakingintoaccounttheguidelinesfromtheInternationalTestCommission(Hambleton,2005;seeDudaetal,thisedition).
Thefullscaleinthefivelanguagesisavailablefromthecorrespondingauthoruponrequest.
ProcedureThePAPAprojectprocedureprotocol(seeDudaetal,thisedition)informeddatacollectionsecuringastandardisedsetofproceduresfollowedinallcountries.
Alldatautilisedinthisstudywerecollectedduringthefirstpartofthesportseason,inspring2011inNorwayandbetweenfallandwinter2011/12intheothercountries.
DatamanagementandanalysisYBRSQdatawerecleanedandscreenedforpatternsofmissingvaluespriortothecreationofaninternationalfileformainanalysis.
Inordertotestformeasurementinvariance,Mplus(Muthén&Muthén,1998-2012)wasused.
WefollowedtherecommendationsproposedbyMillsapandYun-Tein(2004)forcategoricalvariables,andMarsh,NagengastandMorin(2012)forexploratorystructuralequationmodels(ESEM).
Thisprocessinvolvedtestingfournestedmodels.
FirstwetestedtheESEMmodel(model1)againsttheindependentclustersfactoranalysismodel(ICM-CFA,model2).
TheICM-CFAstandsforthetraditionalconfirmatoryfactoranalysis,whichallowsaresearchertotestthehypothesisthataspecificnumberoffactorsarerepresentedbycertainindicators,withallcross-loadingsconstrainedtozero(Skrondal&Rabe-Hesketh,2004).
ESEMrelaxesthisrestrictionandfactorloadingsinallfactorsareestimatedforeachitem,obtainingparameterestimatessimilartothoseobtainedbyclassicalexploratoryfactoranalysisplusthestandarderrorsandgoodnessoffitindicesusuallyassociatedtoCFA(Asparouhov&Muthén,2009).
Therelaxedfactorcross-loadingsarethereasonwhyESEMisRunninghead:MeasurementinvarianceoftheYBRSQalsocalledweakdimensionalitymodelwhereastheICM-CFAiscalledstrongdimensionalitymodel.
Thecomparisonofbothmodelsprovidesatestofthehypothesisaboutcross-loadingsbeingzeroornon-zero(Skrondal&Rabe-Hesketh,2004).
InordertoidentifytheESEMsolution,adecisionshouldbetakenamongavarietyofpossiblerotations.
DuetotheprevioustheoreticalandempiricalknowledgeabouttheinternalstructureoftheBRSQ(Lonsdaleetal.
,2008),wedeemedthetargetrotation(Asparouhov&Muthén,2009;Marshetal.
,inpress)tobethemostappropriate.
Targetrotationestimatesallcross-loadingswiththerestrictionthattheirvaluesareascloseaspossibletozero,whichistheclosestoptionofESEMtotheICM-CFA.
IrrespectiveofESEMortheICM-CFA,themodelthatshowedbetterfitinthefirststepwastakenastheconfiguralmodelandwasusedasthebaselinemodelforthetestsoffactorloading[analogoustoitemdiscriminationparameters]invariance(model3)andfactorloadingplusthresholds[itemlocationparametersanalogoustocategorydifficultyparameters]invariance(model4).
Inotherwords,wetestedconfigural(models1and2),metric(model3)andscalar(model4)measurementinvarianceofthefivelatentfactorsmeasuredbytheYBRSQ.
Duetothecategoricalnatureofthedataandthepresenceofmissingvalues(seeResultssection),theweightedleastsquaresmeanandvarianceadjusted(WLSMV)estimatorwasusedwithpairwisedeletionformissingvalues,bothofthembeingtheMplusdefaultsforcategoricaldata.
WedeemedthisapproachtobeappropriatebasedonGraham(2009)andAsparouhovandMuthén(2010),whojudgethatwhenthepercentageofmissingdataislow,thebiasesandlossofpowerattributabletopairwisedeletioncanbeconsideredinconsequentialandbetterthanusinglistwisedeletionofcases.
Thegoodnessoffitindiceswere2,comparativefitindex(CFI),Tucker-LewisIndex(TLI),androotmeansquareerrorofapproximation(RMSEA).
InanICM-CFAanalysiswithquantitativeindicators,CFIandTLIvalues>.
95andRMSEA.
90andRMSEA.
96forcategoricaldataandmostpapersusingESEM,Runninghead:MeasurementinvarianceoftheYBRSQincludingcategoricaldataESEM(e.
g.
,Myers,Chase,Pierce,&Martin,2011)relyonthemwithsomecaution.
Alltheserecommendationswereconsideredinthispaper.
Inordertocomparenestedmodels,twoindicatorswereevaluated;thechi-squaredifferenceascomputedinMplusforcategoricalvariables,andthedifferenceinCFI(ΔCFI).
Asacut-offvalue,aΔCFI<.
01(Cheung&Rensvold,2002)wasconsideredthemoreappropriateevidenceofincreasinginvariance,duetothesensitivityofchi-squaretothesamplesize.
Standarderrorsandgoodnessoffitindexeswerecalculatedtakingintoaccountthatplayers'responseswereclusteredwithintheirteams.
ResultsTheYBRSQshowed2.
5%ofmissingresponses,with236differentpatternsofmissingdata.
Theitemdistributions(seeTable1)showedsizeablefloororceilingeffectsacrosscountries,withskewnessbeingbetween0.
39and2.
99inabsolutevalue,andkurtosisbetween0.
20and10.
09inabsolutevalue.
TheitemdistributionsshowninTable1justifythetreatmentofdataascategorical.
Besidestheirskewnessandkurtosis,theitemsshowedaconsistentpatternofresultsacrosscountries(notdisplayed,butavailableuponrequest),withlowvaluesforamotivationandexternalregulationitems,highvaluesforidentifiedregulationandintrinsicmotivationitemsandintermediatevaluesforintrojectedregulation.
GoodnessoffitstatisticsforallthemeasurementinvarianceanalysescanbeseeninTable2.
Exceptfor2values,theweakdimensionalitymodel(ESEM)showsverygoodindexesoffit(CFI=.
987,TLI=.
976,RMSEA=.
035,CI95%=0.
033÷0.
037),whereasthestrongdimensionalitymodel(ICM-CFA)wasnottenableacrosscountries(CFI=.
892,TLI=.
872,RMSEA=.
080,CI95%=0.
078÷0.
081,ΔCFI=-.
105).
Thisresultwasreplicatedinalldataanalysesconductedseparatelybycountry(resultsnotshown).
Thus,theweakdimensionalitymodel(model1)wastestedforsubsequentinvarianceacrosscountries.
Resultsshowedgoodindexesoffitbothforfactorloadingsinvariance(model3,CFI=.
972,TLI=.
967,RMSEA=.
041,CI95%=0.
039÷0.
042)andfactorloadingsplusthresholdsinvariance(model4,CFI=.
964,TLI=.
967,RMSEA=.
041,CI95%=0.
039÷0.
042)favouringtheconclusionofstrongmeasurementinvarianceacrosscountries.
ModelcomparisonindexeswereonlymarginallyfavourabletothisconclusionduetotheΔCFIvalueof-.
015whencomparingconfiguralinvarianceversusfactorRunninghead:MeasurementinvarianceoftheYBRSQloadingsinvariance.
Modificationindexessuggestedthattheitem"BecauseIfeelImustcontinue",thethirditemtappingintrojectedregulation(IN3),contributedlargelytothisvalue,butwedecidednottorunapartialinvariancebecausethemixtureofinvariantandnotinvariantfactorloadingsdonothaveaclearinterpretationintheESEMcontext(Millsap,2011).
Thisissueiselaborateduponinthediscussionsection.
EstimatedfactorloadingsandcorrelationsbetweenfactorsaredisplayedinTable3.
Thepatternismostlycoherentwiththeexpectations,andfactorloadingsexpectedtobehigh,aregenerallyhigherthanthecross-loadingsexpectedtobelow.
WiththesoleexceptionofitemIN3fromtheintrojectedregulationfactor,allindicatorsshowedfactorloadingshigherthan.
35ontheirintendedfactors.
Moreover,eachitem'sfactorloadingonitsintendedfactorwashigherthanitsestimatedcross-loadingonnon-intendedfactors,andthesecross-loadingswerebelow.
30inabsolutevaluewithonlytwoexceptions(i.
e.
,IN4tappingintrojectedregulation,andID1tappingidentifiedregulation,bothcross-loadingintotheexternalregulationfactor).
ItemIN3cross-loadedontotheidentifiedandexternalregulationfactors.
Lessnoteworthy,theexternalregulationfactorshowedthelowestfactorloadingsinthiscontext(from.
393to.
512).
Thevaluesofthecorrelationbetweenfactors,rangingfrom-.
34to.
62,provideevidenceofgooddiscriminantvaliditybetweenthemeasuredfivefactors.
Moreover,theirpattern,changingfromhigherpositivevaluesforadjacentconstructstonegativevaluesfordistalconstructs,iscompatiblewiththesimplex-likehypothesisofSDT.
DiscussionThepurposeofthepresentstudywastofurthercontributetothedevelopmentandvalidationprocessoftheBRSQ(Lonsdaleetal.
,2008)byadaptingthisinstrumenttoyouthsportsparticipants(i.
e.
,9-15yearsold)andexploringmeasurementinvarianceacrossfiveEuropeancountries.
ESEManalysesindicatedthatthefiveYBRSQsubscalesshowedmetricinvarianceacrosssamplesofyoungEuropeanathletes.
InspiteofusingESEM–anestimationmethodwhichallowscross-loadingsbetweenfactors–thefivefactorsobtainedfromtheanalysisareinterpretableasintrinsicmotivation,identified,introjectedandexternalregulations,andamotivation,asmostoftheitemstappingtheseconceptsloadedclearlyonRunninghead:MeasurementinvarianceoftheYBRSQcontent-coherentfactors.
TheexceptionistheitemIN3fromtheintrojectedregulationsubscale,whichdemonstratedsizeablecross-loadingswiththeexternalregulationfactorandtheidentifiedregulationfactor.
Furthermore,thecorrelationsbetweenthesefactorsshowedgooddiscriminantpower(anadvantageofESEMisthatitlowersfactorcorrelationsbyallowingcross-loadings)andproducedasimplex-likestructureasproposedbySDT.
Collectively,theresultsfromthisstudyprovideevidencethatthefivesubscalesfromtheYBRSQ,withtheexceptionofoneitem(IN3,IfeelImustcontinue),representagoodoperationalizationofthecontinuumofbehaviouralregulations,ashypothesizedbythetenetsofSDTacrossfiveofthelanguagesspokeninfiveEuropeancountries.
AccordingtoSDT,themeasuredfactorswereexpectedtobeidentifiableandorderedonacontinuumfromintrinsicmotivationtoamotivation(Ryan&Connell,1989);thisassumptionwassupportedinthisstudy.
ThisstudyalsocontributestomethodologicaladvancesinsportpsychologyasitisthefirsttimeanadaptationoftheBRSQforyoungparticipantshasbeenfactoranalyzedcomparingthemoreflexibleESEMtotheusualICM-CFAfactoranalysis.
Ourdataclearlyfavourusingthemoreflexibleweakdimensionalitymodelandsuggestafreshinterpretationofpreviousresultsmayberequired.
ThefactoranalysespublishedtodateontheBRSQhavebeenICM-CFAbased,andhaveshowedacceptablefitindexesandgoodfactorinterpretability(Assoretal.
,2009;Lonsdaleetal.
,2008,2009;Mouratidisetal.
,2010;Viladrichetal.
,2011).
However,theresultsfromICM-CFAhavesometimesfailedtosupportthediscriminantvaliditybetweenexternalandintrojectedregulationsandbetweenidentifiedandintegratedregulations(Lonsdaleetal.
,2008;Mouratidisetal.
,2010).
Theimplicationsofalackofdiscriminantvalidityisthattestsofprocessmodelsincludingmotivationregulationswillbelimitedtomoregeneralfactorsofautonomousversuscontrolledmotivation,suchastheanalysisincludedinChanetal.
(2011),orasingleself-determinationindexconstructedbyaggregatingscoresofallfactors(Vallerand,2001).
TheresultspresentedinthispapersuggestthelackofdiscriminantvalidityidentifiedinpreviousstudiescouldbeattributabletothefactthattheICM-CFAmodelusuallyproducesoverestimatedcorrelationsbetweenfactorsasaconsequenceoffactorcross-loadingsbeingunrealisticallyfixedtozero(AsparouhovRunninghead:MeasurementinvarianceoftheYBRSQ&Muthén,2009;Marshetal.
,2012).
Inthisstudy,themorerealisticESEManalysiswithtargetrotationshowedthatitemcross-loadingscouldbeanalternativeexplanationforthelackofdiscriminantvalidityfoundinpreviousresearch.
Inoursamples,itemIN3wasnotinterpretedasanindicatorofthesamefactorasotherintrojecteditems,suggestingthatendorsingtheitem"IfeelImustcontinue"couldbeduetoasenseofguilt,contingentself-worth(asperintrojection),butitmightalsobebecauseparticipatinginsportispersonallyimportanttooneself(asperidentifiedregulation),orbecauseonewantstoreceivepraisefromothers(asperexternalregulation).
Thesealternativeinterpretationsofthisitemarenotage-dependentandthusraisequestionsaboutthevalidityofthisitemintheoriginalBRSQ.
DespitetheadvantagesofferedbyESEM,onelimitationofthisanalyticalapproachisthedifficultytomanagethedifferentialitemfunctioningattributabletopartialnon-invarianceoffactorloadings.
Inouranalysis,weacceptedmarginalstatisticalgainanddecidednottorunpartialinvarianceanalysisinspiteofthefactthatsomeindicatorssuggestedthatitemIN3couldhavenotperfectlyinvariantfactorloadingsandcross-loadingsacrosscountries.
Myersetal.
(2011)discussedasimilardecisioninanESEManalysisusinggeominrotation.
Alignedwiththeirrationale,weproposethatinthecontextofanESEMmodelwithtargetrotation,itdoesnotmakesensetofreelyestimateforeachgrouponefactorloadingthatwesuspectnottobeinvariant.
Thisisbecauseallfactorloadingstargetedtozeroaresimultaneouslycalculatedaccordingtotherotationcriterionestablished,thustheyarenotindependentfromeachotherandfreeingoneofthemacrossgroupsinduceschangesintheestimationofallothertargetedparameters.
Inouropinion,thisiscompatiblewithMillsap's(2011)warningthatsomecasesofpartialinvarianceareverydifficulttointerpretinanESEMcontextandalsowiththepositiontakenbyAsparouhovandMuthén(2009)thatinthisframework,onlyfullinvariancecanbeestimated.
ConclusionsOverall,thefindingsfromthisstudyprovideinitialevidencefortheYBRSQasameasureofmotivationregulationsandshouldallowfutureresearchtobeconductedonthemotivationofchildrenandadolescentsparticipatinginsportfromanSDTperspective.
Forexample,theYBRSQcouldbeusedtoRunninghead:MeasurementinvarianceoftheYBRSQassesschangesinchildren'ssportmotivationovertimeininterventionstudiessuchasthePAPAproject.
FutureresearchcanalsonowtestprocessmodelspositedbySDTthatincludethefiveregulationsmeasuredbytheYBRSQ,andsuchresearchwouldalsocontributetothenomologicalvalidityofthequestionnaire.
FurtherevidenceofpsychometricpropertiesofYBRSQinEuropeanyouthathletescouldalsobeobtainedbycheckingforstabilityovertimeandcomparingfactormeansacrosscountriescontrollingforsocio-demographicvariablessuchasgender,ageandyearsplaying.
Inadditiontoforwardingasport-specificmeasureofmotivationregulationsforchildrenandadolescents,thepresentstudyalsocontributestomethodologicaladvancesinsportpsychology.
ThefindingsfromthisstudydemonstratetheconsiderablepromisefortheapplicationofESEMformotivationregulationsscalesinsport,andsuggestafreshinterpretationofpreviousresultsfromstudiesinthisareathathavereliedonthemoretraditionalICM-CFA.
Runninghead:MeasurementinvarianceoftheYBRSQReferenceslvarez,M.
S.
,Balaguer,I.
,Castillo,I.
,&Duda,J.
L.
(2009).
Coachautonomysupportandqualityofsportengagementinyoungsoccerplayers.
TheSpanishJournalofPsychology,12,138-148.
Asparouhov,T.
,&Muthén,B.
(2009).
Exploratorystructuralequationmodeling.
StructuralEquationModeling,16,397-438.
doi:10.
1080/10705510903008204.
Asparouhov,T.
,&Muthén,B.
(2010).
Weightedleastsquaresestimationwithmissingdata.
Retrievedfromhttp://www.
statmodel.
com/download/GstrucMissingRevision.
pdfAssor,A.
,Vansteenkinste,M.
,&Kaplan,A.
(2009).
Identifiedversusintrojectedapproachandintrojectedavoidancemotivationsinschoolandinsports:Thelimitedbenefitsofself-worthstrivings.
JournalofEducationalPsychology,101,482-497.
doi:10.
1037/a0014236Blanchard,C.
M.
,Amiot,C.
E.
,Perreault,S.
,Vallerand,R.
J.
,&Provencher,P.
(2009).
Cohesiveness,coach'sinterpersonalstyleandpsychologicalneeds:Theireffectsonself-determinationandathletes'subjectivewell-being.
PsychologyofSportandExercise,10,545-551.
doi:10.
1016/j.
psychsport.
2009.
02.
005Chan,D.
K.
C.
,Hagger,M.
S.
,&Spray,C.
M.
(2011).
Treatmentmotivationforrehabilitationafterasportinjury:Applicationofthetrans-contextualmodel.
PsychologyofSportandExercise,12,83-92.
doi:10.
1016/j.
psychsport.
2010.
08.
005Cheung,G.
W.
,&Rensvold,R.
B.
(2002).
Evaluatinggoodness-of-fitindexesfortestingmeasurementinvariance.
StructuralEquationModeling,9,233-255.
doi:10.
1207/S15328007SEM0902_5Deci,E.
L.
,&Ryan,R.
M.
(1985).
Intrinsicmotivationandself-determinationinhumanbehavior.
NewYork,NY:PlenumPress.
Deci,E.
L.
,&Ryan,R.
M.
(1991).
Amotivationalapproachtoself:integrationinpersonality.
InR.
Deinstbier(Ed.
),NebraskaSymposiumonMotivation:Vol.
38.
PerspectivesonMotivationVolume38.
(pp.
237-88).
Lincoln,NE:UniversityofNebraskaPress.
Runninghead:MeasurementinvarianceoftheYBRSQDeci,E.
L.
,&Ryan,R.
M.
(2000).
The"what"and"why"ofgoalpursuits:Humanneedsandtheself-determinationofbehavior.
PsychologicalInquiry,11,227-68.
Deci,E.
L.
,&Ryan,R.
M.
(2002).
Handbookofself-determinationresearch.
Rochester,NY:UniversityofRochesterPress.
Gagné,M.
,Ryan,R.
M.
,&Bargmann,K.
(2003).
Autonomysupportandneedsatisfactioninthemotivationandwell-beingofgymnasts.
JournalofAppliedSportPsychology,15,372-390.
doi:10.
1080/10413200390238031Graham,J.
W.
(2009).
Missingdataanalysis:Makingitworkintherealworld.
AnnualReviewofPsychology,60,549-576.
doi:10.
1146/annurev.
psych.
58.
110405.
085530Hambleton,R.
K.
(2005).
Issues,designsandtechnicalguidelinesforadaptingtestsintomultiplelanguagesandcultures.
InR.
K.
Hambleton,P.
F.
Merenda,&C.
D.
Spielberger(Eds.
),Adaptingpsychologicalandeducationaltestsforcross-culturalassessment(pp.
3-38).
Mahwah,NJ:LawrenceErlbaum.
Holland,M.
J.
G.
,Sharp,L.
,Woodcock,C.
,Cumming,J.
,&Duda,J.
L.
(2010).
ValidityandreliabilityoftheBehavioralRegulationinSportQuestionnaire(BRSQ)withyouthathletes.
JournalofSport&ExercisePsychology,32,s175-s176.
Hu,L.
T.
,&Bentler,P.
M.
(1999).
Cutoffcriteriaforfitindexesincovariancestructureanalysis:Conventionalcriteriaversusnewalternatives.
StructuralEquationModeling,6,1-55.
doi:10.
1080/10705519909540118Lonsdale,C.
,Hodge,K.
,&Rose,E.
A.
(2008).
TheBehaviouralRegulationinSportQuestionnaire(BRSQ):Instrumentdevelopmentandinitialvalidityevidence.
JournalofSportandExercisePsychology,30,323-355.
Lonsdale,C.
,Hodge,K.
,&Rose,E.
(2009).
Athleteburnoutinelitesport:Aself-determinationperspective.
JournalofSportsSciences,27,785-795.
doi:10.
1080/02640410902929366Marsh,H.
W.
,Hau,K.
-T.
,&Wen,Z.
(2004).
Insearchofgoldenrules:Commentonhypothesis-testingapproachestosettingcutoffvaluesforfitindexesanddangersinovergeneralisingHuandRunninghead:MeasurementinvarianceoftheYBRSQBentler's(1999)findings.
StructuralEquationModeling,11,320-341.
doi:10.
1207/s15328007sem1103_2Marsh,H.
W.
,Lüdtke,O.
,Nagengast,B.
,Morin,A.
J.
S.
,Trautwein,U.
,&VonDavier,M.
(inpress).
Whyitemparcelsare(almost)neverappropriate:Twowrongsdonotmakearight:Camouflagingmisspecificationwithitem-parcelsinCFAmodels.
PsychologicalMethods.
Marsh,H.
W.
,Nagengast,B.
,&Morin,A.
J.
(2012).
Measurementinvarianceofbig-fivefactorsoverthelifespan:ESEMtestsofgender,age,plasticity,maturity,andladolcevitaeffects.
DevelopmentalPsychology.
Advanceonlinepublication.
doi:10.
1037/a0026913.
Millsap,R.
E.
(2011).
Statisticalapproachestomeasurementinvariance.
NewYork,NY:Routledge.
Millsap,R.
E.
,&Yun-Tein,J.
(2004)Assessingfactorialinvarianceinordered-categoricalmeasures.
MultivariateBehavioralResearch,39,479-515.
doi:10.
1207/S15327906MBR3903_4Mouratidis,A.
,Lens,W.
,&Vansteenkiste,M.
(2010).
Feedbackmakesadifference:Themotivatingroleofcommunicatinginanautonomy-supportingway.
JournalofSport&ExercisePsychology,32,619-637.
Muthén,L.
K.
,&Muthén,B.
,O.
(1998-2012).
Mplususer'sguide(6thed.
).
LosAngeles,CA:Author.
Myers,N.
D.
,Chase,M.
A.
,Pierce,S.
W.
,&Martin,E.
(2011).
Coachingefficacyandexploratorystructuralequationmodeling:Asubstantive-methodologicalsynergy.
JournalofSport&ExercisePsychology,33,779-806.
Ntoumanis,N.
(2012).
Aself-determinationtheoryperspectiveonmotivationinsportandphysicaleducation:Currenttrendsandpossiblefutureresearchdirections.
InG.
C.
Roberts&D.
C.
Treasure(Eds).
Motivationinsportandexercise:Volume3(pp.
91-128).
Champaign,IL:HumanKinetics.
Ommundsen,Y.
,Lemyre,P.
N.
,Abrahamsen,F.
,&Roberts,G.
C.
(2010).
Motivationalclimate,needsatisfaction,regulationofmotivationandsubjectivevitalityinsoccer.
Astudyofyoungplayers.
InternationalJournalofSportPsychology,41,216-242.
Runninghead:MeasurementinvarianceoftheYBRSQOmmundsen,Y.
,Lndal,K,&Loland,S.
(inpress).
Sport,childrenandwell-being.
InBen-Arieh,Asher,Casas,Ferran,Frones,Ivar.
andKorbin,JillE.
(Eds.
),HandbookofChildWell-Being.
Theories,MethodsandPoliciesinGlobalPerspective.
Dorcrecht:Springer.
Ryan,R.
M.
,&Connell,J.
P.
(1989).
Perceivedlocusofcausalityandinternalization:Examiningreasonsforactingintwodomains.
JournalofPersonalityandSocialPsychology,57,749-761.
Ryan,R.
M.
,&Deci,E.
L.
(2000).
Self-determinationtheoryandthefacilitationofintrinsicmotivation,socialdevelopment,andwell-being.
AmericanPsychologist,55,68-78.
Ryan,R.
M.
,&Deci,E.
L.
(2002).
Anoverviewofself-determinationtheory:Anorganismic-dialectalperspective.
InE.
L.
Deci&R.
M.
Ryan(Eds.
),Handbookofself-determinationresearch(pp.
3-33).
Rochester,NY:TheUniversityofRochesterPress.
Ryan,R.
M.
,&Deci,E.
(2007).
Activehumannature:Self-determinationtheoryandthepromotion,andmaintenanceofsport,exerciseandhealth.
InM.
S.
Hagger&N.
L.
D.
Chatzisarantis(Eds.
),Intrinsicmotivationandself-determinationinsportandexercise(pp.
1-22).
Champaign,IL:HumanKinetics.
Sarrazin,P.
,Vallerand,R.
,Guillet,E.
,Pelletier,L.
,&Cury,F.
(2002).
Motivationanddropoutinfemalehandballers:A21-monthprospectivestudy.
EuropeanJournalofSocialPsychology,32,395-418.
doi:10.
1002/ejsp.
98.
Skrondal,A.
,&Rabe-Hesket,S.
(2004).
Generalizedlatentvariablemodeling:Multilevel,longitudinal,andstructuralequationmodels.
London:Chapman&Hall/CRC.
Smith,A.
,Ntoumanis,N.
,Duda,J.
L.
,&Vansteenkiste,M.
(2011).
Goalstriving,coping,andwell-beinginsport:Aprospectiveinvestigationoftheself-concordancemodel.
JournalofSport&ExercisePsychology,33,124-145.
Vallerand,R.
J.
(1997).
Towardahierarchicalmodelofintrinsicandextrinsicmotivation.
InM.
P.
Zanna(Ed.
),Advancesinexperimentalsocialpsychology,Vol.
29(pp.
271-360).
NewYork:AcademicPress.
Runninghead:MeasurementinvarianceoftheYBRSQVallerand,R.
J.
(2001).
AHierarchicalModelofIntrinsicMotivation.
InG.
C.
Roberts(Ed.
),Advancesinmotivationinsportandexercise(pp.
263-319).
Leeds:HumanKinetics.
Viladrich,C.
,Torregrosa,M.
,&Cruz,J.
(2011).
CalidadpsicométricadelaadaptaciónespaoladelCuestionariodeRegulaciónConductualenelDeporte.
Psicothema,23,786-794.
Yu,C.
-Y.
(2002).
Evaluatingcut-offcriteriaofmodelfitindicesforlatentvariablemodelswithbinaryandcontinuousoutcomes(Doctoraldissertation,UniversityofCalifornia).
Retrievedfromhttp://statmodel2.
com/download/Yudissertation.
pdfRunninghead:MeasurementinvarianceoftheYBRSQTable1.
ItemDistributionalCharacteristicsofBRSQ.
N%floor%ceilingMSDSkKAM1questionwhyIcontinue72436261.
781.
1991.
4691.
046AM2questionwhyIplay75586461.
711.
1561.
6611.
765AM3don'tknowwhyanymore75235961.
791.
1711.
4291.
062AM4wonderwhat'sthepoint76046251.
731.
1211.
5241.
411EX1pushme75916751.
671.
1431.
6861.
782EX2satisfypeople760544102.
221.
3560.
787-0.
640EX3pressurefromothers75946061.
811.
2161.
4040.
830Ex4otherswillnotbepleased76105561.
871.
1851.
2710.
606IN1guilty756435152.
561.
4590.
415-1.
203IN2ashamedifquit760752102.
041.
3461.
054-0.
200IN3mustcontinue759020333.
351.
520-0.
388-1.
309IN4feellikeafailureifquit758941142.
381.
4550.
641-0.
994ID1benefitsimportant75822564.
350.
895-1.
5452.
401ID2valuebenefits76162594.
400.
882-1.
6552.
768ID3selfdiscipline75757353.
751.
216-0.
736-0.
334ID4learnusefulthings75786383.
881.
140-0.
8770.
081IM1enjoy76092744.
610.
794-2.
6167.
506IM2likeit76221794.
690.
711-2.
98610.
086IM3fun76003644.
430.
950-1.
9023.
353IM4exciting76094514.
171.
076-1.
3841.
350Note:%floor:percentageofplayerswhopickedcategory1(minimum);%ceiling:percentageofplayerswhopickedcategory5(maximum);Sk:Skewness;K:KurtosisRunninghead:MeasurementinvarianceoftheYBRSQTable2.
GoodnessofFitIndexesfortheMeasurementInvarianceModelsofBRSQFactorsacrossCountries.
Model2dfΔ2ΔdfCFIΔCFITLIRMSEA95%CIRMSEAESEMweakdimensionality(model1)1445.
541500.
987.
975.
035.
033-.
037ICM-CFAstrongdimensionality(model2)8616.
4848005902.
181300.
892-.
105.
872.
080.
078-.
081ESEMinvariantFL(model3)2827.
2018001544.
583300.
972-.
015.
967.
041.
039-.
042ESEMinvariantFL,TH(model4)3625.
2521020646.
865220.
964-.
008.
967.
041.
039-.
042Note:df:degreesoffreedom;Δ:differencewithmodel1;CFI:comparativefitindex;TLI:Tucker-Lewisindex;RMSEA:rootmeansquareerrorofapproximation;95%CI:95%confidenceinterval;ESEM:exploratorystructuralequationmodel;ICM-CFA:independentclustersfactoranalysismodel;FL:factorloadings;TH:thresholds.
Allchi-squarevaluesinthetablearestatisticallysignificantwithp<.
004.
Runninghead:MeasurementinvarianceoftheYBRSQTable3.
StandardizedFactorLoadingsandCorrelationsBetweentheFactorsoftheBRSQ(wholesample).
ITEMAMOTIVATIONEXTERNALINTROJECTEDIDENTIFIEDINTRINSICAM1.
801.
005-.
012.
026-.
048AM2.
910-.
016-.
017.
024-.
006AM3.
646.
138-.
001.
031-.
090AM4.
616.
220.
010-.
018-.
071EX1.
271.
393.
154-.
080.
021EX3.
257.
359.
262-.
067.
214EX5.
260.
512.
067.
024-.
141EX7.
066.
503.
359-.
164-.
013IN1-.
041.
024.
679.
069-.
002IN2.
129.
019.
739-.
056-.
019IN3.
033.
297.
133.
545-.
093IN4-.
163.
315.
570.
111-.
154ID1.
041-.
324.
170.
586.
125ID2-.
023-.
260.
123.
635.
148ID3.
033.
247-.
051.
641.
030ID4-.
007.
146.
048.
439.
244IM1-.
037-.
083.
048.
121.
622IM2-.
189-.
045-.
080.
235.
525IM3-.
008.
140-.
105-.
089.
919IM4.
019.
050-.
001.
099.
687AMOTIVATION1.
000EXTERNAL.
6201.
000INTROJECTED.
493.
4861.
000IDENTIFIED-.
162.
082.
1681.
000Runninghead:MeasurementinvarianceoftheYBRSQINTRINSIC-.
344-.
248-.
070.
4651.
000Note:boldface:factorloadingsfreelyestimated,allothertargetedtobezero;italics:factorloadingswithestimatedvalues≥|.
30|inspiteofbeingtargetedtobezero.