Informing-context-and-change-in-young-children-s-sociobehavioral-development

EarlyChildhoodResearchQuarterly29(2014)255–267

ContentslistsavailableatScienceDirect

EarlyChildhoodResearchQuarterly

Informingcontextandchangeinyoungchildren’ssociobehavioraldevelopment–ThenationalAdjustmentScalesforEarlyTransitioninSchooling(ASETS)?

PaulA.McDermotta,?,MarleyW.Watkinsb,MichaelJ.Rovinec,SamuelH.Rikoond

a

UniversityofPennsylvania,UnitedStatesBaylorUniversity,UnitedStatesc

PennsylvaniaStateUniversity,UnitedStatesd

EducationalTestingService,UnitedStates

b

article

info

abstract

Articlehistory:

Received23January2013

Receivedinrevisedform9February2014Accepted14February2014Availableonline1March2014

Keywords:

EarlychildhoodHeadStart

BehavioraladjustmentContext

ItemresponsetheoryTransition

Thisarticlerecountsthedesignandvalidityevidenceforcontextually-speci?cmeasuresofearlychild-hoodsocialandbehavioraladjustmentwithinschoolusingtheAdjustmentScalesforEarlyTransitioninSchooling(ASETS).ThroughprimaryanalysesofdatafromtheHeadStartImpactStudy,arepresenta-tivenationwidesample(N=3077)ofrandomlyselectedchildrenfromlow-incomefamilieswasusedtoinformdevelopmental-transitionalstabilityandchangeinadjustmentacrossnumerousschoolcontexts.Longitudinalexploratoryandcon?rmatoryfactoranalysesyieldedreliableandtemporallycontinuousbehavioraldimensionsassessingthepervasivenessofPeer,Learning,andTeacherContextProblems.EachcontextdimensionwasequatedverticallythroughIRT,withBayesianscoringacrosstwoyearsspanningprekindergartenthrough1stgrade.MultilevelmodelingprovidedsupportfortheconcurrentvalidityofASETScontextualscalesandtheirabilitytoassessfutureriskofacademicandbehavioralproblems.ASETSscalesarealsoshowntorevealdifferential,contextually-based,changetrajectoriesacrossfouryearsofearlyschooltransition.

©2014ElsevierInc.Allrightsreserved.

1.Introduction

Thepastdecadehaswitnessedaphenomenalincreaseinpolicyinitiativesandresearchcenteringonthesocio-emotionalneedsofyoungchildren(Campbell,2001;Egger&Angold,2006;President’sNewFreedomCommissiononMentalHealth,2003;Rescorlaetal.,2007,2011).Motivationstemslargelyfromtheobservationthatprevalenceratesforpreschoolemotionalandbehavioralproblemsapproach20%(Egger&Angold,2006),withearlyanduntreatedproblemsunderminingcriticaldevelopmentalprocessesandpor-tendingmoreseriousandsometimesintractabledisordersatlater

?ThisresearchwasconductedwiththecooperationoftheU.S.DepartmentofHealthandHumanServices,AdministrationforChildrenandFamilies,andsup-portedinpartbytheU.S.DepartmentofEducation,InstituteofEducationSciences(GrantNo.R305C050041-05).Theviewsexpressedrepresenttheviewsoftheauthors,andnotoftheEducationalTestingService.

?Correspondingauthorat:SchoolofEducation,UniversityofPennsylvania,3700WalnutStreet,Philadelphia,PA19104-6216,UnitedStates.

E-mailaddresses:drpaul4@http://wendang.chazidian.com(P.A.McDermott),rikoon@upenn.edu(S.H.Rikoon).

ages(Campbell&James,2007;Feeney-Kettler,Kratochwill,&Kettler,2011;Kataoka,Zhang,&Wells,2002).Oneunderstandableresponsehasbeenavarietyofassessmentdevicestoidentifyanddifferentiatemanifestationsofpreschoolsocio-emotionaldistress(Campbell&James,2007;Feeney-Kettleretal.,2011;Rescorlaetal.,2011),theintentionbeingtoclarifythedistinctnatureofprob-lemsinsuchawaythatmightleadtopreventativeorrestorativeintervention.

Mostcontemporaryinstrumentsforassessmentofearlyemo-tionalandbehavioralproblemsembraceacommonformulation.Sinceyoungchildren,giventheirsocial,conceptual,andlinguisticimmaturity,andlimitedperspectives,areunabletoreportaccu-ratelytherelevantsymptomatologyandincidenceoftheirowndistress(Fulmer&Frijters,2009;Moll&Tomasello,2012;Norwood,2007;vonBaeyer,Forsyth,Stanford,Watson,&Chambers,2009),informedadultobservers(teachers,parents)aretypicallyaskedtorespondtoratingscalesorquestionnairesthatsurveythechild’sreactionsathomeorschool.ThebestexamplesincludethePreschoolandKindergartenBehaviorScales(Merrell,2003),theDevereuxEarlyChildhoodAssessment(LeBuffe&Naglieri,1999),theBehaviorAssessmentSystemforChildren(Reynolds&

http://wendang.chazidian.com/10.1016/j.ecresq.2014.02.004

0885-2006/©2014ElsevierInc.Allrightsreserved.

256

P.A.McDermottetal./EarlyChildhoodResearchQuarterly29(2014)255–267

Kamphaus,2004),andtheAchenbachSystemofEmpiricallyBasedAssessment(Achenbach&Rescorla,2000).Thus,teachersintheclassroomorparentsinthehomeportrayachild’sadjustmentbyindicatingthepresenceorfrequencyofnumerousspeci?csymp-toms,wherethesymptomsaremanifesteitherthroughobservablechildbehaviorsorperceivedchildemotions.Inturn,researchersapplyfactor-analyticprocedurestotheresultantteacherorparentresponsesandtherebydiscoverthatdifferentsymptomstendtogrouptogetherandrevealcommonsurfacesyndromesordimen-sionsthatmoreorlessresembletraditionalclinicalpsychiatricdisorders.Inthisway,agivenchildcanbeassessedbyateacheratschoolandthechild’scurrentsocio-emotionaladjustmentmaybequanti?edforavarietyofdifferenttypesofproblems(aggression,withdrawal,etc.).

Althoughsuchinstrumentationhasbeenimmenselyusefulforadvancingunderstandingofearlychildhoodsocio-emotionalmani-festations,andarguablyforinformingpertinentintervention,ithasbeenourviewthatmostcontemporaryinstrumentsdonotreston?rmtheoreticalgroundsanddonottakeadvantageofthetechnicalcapacitiesavailablefordesignandapplicationofsurveyinstru-ments.Speci?cally,itisourpositionthatpopularratingscalesandquestionnairesaredesignedwithlittleornoattentiontothedis-tinctcontextualframeworkswithintheschool(orhome)ortothesignaturetransitionalnatureofemotionalandbehavioralproblemsaschildrendevelop.Intheintroductorysectionofthisarticle,wediscussthetheoreticalimportofcontextandtransitionforstudyingearlysocio-emotionaladjustment.Wethendemonstratethroughdevelopmentandapplicationofanewnationalmeasure,howcon-textandtransitionplayacentralroleinadvancingunderstandingofearlychildhoodadjustment.

1.1.Contexttheory

Intheassessmentdomain,contexttheoryisprobablybestrepresentedbytheworkofMischel(2004).Mischelpointstoakeyfactorguidingtheconstructionoftraditionalmeasuresofperson-alityandadjustment—theassumptionthatagivenscorelevelonaparticulartraitdimension(e.g.,withdrawal)translatestoagivendispositionforthattypeofemotionorbehavior.Butasresearchandexperiencedictate,peoplesharingthesamescorelevelonatraitordimensionwill,inreality,displayasubstantialrangeofdispositionsfortheanticipatedemotionsorbehaviors.Inpractice,thismakestraditionalassessmentslessaccurateandconsequentlylessuseful.Alternatively,Mischel,Shoda,andMendoza-Denton(2002)illustratehowtheactualdispositionforagivenemotionorbehaviordependsnotonlyonthetraitordimensionalscorelevelbutonthecontextualcircumstanceswhereintheemotionsorbehaviorsareembedded.Thus,childrenwithahighlevelofawithdrawaltraitwillnotreactuniformlyindifferentcontexts,suchaswhenapproachedbyateacherversuswheninvolvedingroupplayversuswhenconfrontedbychallenginglearningtasks.Indeed,knowledgeofthesituationsthatgiverisetoproblembehaviorsisfundamentaltounderstandingthemotivationsbehindproblemsandtheaccuratepredictionoffutureincidence(Zayas,Whitsett,Lee,Wilson,&Shoda,2008).Traditionalinstrumentstendtoignorecontextualframeworksbyregardingsituationalvariationassomesortof“noise”orerror(Mischeletal.,2002).Suchinstrumentsmayfeatureitemsthateitherinquireabouttraitbehaviorwithoutanyreferencetospeci?csituationsunderwhichitmayormaynotemerge,orotherwiseaveragescoresacrossallsortsofsituationstoproduceageneralcompositeontraitbehavior.

Thecontextualviewisentirelyconsistentwiththedevelopmental–ecologicalperspectiveadvocatedbySabolandPianta(2012)forstudyingcontextsthatdifferentiallyin?u-enceteacher–childrelationships;byZayasetal.(2008)andKagan

(2003)whoshowtheroleofcontextforexplainingintra-individualvariationsinbehavioraldispositions;byMian,Wainwright,Briggs-Gowan,andCarter(2011)andThorsen,Goldberg,Osann,andSpence(2008)whofocusonspeci?csituationsthatinvitegoodversusbadreactions;andbySameroff(2010)andBronfenbrennerandMorris(2006),whooffermoreuni?edtheoriestobindnaturalindividualchildpropensitiesandcontextualframeworksinthebroaderstoryofhumandevelopment.Theideathatcontextualspeci?citymakesadifferenceisalsosupportedbyemergentempiricalliteraturedemonstratingthat:(a)youngchildren’swith-drawalandemotionalregulationvaryasafunctionofclassroomcontext(Buss,2011;Goldsmith&Davidson,2004);(b)peer-groupcontextsaffecttheaggressivenessofchildrenwithspecialneeds(Visser,Kunnen,&vanGeert,2010)andpreschoollanguageacquisition(Justice,Petscher,Schatschneider,&Mashburn,2011);(c)manipulationofclassroomstructuralaspectsandlearninglocationscanabateproblembehavior(Kern&Clemens,2007;Wannarka&Ruhl,2008);(d)alternationofindividualandgroupactivitiesandtheamountofteacherinvolvementaffectschildclassroomengagement(Powell,Burchinal,File,&Kontos,2008);and(e)plannedfree-timeandclassroomtransitionsaffectbehavior(Joosten,Bundy,&Einfeld,2012).

1.2.Transitiontheory

EarlychildhoodtransitiontheoryemergesfromtheworkofEntwisleandAlexander(1993)andEntwisle,Alexander,andOlson,(2005),withseminalconnectionstoPiagetianandEriksonianconceptsofstagetheory.Transitiontheoryessentiallyarguesthatchildren’sdevelopmentalstatusismultifacetedandcon-stantlychanginginresponsetoontogeneticandenvironmentalin?uences.Itholdsthatchildren’smoreorlesssuccessfuladap-tationstothosein?uencessetthetemplateforfuturecapacitiestoadaptandthat,aspertainstolong-termacquisitionofcop-ingmechanismsandcognitiveachievements,themostcriticaldevelopmentalperiodsarethoseproximatetomajortransitions.Asresearcherspointout,suchtransitionperiodsinearlychild-hoodeducationincludemovementintoandthroughpreschoolandprogressionintoregularkindergartenand?nally?rstgrade(Benner&Crosnoe,2011;Buss,2011;Goldsmith&Davidson,2004;Hemmeter&Ostrosky,2006;Pianta,Cox,&Snow,2007;vonSuchodoletz,Trommsdorff,Heikamp,Wieber,&Gollwitzer,2009).

Thus,whereasprekindergartenentrywilloftenprovideachild’s?rstexposuretopart-orfulldayschedulesorganizedaroundgroupmealsandnapsandindividualorcompanionplay,kindergartenand?rst-gradeactivitiesbegintosupplantdiscoverylearningwithmoredeliberateandstructuredactivitiesemphasiz-inggroup-centeredcommoncurriculathatencourageself-relianceandcompetition.Eventually,desksreplaceplaycircles,vocalitybecomesimperative,literacybecomesfundamentaltowhatistran-spiringintheclassroom,andacademicfailureorretentionbecomerealprospects.Consequently,thecausalcentralityofearlyschooltransitiontolong-termchilddevelopmenthasessentiallyrisentoameta-theoreticallevelthatregardsearlytransitionsinschoolingasamajordevelopmentalmilestone(Eivers,Brendgen,&Borge,2010).

1.3.Innovativeinstrumentation

Theinstrumentationandmethodsforassessingthecontextualnatureofchildren’sschoolsocio-behavioraladjustmentwas?rstsuggestedbyStott(1966)andimplementedfullyinthedevelop-mentandnationalstandardizationoftheAdjustmentScalesforChildrenandAdolescents(ASCA;McDermott,1993;McDermott,Steinberg,&Angelo,2006).Incontrasttothetraditionalformat

P.A.McDermottetal./EarlyChildhoodResearchQuarterly29(2014)255–267

257

ofclassroomratingscales,whereteachersareaskedtorespondtolistsofsymptomaticbehaviorsbyindicatingtheirgeneralintensityorfrequency,theASCApresents122problembehavioritemsembeddedin22differentcontextsinvolvingpeerinterac-tions,teacher,learningtasks,groupactivity,organizedgames,andfreeplay.Theteacherindicateswhethereachproblembehaviortypi?esthechild’sbehavioroverthepastmonth,whereasetofmultiplepossiblebehaviorsappearswithineachcontext.Alsounliketraditionalratingscales,ASCAembedsoneortwohealthyorcommonplacebehaviorswithineachcontext.Thus,teachersareprovidedalternativenormalbehavioralvariantsasameasuretoavertresponsebiasinducedbyexclusivepresentationofproblembehaviorchoices.Thenatureandseverityofchildren’sadjustmentproblemsaremeasuredintwoways:First,theproblembehaviorsaregroupedbyfactoranalysestorevealmultipledimensionswhosecomponentbehaviorsindicatecommonsurfacesyndromes(suchasoppositionality,dif?dence,andavoidance).Thesedimen-sionsarecalledphenotypesbecausetheyembodybehaviorsthatsharesimilarappearanceandfunction,althoughunlikeotherratingscales,theyactuallyre?ectspeci?ctypesofproblematicbehaviorsthatarepervasiveacrossmultipledifferentcontexts.Inthisway,maladjustmentisde?nedbyitsgeneralityacrossschoolcontextsratherthatitsemergenceinisolatedcircum-stances(thelatterphenomenonactuallysuggestingatransientreactiveproblemratherthanamoregeneralmaladjustment;Horn,Wagner,&Ialongo,1989).Second,thecontextsthemselvesaregroupedbyfactoranalysestorevealthesituationsthatgiverisetothephenotypicproblembehavior.Thesecontextualdimensionsarecalledsitutypesandindicatewhetherproblemsemergeincontextsinvolvingtheteacher,classmates,orstructuredlearning.

TheASCAisdesignedforchildreninkindergartenthrough12thgrade.Asameanstoaccommodatepreschoolchildren,ASCAwasrevisedandvalidatedforapplicationwithHeadStartchildren(Bulotsky-Shearer,Fantuzzo,&McDermott,2008;Noone-Lutz,Fantuzzo,&McDermott,2002)andnamedtheAdjustmentScalesforPreschoolIntervention(ASPI).AswithASCA,ASPIyieldsscoresonphenotypedimensionsthatindicatewhattypesofprob-lembehaviorexistandsitutypedimensionsthatinformwhenandwheretheyexist,thusavailingimportantcluesastomotiva-tionandpotentialintervention.Morerecently,thetechnologytoassesschildren’stransitionfrompreschoolintoformalschoolingbecameavailable.AlloftheoriginalASPIitemsandtheiruniquecontextualformatswereadministeredintactforthenationalHeadStartImpactStudy(U.S.DepartmentofHealthandHumanServices[DHHS],2010a),spanningtwoyearsofpreschoolthroughkindergartenand?rstgrade.BecausethenationalsamplereachedhorizontallywellbeyondHeadStartintootherpublicandprivatepreschoolsandverticallythroughkindergartenand?rstgrade,theinstrumentwasrenamedtheAdjustmentScalesforEarlyTransi-tioninSchooling(ASETS).McDermott,Watkins,Rovine,andRikoon(2013)resolvednewphenotypedimensionsforASETS,calibrat-ingthemthroughitem-responsetheory(IRT)andvalidatingthemforlongitudinalassessment.Asyet,contextually-basedsitutypedimensionshavenotbeenestablishedorvalidatedforthenewnationalinstrument.

Thisarticlereportsthelongitudinalfactoranalyses,IRTscal-ingandscoring,andconcurrentandpredictivevalidityofsitutypedimensionsofproblembehaviorassessedbyASETS.Itfurtherdemonstratesviamultilevelmodelingtheutilityofthecontex-tualdimensionsforassessingrelativeriskofpreschoolchildren’slateracademicnonpro?ciency,fordetectingchangeinadjustmentlevelsoverthelongtransitionfromearlypreschoolintoformalschooling,andforidentifyingthesignaturedevelopmentaltra-jectoriesthatearmarkeventuallysuccessfulversusunsuccessfullearners.

2.Method

2.1.Participants

TheHeadStartImpactStudy(HSIS;DHHS,2010b)wasanationwiderandomizedcontroltrialdesignedtodeterminetherel-ativeeffectivenessofHeadStartandcomparableprekindergartenprograms.Participantsweredrawnrandomlyfrom223prekinder-gartenagenciesacrossallgeographicregionsoftheUnitedStates,providedthateachchildwaseligibleforHeadStartentry(essen-tiallyafamilyincomebeloworclosetothefederalpovertylevel).Theyoungestchildrenwereenrolledinprekindergarteninaca-demicyear2002–2003(AY0203)andfollowedthroughAY0506.ASETSwascompletedbyeachchild’sclassroomteacherattheendofthe?rstyearofprekindergarten(PreK1)andsecondprekinder-gartenyear(PreK2),thekindergartenyear(K),and?rst-gradeyear(1stgrade).

BecausenotallchildrenrandomlyselectedforprekindergartenentryactuallyenteredschoolforPreK1,andbecauseothersdidnotenterprekindergartensettingsthatwouldprovideateacherand/orclassroom-typeenvirons,thenationalsamplesizeincreasedaschil-drenmovedfromPreK1to1stgrade(i.e.,PreK1N=1377,PreK2N=2764,KN=2873,1stgradeN=3077).ConsideringtheASETSfullnationalsample(N=3077),Mageatentrytothestudywas4.0years(SD=.5),with49.6%ofchildrenbeingfemales,37.8%His-panic,29.5%AfricanAmerican,32.7%Whiteorotherrace/ethnicity,25.7%primarilySpanish-speakingatentry,12.8%identi?edwithspecialneeds,and82.7%residinginurbanareas.DuringPreK1,childrenattended540preschoolcenters(867classrooms)andduringPreK21032centers(1815classrooms),whileduringK,childrenattended1469schools(2280classrooms)andduring1stgrade1617schools(2576classrooms).ThroughPreKyears,asmuchas80%ofclassroomswerenotassociatedwithconventionalschools(approximately60%beingpart-dayenvironssuchasdaycareorothernon-schoolcenters),withabout90%ofpost-PreKclassroomsaf?liatedwithpublicschools.Detailedsamplecharac-teristicsarereportedbyMcDermottetal.(2013)andDHHS(2010a,2010b).

2.2.Instrumentation:sociobehavioraladjustment

Table1presentsdescriptivestatisticsforallinstrumentationusedinthecurrentstudybydevelopmentallevel.ASETScontains134behavioralindicatorsembeddedin22situationalcontexts.Eachindicatormaybecheckedornotbytherespondingteachertodescribethechild’sbehavioroverthepastmonth.The22contextscoverrelationshipswiththeteacherandwithotherchildren,cop-ingwithclassroomexpectations,anddemeanorduringgamesandplay.Atypicalcontextinquires,“Howdoesthechildreacttocorrec-tion?”Withinthatcontext,theteachermaydescribechildbehaviorbyendorsingoneofmoreofthefollowingindicators:“Improvesforthemomentbutitdoesnotlastlong,”“Acceptscorrectionwithoutfuss,”“Takescorrectionbadly(sulky,muttering,etc.),”“Answersbackaggressively,makesthreats,orcreatesadisturbance,”and“Cries.”Inthismanner,eachcontextpresentsthreetosevenrel-ativelynegativeproblembehaviorindicatorsthataretheoreticallyandempiricallyre?ectiveofapotentialsurfacesyndromeorphe-notype(Aggression,AttentionSeeking,Reticence/Withdrawal,orLowEnergy;seeMcDermottetal.,2013)andmostcontextsprovideoneortwopositiveorhealthybehaviorchoices.Altogether,ASETSfeatures112problemand22healthyindicatorswhere,giventheintendedpurposeoftheinstrument,thenumberofproblemindi-catorsendorsedbytheteacherforagivencontextcomprisesthatcontext’sscoreforproblembehavior(scorerangesvaryingfrom0–3to0–7,dependingonthenumberofproblembehaviorsembed-dedindifferentcontexts).Moreover,allcontextualandindicator

258

P.A.McDermottetal./EarlyChildhoodResearchQuarterly29(2014)255–267

Table1

Descriptivestatisticsforinstrumentationbydevelopmentallevel.

Prekindergarten1

Prekindergarten2

Kindergarten

FirstGrade

Instrumentandcontexts/subscales

#ofindicators

N

M(SD)

N

M(SD)

N

M(SD)

N

M(SD)

AdjustmentScalesforEarlyTransitioninSchooling(contextsa)

1377

5444548434434764784364

2764

.5(.6).3(.5).5(.6).2(.4).4(.6).6(.6).6(.8).5(.7).3(.6).6(.8).6(.7).3(.5).4(.7).5(.8).7(.9).3(.7).5(.8).8(1.0).4(.6).5(.6).5(.8).5(.8)

2873

.4(.6).2(.5).5(.6).2(.4).3(.5).6(.6).6(.8).4(.7).4(.6).5(.8).4(.6).3(.5).3(.6).4(.8).6(.9).3(.6).4(.8).7(1.0).3(.5).4(.6).5(.9).4(.7)

3077

.3(.5).2(.5).5(.6).2(.5).4(.6).6(.6).5(.8).5(.7).4(.6).5(.8).5(.7).3(.5).3(.6).5(.9).6(.0).2(0.6).4(.8).6(.9).3(.5).4(.6).5(.9).4(.7)

Greetingtheteacher

HelpingtheteacherwithjobsAnsweringquestionsSeekingteacher’shelpTalkingwiththeteacher

Valuingtheteacher’sattentionGeneralmannerwiththeteacherBehavingintheclassroomTellingthetruth

Reactingtocorrection

PayingattentionintheclassroomCopingwithnewlearningtasksGettinginvolvedinclassactivityWorkingwithhands(artwork,etc.)SittingduringdirectedactivitiesRespectingothers’belongingsTakingpartingameswithothersFreeplay(self-choiceactivity)Havingcompanions

BehavingwhilestandinginlineGettingalongwithagematesHandlingcon?ictswithothers

PiantaChild–TeacherRelationshipsScaleClosenessCon?ict

PositiveRelationship

.3(.5).2(.4).5(.6).3(.5).4(.6).5(.6).5(.8).5(.7).4(.7).5(.8).6(.8).3(.5).3(.6).5(.9).6(1.0).2(.7).4(.8).6(.9).2(.5).4(.6).5(1.0).4(.7)

7815

27472743

31.0(4.0)13.7(6.3)

305830503059

29.7(4.6)14.2(6.9)63.3(9.7)

Parentratingscale

TotalBehaviorProblems

14

2626

5.5(3.6)

3059

4.9(3.9)

PeabodyPictureVocabularyTest,ThirdEdition

2699

297.0(37.5)

2900

359.6(30.5)

Woodcock–JohnsonIIITestofAchievementLetterWordIdenti?cationAppliedProblemsPre-AcademicSkillsSpelling

BasicReadingSkillsWordAttack

QuantitativeConceptsMathematicsReasoning

2700268326832701

330.0(27.9)399.3(22.4)368.1(21.0)374.8(25.5)

2873287528772879

449.8(32.3)467.7(31.0)461.3(17.3)457.9(17.2)

LanguageandLiteracyAbilityMathematicsAbilitySocialScienceAbility

222

218821822177

.7(.5).8(.4).8(.4)

a

Contextdescriptionsareabbreviatedforconvenientpresentation.

languageisdistinctlybehavioral,avoidingclinicaljargonorneces-sityforspeculationaboutunobservableinternalprocessessuchaschildren’sthoughtsorfeelings.

2.3.Instrumentation:externalvaliditymeasures

ASETSscoreswerevalidatedagainstseveralteacherandpar-entratingsanddirectassessments.Resultsarereportedherefortwodevelopmentallevels(PreK2and1stgrade),althoughavail-ableforalllevels.Onlyarepresentativepresentationofresultswasfeasible,giventhelimitationsofspace.Itwasthoughtappropri-atetopresentresultsfromtheculminatingpointofthePreKlevels(PreK2)andpost-PreKlevels(1stgrade)becausetheyallowedthemaximumamountofparticipantdatawherepsychometricprop-ertiesfortheexternalmeasureswereacceptable.CertainmeasuresadministeredforHSISwereeliminatedfromthecurrentstudybecauseatagivendevelopmentallevel,theyfailedtoproducesuf-?cientdatatoyieldreasonablestatisticalpower,orbecausetheoriginalinstrumentswerealteredforHSISwithoutreportofreq-uisitepsychometricsupport(Smith,McCarthy,&Anderson,2000),orbecausetheyfailedtoproduceminimallyadequatereliability

(viz.,≥.70asrecommendedbyFabrigar,Wegener,MacCallum,&Strahan,1999andNunnally,1978)fortheHSISpopulation(DHHS,2010b,pp.3.32–3.43).

Teacherratings.ThePiantaStudent–TeacherRelationshipsScale(Pianta,1996)features15items,suchas“Thischildeasilybecomesangryatme,”ratedona5-pointscalerangingfrom1=“De?nitelydoesnotapply”to5=“De?nitelyapplies.”Threesubscalesareavailable:Closeness(7items),Con?ict(8items),andTotalPosi-tiveRelationship(15items).Substantialconcurrentandpredictivevalidityevidenceisprovided(Pianta,2001;Pianta&Stuhlman,2004)andinternalconsistencyfortherelevantHSISdevelopmen-tallevelsrangedbetween.73and.82forCloseness,.76and.89forCon?ict,and.88and.89forTotalPositiveRelationship(DHHS,2010b).TeacherreportofAcademicAbilityisratedatthecloseof1stgradeforLanguageandLiteracy,Mathematics,andSocialSci-ence,asbasedonobservedattainmentofmultipleskillscomparedtotheattainmentofpeers(DHHS,2010b).Performanceisratedaseither0=“BelowAverage”(nonpro?cient)or1=“Pro?cient.”Sinceeachmeasureisasingleindex,internalconsistencyesti-matesareinfeasible.Rather,theappropriatestandarderroroftheMisreportedhere,whereSEMforLanguageandLiteracy=008,

P.A.McDermottetal./EarlyChildhoodResearchQuarterly29(2014)255–267

259

Mathematics=.008,andSocialScience=.007.Giventhediscretescaling,subsequentstatisticalanalysesapplylogitlinkfunctionsandtheBernoulliresponsedistribution.

Parentratings.Aparentratedeachchild’saggressiveorde?-ant,hyperactive,andwithdrawnordepressedbehaviorusingtheTotalBehaviorProblemsscale.Thescalecontained14dichotomousitems,suchas“Isdisobedientathome”and“Feelsworthlessorinferior.”DevelopmentandvalidityevidenceareprovidedfortheFACESnationalstudy(DHHS,2001,p.2.27)andforHSISinDHHS(2010b).AdditionalvalidityevidencewasreportedbyVaden-Kiernanetal.(2010)andZiv,Alva,andZill(2010).FortheHSISPreK2and1stgradesamples,internalconsistencyranged.78–.79.FortheHeadStartandkindergartensamplesasreportedfortheFACESnationalstudy,internalconsistencyranged.76–.80.

Directassessments.ThePeabodyPictureVocabularyTest,ThirdEdition(PPVT;Dunn,Dunn,&Dunn,1997)assessesreceptivevocabulary.Criterionvalidityevidenceisabundant(Dumont&Willis,2006;Salvia,Ysseldyke,&Bolt,2007)andreliabilityindexesranged.70–.78fortheHSISpopulation.Also,varioussub-scalesoftheWoodcock–JohnsonIIITestsofAchievement(WJ;Woodcock,McGrew,&Mather,2002)wereadministered.ThepresentstudyusedscoresfromLetter-WordIdenti?cation(let-terandwordreadingskills;HSISinternalconsistencyranging.90–.91),AppliedProblems(solvingpracticalmathproblemsbyrecognizingprocessandcountingorcalculating;internalcon-sistency=89–.90),Spelling(writinglettersandwords;internalconsistency=.78–.81),Pre-AcademicSkillscluster(compositeofthethreeprecedingsubscales;internalconsistency=.76–.78),WordAttack(applyingphonicandstructuralanalysisskills;inter-nalconsistency=.93–.94),BasicReadingSkillscluster(compositeofthethreePre-AcademicSkillsclustersubscalesandWordAttack;internalconsistency=.90–.91),QuantitativeConcepts(identifyingnumberconceptsandrecognizingpatternsandmissingaspects;internalconsistency=86–.87),andMathematicsReasoningcluster(compositeofAppliedProblemsandQuantitativeConcepts;inter-nalconsistency=71–.78).AmplevaliditysupporthasbeenreportedfortheWJachievementsubscales(Dumont&Willis,2006;Salviaetal.,2007).

2.4.Procedure

TeachersrespondedtoASETScontextsinthespringsemestersofAY0203–AY0506.Thevariouscriterionmeasureswereadmin-isteredduringthesamesemesters,withPPVTandWJscalesappliedbytrainedtechniciansandteacherreportsonchildaca-demicabilityprovidedatthecloseof1stgrade(DHHS,2010a,2010b).Theaveragenumberofchildrenassessedperclass-roomduringAY0203=1.59,AY0304=1.52,AY0405=1.26,andAY0506=1.08.Factoranalysesandverticalscaleequatingeachrequiredamutually-exclusivecross-sectionalsamplingacrossPreK1,PreK2,K,and1stgrade.Tothisend,weconsecutivelydrewatrandomonechildfromeachdevelopmentallevel(nochildbeingdrawntwice)untilasampleof1600childrenwascon-structed,with400differentchildrenrepresentingeachlevel.Thetotalsizeofthissamplewasdictatedbythenecessityto(1)generateforstatisticalpowerpurposesthelargestsamplepossiblewith-outredundantmembershipand(2)representeachdevelopmentallevelequally.Thissamplewastermedthecalibrationsample.Thissamplewasimportantbecauseitwasusedinsubsequentlongitu-dinalfactoranalysestoensurethateachdevelopmentalagegroupwasrepresentedequally,providingsuf?cientstatisticalpowerforsubsequentcon?rmatoryanalysesinvestigatingthelongitudinalinvarianceofthefactorstructure(seebelow).Thisalsoprecludedanysourcesofwithin-childvarianceinthecalibrationsample,asrequiredforverticalequatingandderivationofscoringparameters.Thereafter,thecalibrationsampleof1600wasrandomlybifurcated

toformanexploratoryfactoranalysis(EFA)subsampleof800andcon?rmatoryfactoranalysis(CFA)subsampleof800.

Factoranalyses.Inasmuchasthepointscalesforthe22con-textswereessentiallyordinal,itwasnecessarytotreateachscoredistributionascategorical.Numerousresearchershavedemon-stratedthespuriousfactorsandunstabledimensionsthatcanarisewhencategoricaldataaretreatedascontinuousdatainEFA(Bernstein&Teng,1989;McDonald&Ahlawat,1974;Mislevy,1986;Mooijaart,1983;Muthén,1987;Waller,2001).Alternatively,perWaller’s(2001)recommendation,weappliediterativecommonfactoringwithasmoothedpolychoricmatrix.Speci?cally,two-stagemaximum-likelihoodestimation(Olsson,1979)wasusedtoproduceaninitialpolychoriccorrelationmatrixamongthe22con-textscoresfortheEFAsubsample,andthematrixwassmoothedforpositivesemide?nitenessthroughleast-squaresapproximationoftheoriginalmatrix(Knol&Berger,1991).Thesmoothedmatrixwassubmittedforminimumaveragepartialling(MAP;Garrido,Abad,&Ponsoda,2011;Velicer,1976)tosuggestthenumberoffactorsforretentionandthereaftersubmittedtoiterativecommonfactoringwithvarimax,equamax,andpromaxrotation.Theidealstructurewasthatwhichsatis?edmultiplecriteria;namely,thesolutionmust(a)approximatesimplestructureasre?ectedinmaximumhyperplanecount(Yates,1987)andcoverageofcontexts,(b)haveatleastfoursalientloadingsperfactorwhereloadings≥.40indicatesalience,(c)produceinternallyconsistentfactors(i.e.,r≥.70),and(d)maketheoreticalsenseintermsofparsimoniouscoverageofthedataandcompatibilitywithleadingresearchinthearea(Fabrigaretal.,1999).

ThefactorstructurebasedonthesalientmarkersfromtheidealEFAsolutionwereanalyzedfortheCFAsubsampleusingmaximum-likelihoodestimationundertheSatorra–Bentlerscaleddifferencechi-squarefornonnormaldata(Satorra&Bentler,2001),seekingacceptable?twheretheRootMeanSquaredErrorofApproximation(RMSEA)≤.08andComparativeFitIndex(CFI)≥.90(Marsh,Liem,Martin,Morin,&Nagengast,2011).

Scaling.Factordimensionswerescaledvertically,joiningPreK1toPreK2,PreK2toK,andKto1stgrade.Foreachdimension,anumberofcontextswereidenti?edaslinkingcontextsthroughmultiple-groupIRTanalysis(Muraki&Bock,2003)ofDifferentialItemFunctioning(DIF).Alinkingcontextisonethatappearssimul-taneouslyattwoadjacentdevelopmentallevels.DIFwasassessedthrough??2testsoftheresiduals(basedonexpectedcomparabilityofcontextdif?cultyparameters)forlinkingcontextsacrossadja-centdevelopmentallevels(e.g.,PreK1versusPreK2).Contextsdisplayingstatisticallysigni?cantDIFweredismissedaspotentiallinkingcontexts.One-thirdthenumberofcontextscomprisingadimensionforagivenlevelwereselectedaslinkingcontexts,beingchosensoastobestdistributelinkingcontextsacrossdif?cultylevelscoveringthedimension’sdistribution.Verticalequatingwasaccomplishedwiththelongitudinalcalibrationsample(N=1600)usingmultiple-groupIRT(ProgramPARSCALE;Muraki&Bock,2003)testingboththeGeneralizedPartialCreditModel(GPCM;Muraki,1992)andtheGradedResponseModel(GRM;Samejima,1996).ResultantitemparameterswereappliedfortheASETSfullnationalsample(N=3077),withscorescalculatedviaexpectedaposteriori(EAP)estimation(Thissen&Wainer,2001),wherethescaledscore(SS)M=50andSD=10atPreK1,thereferencelevel.Externalvalidity.Allvalidityanalyseswereperformedusingavailabledatafromthefullnationalsample.Product–momentcorrelationswerecomputedtoshowthedirectionandstrengthofrelationshipsbetweenASETSdimensionscoresandscoresfortheexternalvaliditymeasures.Giventhevolumeofdataacrossdevelopmentallevels,reportingislimitedtothemostrepresentativelevels;i.e.,PreK2(culminatingthePreKperiod)and1stgrade(culminatingthepost-PreKperiod).Sincethedatawerenestedwithinteachers/classrooms,relationshipsalso