自我控制哪里而来呢？.pdf

自我控制与问题行为的关系什么呢？

CORRECTEDMARCH23,2010;SEELASTPAGE

DevelopmentalPsychology2010,Vol.46,No.1,245–257

©2010AmericanPsychologicalAssociation0012-1649/10/$12.00DOI:10.1037/a0016538

WhereSelf-ControlComesFrom:OntheDevelopmentofSelf-Controland

ItsRelationshiptoDevianceOverTime

AlexanderT.VazsonyiandLiHuang

AuburnUniversity

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Thecurrentstudytestedasetofinterrelatedtheoreticalpropositionsbasedonself-controltheory(M.R.Gottfredson&T.Hirschi1990).Datawerecollectedon1,155childrenat4.5years,at8.5years(3rdgrade),andat10.5years(5thgrade)aspartoftheNationalInstituteofChildHealthandHumanDevelopmentlongitudinalstudyovera6-yearperiod.Findingsbasedonsimplestructuralequationmodelsandlatentgrowthmodelingofdevelopmentaltrajectoriessuggestthat(a)therewasgreatconstructstabilityofself-controlanddevianceoverthe6-yearperiod,(b)therewaspositivegrowthinself-controltrajectoryovertime,(c)parentingpredictedthistrajectorybutalsoexplainedvariabilityinself-controlatinitialstatus,(d)therewasadecliningdeviancetrajectoryovertime,(e)self-controlatinitialstatusreducedtheunexplaineddeviancevarianceby44.8%,and(f)boththeinterceptandslopefactorssharedabout75%ofthevariancebasedongrowth-to-growthcurvepredictivemodelsofself-controlanddeviance.Findingsarediscussedintermsoftheirimplicationsforself-controltheoryandfutureempiricalwork.

Keywords:generaltheoryofcrime,selfregulation,parenting,age-crimecurve,delinquency

Self-controltheoryremainsoneofthemostheavilyresearchedandcitedtheoriesofdeviance,delinquency,andcrime(Cohen&Farrington1999;Wright,Bryant,&Miller,2001);thisisduetoitsradicaldeparturefrommostprevioussociologicalexplanationsbutalsotothefactthatitsapparentsimplicityandparsimonyhavestimulatednumerousempiricaltests.GottfredsonandHirschi(1990)haveprovidedahighlytestableetiologicalroadmapthatcentrallyfeatureslowself-controlasakeyprobabilisticexplana-toryconstructusefulintheexplanationofcrimeanddeviance.Althoughlowself-controlhasbeenfrequentlymisinterpretedasasoledeterministiccauseofcrimeintheliterature,GottfredsonandHirschihavenotascribedexclusiveimportancetoit(seealsoGottfredson,2006);infact,lowself-controlissimplyonekeyingredientpartofthehedoniccalculus—thedecisionprocessinwhichanindividualengagespriortoviolatingsocialnorms.Got-tfredsonandHirschidefinedlowself-controlasapersistentindi-

AlexanderT.VazsonyiandLiHuang,DepartmentofHumanDevelop-mentandFamilyStudies,AuburnUniversity.

ThisstudywasconductedbytheNationalInstituteofChildHealthandHumanDevelopment(NICHD)EarlyChildCareResearchNetworksup-portedbyNICHDthroughacooperativeagreementthatcallsforscientificcollaborationbetweenthegranteesandtheNICHDstaff.ThecontentissolelytheresponsibilityoftheauthorsanddoesnotnecessarilyrepresenttheofficialviewsoftheNationalInstituteofChildHealthandHumanDevelopmentortheNationalInstitutesofHealth.

PreviousversionsofthisarticlewerepresentedatthebiennialmeetingoftheSocietyforResearchonChildDevelopment(March2007,Boston,Massachusetts)andtheannualmeetingoftheAmericanSocietyofCrim-inology(November2007,Atlanta,Georgia).

CorrespondenceconcerningthisarticleshouldbeaddressedtoAlex-anderT.Vazsonyi,DepartmentofHumanDevelopmentandFamilyStud-ies,284SpidleHall,AuburnUniversity,Auburn,AL36849.E-mail:

245

vidualbehavioraltendency,onewhichisestablishedearlyinlifethroughsocializationexperiencesandpressures,largelylocatedinthefamily,thoughtheydonotexcludeothercontrol,restraint,orsocializationmechanisms(e.g.,teachersatschool).DespitethefactthatHirschi(2004)hasslightlydistancedhimselffromtheoriginaldefinitionoflowself-control(alsosimilarlydiscussedbyGottfredson,2006;Piquero&Bouffard,2007)andhowitshouldbemeasured,individualslowinself-controlcanbedefinedasbeing“impulsive,insensitive,physical(asopposedtomental),risk-taking,short-sighted,andnonverbal”(Gottfredson&Hirschi,1990,p.90).Accordingtothetheory,thisindividualpropensityinteractswithsituationalopportunitiesthatmayormaynotresultinthecommissionofdeviant,criminal,oranalogousacts,thoughitseemsclearthatthetheoryprimarilyconcernsitselfwiththedevelopmentandimportanceofself-control.

Amodestnumberofrecentstudieshavetestedwhatmostscholarshaveinterpretedtobeoneofthekeyempiricalissuesrelatedtoself-control,somethingwefindofinterestbutoflimitedtheoreticalandpracticalimportance,namely,whetherself-controlisstableovertime,mostlybasedonvariable-centeredstrategies(Hay&Forrest,2006;Mitchell&MacKenzie,2006;Turner&Piquero,2002;fordevelopmentalstudies,seeBradley&Corwyn,2005,orEisenbergetal.,2005).Ontheotherhand,littleempiricalworkhasexaminedwhatwebelievetobeperhapsthemostimportantpredictionbyself-controltheory:howself-controlde-velopsduringthefirstdecadeoflife.Thus,thecurrentstudyaddressedthisandassociatedgapsintheliterature,includinganexaminationofhowdevianceco-evolvesovertime.Thoughnotexplicitlydiscussedbyself-controltheory,thisparallelquestionlogicallyfollows;asself-controldevelopsandincreasesovertimeduringthefirstdecadeoflife,sotoodoesdeviancedecreaseduringthesametimeperiod,despitesubsequentincreasesduringmidtolateadolescence.Thecurrentstudyalsore-examinedthestability

自我控制与问题行为的关系什么呢？

246

VAZSONYIANDHUANG

thesisinaninitialstepbasedonanationalsampleofchildrenbetweentheagesof4.5yearsand10.5years.

Wetestedfiveinterrelated,theoreticallydrivenhypothesesus-ingalatentgrowthcurveanalyticapproachthatincludedbothunconditionalandconditional(withpredictors)models:(a)whetherself-controldevelopsorincreasesovertimeduringthefirstdecadeoflife(unconditional),(b)whetherpositiveparentingexplainsvariabilitythedevelopmentinself-control(conditional:initialstatusandtrajectory),(c)whetherdeviancedevelopsordecreasesovertime(unconditional),(d)whetherself-controlac-countsforvariabilityinthedecreaseofdevianceovertime(con-ditional:initialstatusandtrajectory),and(e)whethertheself-controlinterceptandslopepredictthedevianceinterceptand.

lyslope.

daor bdeLiteratureReview

tainmScholarshipthatfollowedtheoriginalspecificationoflowself-esscontrolhasprovidedimpressiveempiricalsupportfortheconstructi deofself-controlanditsassociationwithcrimeanddeviance(e.g., bGrasmick,Tittle,Bursik,&Arneklev,1993)butalsowithanalo-tot ogous,norm-violatingbehaviors(e.g.,accidents,gambling,orothern siimprudentbehaviors;Arneklev,Grasmick,Tittle,&Bursik,1993), dnalthoughthishasnotbeenwithoutsubstantialconceptualcriti-a recismsofthetheory(e.g.,Akers,1991;Benson&Moore,1992;sul Geis,2000;Goode,2008;Muraven,Pogarsky,&Shmueli,2006;auWikstro¨m&Treiber,2007).Empiricalsupporthasbeenfoundidvidindependentofmethodology,whetherbehavioralorattitudinalni (cognitive)measuresofself-controlwereusedandacrossavarietyeht ofpopulations.Thissupporthasalsobeenextendedtosamplesfo eoutsidetheUnitedStates(Tittle&Botchkovar,2005;Vazsonyi,s uPickering,Junger,&Hessing,2001;Vazsonyi,Wittekind,Bellis-lanton,&VanLoh,2004)butalsotodifferentracialandethnicsoregroupswithintheUnitedStates(Morris,Wood,&Dunaway, pe2007;Vazsonyi&Crosswhite,2004).Whatremainsanunad-ht rdressedanduntestedpartofthetheoryiswhenandhowself-of ycontroldevelops(cf.Brannigan,Genmell,Pevalin,&Wade,2002)leloandtowhatextentitpredictsdevelopmentalchangesindeviance.

s dedneWhereDoesSelf-ControlComeFrom?

tni si GottfredsonandHirschi(1990)havewrittenextensivelyaboutelciwhereself-controlcomesfrom,acknowledgingtheimportanceoftra researchoverthepasthalfcenturysincetheseminalworkbythesihTGluecks(1950;1962)ontheeffectsofearlychildhoodsocializa-tionexperiencesandfamilyprocesses.Theyhavearguedthatself-controldevelopsearlyinlife,throughsocializationpressuresappliedbothbycaregiversandbyinstitutionsinvolvedinsocial-izationprocesses,liketheschool.Theyalsosuggestedthatself-controlisestablishedearlyinlife,thatitisconsolidatedbyage8or10,andthatitislargelystable(intermsofrankorder,notabsolutelevels)thereafter.Childrearing—theparticularfamilyclimate,associatedfamilyprocesses,andspecificparentingbehav-iors—iskeyinthesocializationofself-control.Intheabsenceofpositivesocializationand(mostly)positiveparentingbehaviors,suchaspositiveaffect,supervisionormonitoring,anddiscipline,childrenarelesslikelytolearntodelaygratification,tobesensi-tivetoothers,andtoplanfortomorrow.GottfredsonandHirschialsoacknowledgedtheimportanceofmodelingorrewardingpos-

itivebehaviors,althoughtheysuggestedthatmodelingaloneisunabletoshapeandinstillself-control.

Despitethestrengthsofthetheoreticalmodelanditspredictionsfortheetiologyofself-controlandofdeviance,andgiventhelimitationsofhowparentingandsocializationbyparentshavebeendiscussed,anumberoftheoreticallyunaddressedshortcom-ingsneedtobehighlighted.First,GottfredsonandHirschididnotdiscusstheimportanceofsociallearninginsufficientdetail,de-spitethefactthatmuchevidencesupportsthis(Dixon&Johnson,2007).Second,theydidnotdealwithwhatdevelopmentalistshaverecognizedforsometime,namely,thatthesocializationprocessmaybe“circular”orbidirectional(Lerner,2002).Third,largelyrelatedtoveryrecentinsightsgainedfromneuroimagingstudies,theydidnotaddressthelinkbetweenbiologicaldevelopmentormaturationalchangesandthedevelopmentofself-control(Stein-berg,2004;2008).Fourth,althoughtheoreticallyGottfredsonandHirschidiscussedtheimportanceoftheschoolcontextinthedevelopmentofself-control,thecurrenteffortdoesnotfocusonothersourcesofvariabilityinthedevelopmentofself-control(Turner,Piquero,&Pratt,2005).Finally,theyalsodidnotprovidedirectlinkstointerventionworkondevianceeventhoughtheirtheoreticalworkhasdirectimplicationsforit(e.g.,forsubstanceuseprevention;e.g.,Tolan,Szapocznik,&Sambrano,2006).

DevelopmentalStudies

Scholarsinthedevelopmentalscienceshavefocusedonhowself-controlandself-regulationdevelopininfantsandyoungchil-drenforsometime(Kopp,1982),althoughmostwork,includingworkonolderchildren,hasbeenconductedrecently(Colman,Hardy,Albert,Raffaelli,&Crockett,2006;Kochanska,Aksan,Penney,&Doobay,2007;Kochanska,Coy,&Murray,2001;Raffaelli,Crockett,&Shen,2005;Rothbart,Ellis,Rueda,&Posner,2003).Kopp’sseminalworkidentifiedtheearlyinfant–caregiverattachmentandreciprocalrelationshipaskeyinunder-standingandforetellingthedevelopmentofself-controlorself-regulationandhowthisprocessshiftsfrombeingexternaltotheinfanttobeinginternaltothechild.Morerecently,Eisenbergandcolleagues(2005)havearguedthatalthoughindividualdiffer-ences,suchastemperament,arepresentearlyinlife,continuedexperiencesfromcaregiversorparentscontributetothedevelop-mentandobserveddifferencesinself-controlorselfregulationduringchildhoodandadolescence.Theyhavealsopointedouthowdifferentterminologyandattimesoperationalizationshavebeenusedinthedevelopmentalliteraturetorefertoself-controlandselfregulation(Eisenberg,Champion,&Ma,2004).Inaddition,con-sistentwithearlyworkbyMischel,Shoda,&Rodrigues(1989),theirworkhassupportedlongitudinallinksbetweenself-controlandlateradjustmentproblems,includingexternalizingbehaviorsordeviance(Eisenbergetal.,2005;Zhouetal.,2007;seealsoBrody&Ge,2001;Crockett,Raffaelli,&Shen,2006).

EmpiricalTestsofSelf-ControlTheory

Verylittleempiricalworkhasexaminedhowself-controlde-velopsduringchildhoodorearlyadolescence.Asmallnumberofstudieshavetestedthestabilitythesisofself-control(discussedsubsequently),primarilyfocusedonyouthsoryoungadults,notonchildren.Othershavetestedthesalienceofparentingeffortson

This document is copyrighted by the American Psychological Association or one of its allied publishers.

自我控制与问题行为的关系什么呢？

SELF-CONTROLANDDEVIANCEOVERTIME

247

self-controlanddevianceorcrime,mostlyonthebasisofcross-sectionaldata(Hay,2001;Hope.,Grasmick,&Pointon,2003;Jones,Cauffman,&Piquero,2007;Latimore,Tittle,&Grasmick,2006;Perrone,Sullivan,Pratt,&Margaryan,2004;Pratt,Turner,&Piquero,2004;Unnever,Pratt,&Cullen,2003;Vazsonyi&Belliston,2007;Wright&Beaver,2005).Supportforthislatterlinkhasbeenquiteconsistent,althoughinsomeoftheseefforts,researchershavecommittedatheoreticalmisattributionbyhypoth-esizingparentingeffectsshouldbediminishedornotpresentduringadolescence;ineffect,nodirectparentingeffectsshouldremainonceself-controlisdevelopedduringchildhood.

Tothebestofourknowledge,nopreviousworkhastestedtheprocessofself-controldevelopmentduringthefirstdecadeoflife.

lyframedbyself-controltheory,amatterthatappearstobeatthedaoverycoreofthetheory.Ofcourse,workindevelopmentalsciencesr bdhasdoneso,aspreviouslyhighlighted.However,muchlikeworketaincriminologyonself-control,mostofthisworkemployedainmvariable-centeredanalyticapproachthatdidnotpermitthestudyofessdevelopmentalchangesinself-controlduringthefirstdecadeofi delife,acorefeatureandpredictionbyself-controltheory.

btot onATheoreticalTreatmentofWhereSelf-Control si dComesFrom

na reAnadditionalandequallyinterestingtheoreticalandempiricalsl uquestionrelatedtowhereself-controlcomesfromis“whereindi-auvidualsstart”—inotherwords,towhatextentdowefindindivid-idvidualdifferencesinself-controlpriortosocializationpressuresandni effects?ThoughnotdiscussedextensivelybyGottfredsonandeht Hirschi(1990),theyexplicitlyacknowledgedpre-existingindivid-fo eualdifferencesinself-control,againafrequentlymisunderstoods ufactintheliteraturesincethepublicationofthetheory.Indescrib-laningpotentialsourcesofvariationinself-control,theylistedtwosoresources:socializationpressuresbycaretakersand,priortothisin petermsofadevelopmentalsequence,“thevariationamongchildrenht rinthedegreetowhichtheymanifestsuchtraitstobeginwith”of y(Gottfredson&Hirschi,1990,p.96).Partofreasonwhytheysaidlelosolittleaboutpre-existingindividualdifferencesisrelatedtothes dbeliefbytheauthorsthateffectivesocializationcanoccurinde-ednpendentofinitiallevelsofself-control:“Effectivesocialization,etnhowever,isalwayspossiblewhatevertheconfigurationofindi-i si vidualtraits”(1990,p.96).Thus,ineffect,theyacknowledgedelciindividualdifferences,perhapsbiologicalpredispositions,buttra chosetofocusonpositivesocializationprocesses—intheend,thesihTonlyfruitfulavenueintermsofmeaningfulchange.

ExamplesofworkinwhichthisquestionhasbeenexaminedincludeastudybyBranniganandcolleagues(2002);theyfoundstrongevidenceofsimilaritiesindevelopmentalprocesses,namelyinthepatternsofassociationsbetweenparentingmeasures(hos-tility)andchildhoodhyperactivity(aproxyforself-control,thoughaclinicalmeasure),acrossdifferentagecohortspartoftheirstudy.Theauthorsconcludedthatnextstepsinresearchmustincludemodelsofintraindividualtrajectoriesofself-controlbutalsoas-sessmentsofpotentialgeneticorindividualdifferences,excludinganysocializationeffects.Again,todate,mostempiricaleffortsonself-controldevelopmenthavesimplyexaminedthestabilityofself-controlovertime.Inmorerecentrelatedworkonthelinksbetweenparentinganddeviance(basedonaCanadiannationalsampleinwhichaclusteranalyticapproachwasused),Pevalin,

Wade,andBrannigan(2003)foundthatparentingpredictedthelikelihoodofchildrenengaginginantisocialbehaviorsovertime.Finally,onthebasisofare-analysisoftheGluecks’data,Hard-wickandBrannigan(2008)providedfurthersupportforachildrearingmodelandthesalienceofself-controlinunderstandingdeviance,excludingtheeffectsofwhattheytermedsociogenicfactors(informalsocialcontrolorsocialcapital).Atthesametime,theyalsoconcludedthatontogeneticinfluencesappeartolargelysubsumesociogenicinfluencesinthedevelopmentofdeviance,furthersupportingmodelsthatincorporatebidirectionalsocializa-tionprocessesinchildrenandyouths.

WhereDoesSelf-ControlComeFromandWhereDoesItGo?

Onceself-controlis(orisnot)establishedbymiddletolatechildhood(bytheageof8or10)accordingtotheory,itshouldremainlargelystableoverthelifecourse.Thispredictiondoesnotnecessarilyimplythatonceself-controlisestablished,absolutelevelsofself-control(inindividuals)willremainstable,fixed,orimmutable.Infact,GottfredsonandHirschi(1990)suggestedthattherelativerankings(differences)betweenpeopleremainstablethroughoutlife;thismeansthatdifferencesinself-controlbetweenonepersonandanothershouldremainstable,andthus,anindi-vidual’srankorderingatage10shouldbesimilartothoseatages20,30,andsoforth.ThisalsomeansthatGottfredsonandHirschididnotdiscountthepossibilityofwithin-individualchangesoverthelifecourse(maturationalreform,age–crimecurve),anaddi-tionalandlargelymisunderstoodtheoreticalproposition;ofcourse,thisisinadditiontothespecificationofhowself-controldevelopsandchangesduringthefirstdecadeoflife.Amodestnumberofempiricaleffortshavesimplytestedthe(construct)stabilityofself-controlovertime;someresearchersconcludedthattheyfoundsupportfortheoreticalpredictions(Arneklev,Cochran,&Gainey,1998;Turner&Piquero,2002),whileothersdidnot(Burt,Simons,&Simons,2006;Mitchell&MacKenzie,2006;Winfree,Taylor,He,&Esbensen,2006).

Recently,andmostrelevantforthecurrenteffort,HayandForrest(2006)useddatafromtheChildandYoungAdultSupple-mentoftheNationalLongitudinalStudyofYouth(BureauofLaborStatistics,1986)http://wendang.chazidian.comingadata-driven,semiparametricgroup-basedanalysistomodeldivergentdevelopmentalpathwaysovertime,theyaddressedthestabilityquestionofself-controlandwhetherparentingcouldexplainvariabilityinself-controlbeyondchildhood.Theyfoundevidenceofstrongabsolute(rankorder)andrelative(growth/decline)stabilityinthemajorityofthesample(84%).Theyalsofoundthatself-controlstabilityseemedlargelyestablishedbyage7andthatparentingcontinuedto“matter”intoadolescence;theyinterpretedbothofthesefindingsasbeingtheoreticallyinconsistent.

Itappearsthatnoneofthepreviouseffortshaveconsideredthefulltheoreticalimplicationsinthedevelopmentoftheory-driven,testablehypothesesbasedonself-controltheory.Infact,thesta-bilityissueseemstobearelativelyunimportantissue.Self-controlmustdevelopinthefirstplace,andGottfredsonandHirschi(1990)havesuggestedthatitdoessoduringthefirstdecadeoflife.Thus,perhapsthemoreinterestingandpressingquestionbasedontheory

This document is copyrighted by the American Psychological Association or one of its allied publishers.

自我控制与问题行为的关系什么呢？

248

VAZSONYIANDHUANG

iswhetherthereappeartobedevelopmentalchangesduringthefirstdecadeoflifewhenchildrendevelopgreaterself-controlovertimeashypothesized.Arelatedquestionis,dosocializationpres-sures,broadlyconstruedsuchasparentingefforts,accountforvariabilityinthesehypothesizedchangesbutalsoforvariabilityininitiallevelsofself-control(attheonsetofthestudy,whichalsoimplicatesmechanismsofsocialization)?

Finally,aprediction—andagain,onelargelyoverlookedinthetheorybutalsooverlookedbyscholars—isrelatedtothe“devel-opment”ofdeviance;basedontheoreticalpredictionsrelatedtothedevelopmentofself-control,wewouldalsoexpectparalleldevelopmentalchangesindevianceovertime.Itfollowsthatifchildren,duetosocializationpressures,developgreaterself-.

lycontrolthatismeasurableovertime,thendeviancewillalsobedaomanifestasequaland“opposite”or,atleast,inverselyrelatedr bddevelopmentalchanges,suchthatwewouldexpecttofindmea-etasurabledevelopmentaldecreasesindevianceovertime;wewouldinmalsoexpectsignificantvariabilityexplainedatinitialstatusofessdeviancebyself-control.Onthebasisoftheseconsiderations,wei detestedinthecurrentinvestigationthefollowinginterrelatedhy- bpothesesinastructuralequationmodeling(SEM)andalatenttot ogrowthmodeling(LGM)analyticframework.

n si dnaTheCurrentInvestigation

resl ua1.ReplicationofStabilityThesis

uidviInitially,tocontextualizethecurrentstudywithpreviouswork,dniwedecidedtoexaminehowpreviouseffortshaveinterpretedand ehttestedtheself-controldevelopmentissue,namely,whetherthere fo existsstabilityinself-controlovertime(avariable-centeredap-esproach);wealsotestedtheparallelquestionsonmeasuresof ulanparentinganddeviance.

sore pe2.DevelopmentalChangesinSelf-ControlOverTime

ht rofA.Itwashypothesizedthatself-controlwouldincreaseover yletime(unconditionalgrowthmodel).

losB.Toaddresshowmuchparentinghadaffectedself-control dedpriortoTime1assessment(conditionalgrowthmodel),wealsonethypothesizedthatmeasuresofparentingwouldaccountforsignif-ni sicantvariabilityinself-controlatinitialstatusaswellassignificanti elamountsofvarianceinself-controlgrowthovertime.

citra sih3.DevelopmentalChangesinDevianceOverTime

TA.Itwashypothesizedthatdeviancewoulddecreaseovertime(unconditionalgrowthmodel).

B.Toaddresshowmuchself-controlhadaffecteddeviancepriortoTime1assessment(conditionalgrowthmodel),wealsohypothesizedthatmeasuresofself-controlwouldaccountforsignificantvariabilityindevianceatinitialstatusaswellassig-nificantamountsofvarianceindeviancechange(decline)overtime.

4.Growth-to-GrowthPredictiveRelationshipsOverTime

Itwashypothesizedthattheself-controlinterceptwouldposi-tivelypredictthedevianceslope,thattheself-controlintercept

wouldnegativelypredictthedevianceintercept,andthattheself-controlslopewouldnegativelypredictthedevianceslope.

Method

SampleandProcedures

ThedataforthisstudywerebasedontheNationalInstituteofChildHealthandHumanDevelopment(NICHD)EarlyChildCareResearchNetworkStudyofEarlyChildCare(2001;fordetailsaboutrecruitmentproceduresandsample,seetheoriginalarticle,whichislistedinthereferences).Familieswithnewbornswererecruitedfromhospitalsat10differentdatacollectionsitesacrosstheUnitedStates,anddatacollectioncommencedin1991.Thesamplingplanincludedrecruitingfamilies(a)inwhichmothersplannedtoworkortogotoschoolfulltime(60%)orparttime(20%)inthechild’sfirstyear,aswellassomewhoplannedtosayathomewiththechild(20%),and(b)whichwererepresentativeonmeasuresofdemographicdiversity(economic,educational,andethnic)ateachsite.Bothtwo-parentandsingle-parentfamilieswereincluded.Someexclusionarycriteriaincluded(a)mothersyoungerthan18atthetimeofthechild’sbirth,(b)familieswhodidnotanticipateremaininginthecatchmentareaforatleast3years,(c)childrenwithobviousdisabilitiesatbirth,and(d)moth-ersnotsufficientlyconversantinEnglish.Assessmentshavecon-tinuedtothepresentday.ThedatacollectionprotocolandstudywerereviewedbyasteeringcommitteesupervisedbytheNICHDandwerereviewedannuallybyinstitutionalreviewboardsofthe10participatinginstitutionsresponsiblefordatacollection.

Ofthe1,526familiesinitiallyrecruited,1,364(89%)http://wendang.chazidian.comrgelyduetotheoreticalandmeasurementconsiderationsaswellasmeasurementconsistencyforLGManalyses,wefocusedonthreeassessmentsofthechildren:atages4.5years(54-monthassessment),8.5years(3rdgrade),and10.5years(5thgrade).Datawereavailablefrom1,155families,whichrepresentedalossof209familiesfrombaselineassessment.Someadditionaldataweremissingonspecificconstructsinthefinalstudysampleduetononresponseorincompleteanswers;fullinformationmax-imumlikelihood(FIML)inAMOSsoftware(SPSS,Chicago,IL)wasimplementedtoaddressthis.Tofurthercontextualizethesample,weconductedattritionanalysesonthetwogroupsinwhichwecomparedkeydemographiccharacteristicsbetweenthefamilieslosttoattritionandthestudysample(childsex,childrace/ethnicity[AfricanAmericanvs.EuropeanAmerican],mater-nalage,maternalrace/ethnicity,maternaleducation,andfamilystructure[twoparentsvs.other]).Resultsprovidedevidencethatthenumberofboyslostversusthenumberofgirlslostdidnotdifferinthetwogroups(studysamplevs.lostfamilies:male:51.2%vs.54.5%;female:48.8%vs.45.5%,?2(1,n?1,364)?0.81,p?.21).Theydiddifferintermsofracial/ethnicdistribu-tion,withfewerAfricanAmericanyouthremaininginthesample(EuropeanAmerican:74.2%vs.81.6%;AfricanAmerican:19.1%vs.11.8%,?2(1,n?1,364)?8.64,p?.05).Childrenremaininginthesampleweremorelikelytohavebettereducatedmothers,M?14.40yearsversus13.33yearsofeducation,F(10)?32.38,p?.001,andmorelikelytobefromtwo-parentfamilies,78.6%versus66.3%,thanfromotherfamilystructures,21.4%versus33.7%,?2(1,n?1,364)?14.75,p?.001.Mothersofpartici-

This document is copyrighted by the American Psychological Association or one of its allied publishers.

自我控制与问题行为的关系什么呢？

SELF-CONTROLANDDEVIANCEOVERTIME

249

pantswerealsoslightlyolderonaverage,28.6yearsversus26.4years,F(1)?23.31,p?.001,andlesslikelytobeofminoritystatus,17%versus27%,?2(4,n?1,364)?10.49,p?.05,incomparisontomothersofnonparticipants.

Measures

Demographicvariables.Duringhomeinterviews,mothersreportedthechild’ssex(male:51.2%,female:48.8%)andthefamilystructure(subsequentlyupdatedateachassessment),whichwascodedas“two-parent”(twoparents,two-parentextendedorextendedandaugmentedfamilyandtwo-parentaugmentedfam-ily)and“other”inthecurrentstudy.Thesampleincluded78.6%two-parentfamilies(72.9%withtwobiologicalparents)and21.4%other.Mothersalsoreportedwhethertheirfamilyreceivedfoodstamps,whichwasusedasanindicatorofsocioeconomicstatus(SES,codedas0?lowSES[68.8%]and1?averageSES[31.2%]).

Parenting–childrelationshipquality(“parenting”).Toas-sesstheaffectivequalityoftheparent–childrelationship,weuseda15-itemshortformoftheChild–ParentRelationshipScalebyPianta(1995).Forconceptualreasons,aninterestinthewarmth/securitydimension,andreplicationsofthethreeoriginalsubscales(warmth/security,anger/dependence,andanxiety/insecurity)withprincipalcomponentsanalysis(PCA)datareduction,weretainedeightitems(Factor1)withconsistentlyadequatefactorloadingsoverthethreetimepoints;oneitemwasreversecoded(additionaldetailsavailableuponrequest).Itemswereratedona5-pointLikert-typescalerangingfrom1(definitelydoesnotapply)to5(definitelyapplies).Ascalescorewascomputedbyaveragingtheitems.Reliabilityanalysesprovidedevidenceofadequateinternalconsistencyacrossthethreetimepoints(?range:.63–.73).Table1includesdescriptivestatisticsofthestudyconstruct.

Self-control.MotherscompletedtheSocialSkillsRatingSys-tem(SSRS,Gresham&Elliott,1990),whichincludesaself-controlsubscaleconsistingof10itemsratedona3-pointLikert-typescale(0?never,1?sometimes,2?veryoften).Reliabilityanalysesprovidedevidencethatthescalewasinternallyconsistentovertheentirestudyperiod(?range:.79–.82;seeTable1foradditionalinformation).

Deviance.TheChildBehaviorChecklist(CBCL;Achenbach,1991)wasusedtoassesschildandearlyadolescentdeviancebasedTable1

DescriptiveStatisticsofMainStudyConstructs

VariableParenting4.5yrs8.5yrs10.5yrsSelf-control4.5yrs8.5yrs10.5yrsDeviance4.5yrs8.5yrs10.5yrs

No.ofitems888101010888

N1,0771,0281,0201,0571,0281,0211,0571,0261,020

M4.664.254.201.301.361.390.500.370.31

SD0.350.320.350.310.340.330.350.320.32

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

onmotherreports.Thefullmeasureincluded33itemsthatde-scribeavarietyofdeviantbehaviors,includinglying,cheating,stealing,andfighting.BasedonPCAdatareductionanalysesondatafromeachofthethreetimepoints,whichyieldedmultiplefactors,an8-itemscalewasdevelopedandusedinthecurrentstudy(additionaldetailsavailableuponrequest).Theseitemsloadedonasinglefactoranddidsoconsistentlyacrossassessmentpoints,withadequatefactorloadings.Itemswereratedona3-pointscale(0?nottrue,1?somewhatorsometimestrue,and2?verytrueoroftentrue),andparentsindicatedhowtrueeachstatementwasfortheirchildbasedonatimeframeofthepreced-ing2months.Thescalewasreliableateachtimepoint(?range:.78–.80;seeTable1).

Results

Table1includesdescriptivestatisticsforeachmeasure.Find-ingsprovidedsomeevidenceofnonnormality.Enders(2001)suggestedthattheFIMLprocedureinAMOSprovidesthebestestimatesfornonnormaldata;however,healsopointedoutthatmodelrejectionratesareinflatedbecauseofbiasedstandarderrorestimates.Becauseofsomeobservedskewnessinparentinganddeviancemeasuresbutalsobecauseofourdesiretobeconserva-tive,parentinganddeviancemeasuresweretransformedusinglogandexponentialfunctions(Tabachnick&Fidell,1996).Totestwhethertransformationaffectedstudyfindings,werepeatedmodeltestsbeforeandaftertransformations.Findingsprovidedevidencethatneithermodelfitnorparameterestimatesweremateriallyaffectedandthatresultswerehighlysimilarbasedbothontheoriginalandthetransformeddata(additionaldetailsavailableuponrequest).Thus,duetotheapparentrobustnessoffindingsde-spitesomeevidenceofnonnormality,theoriginaluntrans-formeddatawereusedinallsubsequentanalyses.Second,bivariatecorrelationsamongparenting,self-control,anddevi-ancewerecomputed.Resultsshowedthatallthecorrelationsforthethreetimepointswerestatisticallysignificantandintheexpecteddirections.Parent–childrelationshipqualitywaspos-itivelyassociatedwithself-controlandnegativelyassociatedwithdeviance,whileself-controlwasnegativelyassociatedwithdeviance(seeTable2).

Skewness?1.57?1.32?1.390.09?0.15?0.120.680.881.20

SE0.080.080.070.070.070.070.070.080.07

Kurtosis3.384.643.41?0.33?0.63?0.270.260.291.28

SE0.150.150.150.150.150.150.150.150.15

?0.630.660.730.790.820.810.780.790.80