[13年14页]儿童中期的行为问题：母体窘迫的预期作用、儿童依恋及母婴联结

JAbnormChildPsychol(2013)41:1311–1324DOI10.1007/s10802-013-9764-6

BehaviorProblemsinMiddleChildhood:ThePredictiveRoleofMaternalDistress,ChildAttachment,andMother-ChildInteractions

KarineDubois-Comtois&EllenMoss&ChantalCyr&KatherinePascuzzo

Published#online:10June2013

SpringerScience+BusinessMediaNewYork2013

AbstractTheobjectiveofthestudywastoexaminethelongitudinalrelationbetweenearlyschool-agemeasuresofmaternalpsychosocialdistress,qualityofmother-childin-teractions,andchildattachmentbehavior,andbehaviorproblemprofilesinmiddlechildhoodusingamulti-informantdesign.Participantswere243French-speakingmother-childdyads(122girls)whowerepartofanongoinglongitudinalproject.Maternalpsychosocialdistresswasassessedwhenchildrenwerebetween4and6yearsofage.Mother-childinteractivequalityandattachmentpat-ternswereobservedatage6duringalaboratoryvisit.Atage8.5,externalizingandinternalizingproblemswereassessedusingmotherandchildreports.Resultsshowthatmaternalpsychosocialdistresspredictedlatersocialadapta-tionreportedbythechildthroughthemediationofmother-childinteractions.Analysesalsorevealedthathighermater-nalpsychosocialdistressandcontrollingattachmentpat-terns,eitherofthepunitiveorcaregivingtype,significantlypredictedmembershipinbothchildinternalizingandexter-nalizingclinicalproblemgroups.Lowermother-child

K.Dubois-Comtois(*)

DepartmentofPsychology,UniversitéduQuébecàTrois-Rivières,C.P.500,

Trois-Rivières,QuébecG9A5H7,Canadae-mail:karine.dubois-comtois@uqtr.ca

K.Dubois-Comtois

HôpitalduSacré-CoeurdeMontréal,Montréal,Québec,CanadaE.Moss:C.Cyr

Departmentofpsychology,UniversitéduQuébecàMontréal,Montréal,Québec,Canada

K.Pascuzzo

Departmentofsocialpsychology,McGillUniversity,Montréal,Québec,Canada

interactivequality,malegender,andchildambivalentat-tachmentwerealsopredictorsofexternalizingclinicalproblems.

KeywordsMiddle.childhood.Behaviorproblems.ChildattachmentLongitudinalstudy.Maternaldistress.Mother-childrelationship.Mother-childinteraction

Betweentheagesof6and12,childrenexperienceimportantsocializationchallengesandcertainmaladaptedchildsubgroupsbecomeincreasinglydemarcated(Achenbach1984).Researchhasshownthatthedevelopmentofchildbehaviorproblemsinmiddlechildhoodisinfluencedbydifferentaspectsofthefamilyenvironment,includingmaternalpsychosocialfunctioningandthequalityofchildattachmenttoparent(DeKlyenandGreenberg2008).However,thewayinwhichthesefactorsimpactbehaviorproblemprofilesinmiddlechildhoodisunclear,evenmoresowhenconsideringdifferentattachmentprofiles.Ofparticularinterestisthebehavioraltrajectoryofchildrenclassifiedinthedisorganizedandcontrollingattach-mentsubgroups,high-riskgroupsforwhichlittleinknownduringthemiddlechildhoodperiod(Mossetal.2004)andforwhichtheriskofpsychopathologyissignificantlygreatercom-paredtootherattachmentgroups(Lyons-RuthandJacobvitz2008).Researchintheareaofchilddevelopmentalpsychopa-thologyhasconsistentlyshownthegreaterimpactofproximalvariables(i.e.,closelyrelatedtothechild)overdistalvariables(i.e.,affectingfamilymembersorthecommunity)onchildbehavioralmaladaption(CicchettiandValentino2006).Accordingly,theobjectiveofthecurrentstudyistwofold:1)toexaminethemediatingroleofmother-childinteractionqualityintheassociationbetweenearlymaternalpsychosocialdistressandchildbehaviorproblemsinmiddlechildhood;2)toexamineinmoredetailtheassociationbetweenchildclinicalbehaviorproblemsandthedistinctivemother-childattachmentpatterns

1312usingdifferentcategoriesofattachmentgroups,thatis,orga-nized,disorganizedandcontrolling.Afewyearsback,wereportedonthefamilycharacteristicsandchildbehavioraltra-jectoriesofdisorganizedandcontrollingchildrenbetweentheagesof2and5(Moss,CyrandDubois-Comtois2004).Thisstudyisalongitudinalfollow-upofthesechildrenatage8.5,thuscontributingtothecurrentlackofinformationonattach-mentdisorganizationduringmiddlechildhood.

MaternalPsychosocialFunctioning

Maternalpsychosocialfunctioninghasbeenshowntoplayanimportantroleintheemergenceofchildbehaviorproblems.Motherssufferingfrompsychopathologyaremorelikelytoshowalackofadequateparentalcare(Middletonetal.2009)andtotransmitbiologicalsusceptibilitiestotheirchildren(Haroldetal.2011).Ameta-analysis,including134studieswithchildrenbetweentheagesof2and18,showedthatmaternalmentalhealthproblemswereassociatedwithbothexternalizingandinternalizingproblemsintheiroffspring(ConnellandGoodman2002).Maternaldepressionandanx-ietywerethestrongestpredictorsofchildbehaviorproblems,withgreatereffectsizesforyoungerchildren(ages1–6years).However,mostofthestudiesincludedinthemeta-analysiswerecross-sectional,thuslimitingourunderstandingofthelong-termconsequencesofmaternaldepressionandanxietyonchildfunctioning.Inaddition,themajorityoflongitudinalstudieshavenotgonebeyondthepreschoolperiod(Bureauetal.2009a;Lyons-Ruthetal.1997;Munsonetal.2001).Onestudythatdidincludethekindergartenperioddidnotshowasignificantassociationbetweenmaternaldepressionduringthetoddler/preschoolperiodandthedevelopmentofchildinternalizingsymptoms(Essexetal.2001).Theinfancyperi-odhasbeenshowntobeacriticaldevelopmentalperiodduringwhichchildrenareparticularlysusceptibletotheneg-ativeconsequencesassociatedwithmaternaldepressionandanxiety(Bureauetal.2009a).However,theseassociationshavebeenunderinvestigatedinthepre/school-ageperiod.

QualityofMother-ChildInteractionsandChildAttachment

Studiesthathavetestedwhetherparent-childinteractionsandchildattachmentactasmediatingprocessesbetweenmoredistalparentalvariablesandchildadaptationduringmiddlechildhoodarescarce.Inonerecentstudyconductedovertheschool-ageperiod(childaged4–10years),mothers’percep-tionsoftheirpositiveandnegativebehaviorstowardstheirchildfullymediatedthelinkbetweenmotherandchildanti-socialbehaviorsanddepression(Haroldetal.2011).Apre-viousstudyalsofoundthatinconsistentmaternaldiscipline

JAbnormChildPsychol(2013)41:1311–1324

strategiespartiallymediatedthelinkbetweenmaternaldistressandchildaggressionatage9–12,whencontrollingforsocio-economicstatusandmaternalstress(Barryetal.2009).Bothstudiessuggestthatinadequatematernalbehaviorstowardthechildrepresentriskmechanismsforthetransgenerationaltrans-missionofmaladaptationinmiddlechildhood.However,alimitationofthesestudiesisthatallmeasureswereassessedusingmaternalself-reports,whichreflectonlymothers’per-ceptionofthechild’scharacteristicsandrelationshipwithher,andincreasespotentialcolinearitybetweenvariables.Arecentstudyconductedinthetoddlerperiodandincludingmotherobservationsshowedsignificantindirectassociationslinkingmaternaldepressionwithchildexternalizingbehaviorsviachild-criticism,withchildinternalizingbehaviorsviaself-andchild-criticism,andwithchildattachmentviaself-criticism(Graveneretal.2012).

Bowlby(1969/1982)theorizedthatdifferencesinthequal-ityofearlyrelationshipswithcaregivershasalong-termimpactonchildadaptationandmentalhealth,giventhatparent-childinteractionsshapechildren’sinternalworkingmodelsofselfandrelationshipwithothers.Mother-childinteractionisamu-tualprocessduringwhichthemotherguidesandsupportsthechildinexploringaffectsandthoughts,thusassistingintheorganizationofchildemotionalexperiences,andbyextensionthedevelopmentofemotionalandsocialabilities(Thompson2008).Studieshaveshownthatpreschoolchildrenwhohaveexperiencedsupportiveandsensitivecareandharmoniousin-teractionswiththeirmothersaremorelikelytodevelopasecureattachment(Dubois-Comtoisetal.2011;NICHD2001),bemoresociallyadaptatedandusemoreeffectiveemotionregu-lationstrategiesthantheirinsecurepeers(Easterbrooksetal.2012).Alternatively,childrenexperiencingunavailableorin-consistentcareandinvolvedindisruptedmother-childinterac-tionsaremorelikelytodevelopaninsecureattachment,symp-tomsofpsychopathology,andemotionaldysregulation(Carlson1998;Smeekensetal.2007).Someofthesechildren,inthefaceofmoreextremeinsensitivity,suchasfrighteningorfrightenedbehavioroftenobservedinmoredistressedcare-givers,areunabletosuccessefulyusetheircaregivertoregulateemotionandbehavior(Lyons-RuthandJacobvitz2008).Thesechildren,showingdisorganizedpatternsofattachmentatinfan-cyorpreschoolage,havefailedtodevelopacoherentstrategytogainaccesstothecaregiverintimesofstressandareconsideredtobethemostat-riskformaladaptationandpsy-chopathology(MainandSolomon1990).Itisnowestablishedthatoverthepreschoolperiod,aremarkableshiftoccursinthebehaviorofdisorganizedchildrensuchthattheydevelopcontrolling-typebehaviorswiththeparentofacaregivingorpunitivetype(Mossetal.2004).Childrenwhoareclassifiedascontrolling-punitiveusehostileanddirectivebehaviorswiththecaregiverthatmayincludeharshcommandsandoccasionalphysicalaggressiontowardtheparent;whereaschildrenclas-sifiedascontrolling-caregiving,whomalsodirecttheparent’s

JAbnormChildPsychol(2013)41:1311–1324activitiesandconversationalexchangesbystructuringinterac-tions,dosoinahelpfuland/oremotionallypositivecherrymanner.Disorganizedpreschoolerswhodonotdevelopthesecontrollingpatterns,classifiedasbehaviorallydisorganized,continuetodisplaydisordered,incompletemovements,confu-sion,andapprehensionwiththeparentandshownocoherentstrategyforseekingproximity.

Recentmeta-analyticreviewshaveshownthatattach-mentdisorganizationincreasestheriskofdevelopingexter-nalizingproblems(Fearonetal.2010)andismorestronglyassociatedwithexternalizingthaninternalizingsymptoms(Grohetal.2012).However,thesemeta-analyseshavenotprovidedinformationaboutthedifferentdisorganizedsub-groups.Theonlyfourstudiesthathaveassessedattachment-relatedbehaviorproblemsassociatedwiththedifferentdis-organizedsubgroupsbetweenpreschoolandmiddleschoolage(between3and9yearsolds)haveyieldedmixedresults(Bureauetal.2009b;Mossetal.2004;O’Connoretal.2011;Teti1999).Specifically,controlling-punitivechildrenhaveconsistentlybeendescribedbyteachersaspresentingmoreexternalizingproblemsthansecurechildren(Mossetal.2004;O’Connoretal.2011),butmothersconsistentlyreporthigherlevelsofbothinternalizingandexternalizingproblems(Bureauetal.2009b;O’Connoretal.2011;Teti1999).Asforbehaviorallydisorganizedchildren,higherlevelsofinternalizingandexternalizingproblemswerefoundforthesechildrenincomparisontosecurechildreninboththehome(Bureauetal.2009b;O’Connoretal.2011)andchilddaycare(O’Connoretal.2011)settings,whereasMossetal.(2004)foundonlymarginallygreaterlevelsofexternalizingproblemsintheschoolsetting.Asforcontrolling-caregivingchildren,theyhavebeenidentifiedbyteachersaspresentinghigherlevelsofinternalizingprob-lemsincomparisonwithsecurechildreninonestudy(Mossetal.2004),butressembledtheirsecurecounterpartsinthreeotherstudies(Bureauetal.2009b;O’Connoretal.2011;Teti1999).

Insecureavoidantandambivalentchildrenaregenerallyconsideredtobesituatedmidwayontheriskcontinuumbetweenthesecureanddisorganizedgroups.However,thereareconsiderableinconsistenciesintheliteratureregardingattachment/behaviorproblemassociationsforthesegroups.Amonginfancyandpreschoolstudiesusingafour-wayclassificationsystem,severalhavefoundavoidantchildrentoshowhigherlevelsofinternalizingproblemsbymiddlechildhoodthansecurechildren(Lyons-Ruthetal.1997;Mossetal.2006),whereasothersreportedgreaterextenalizingproblemsforthisgroup(FearonandBelsky2011;Munsonetal.2001).Infact,resultsofmeta-analysesrevealedthatavoidanceisassociatedwithhigherlevelsofinternalizing(Grohetal.2012;Madiganetal.2012)andexternalizingproblems(Fearonetal.2010),withaweakereffectforthelatter.Inaddition,nosignificant1313

differenceswerefoundbetweenstudiesusinginfantandpreschoolmeasuresofattachment.However,whenexamin-ingonlypreschoolsamples,avoidantchildrenwerenotmorelikelytoshowinternalizingorexternalizingproblems(Fearonetal.2010;Grohetal.2012).Withrespecttoambivalentattachment,studiesreportfewerassociations(FearonandBelsky2011;Lyons-Ruthetal.1997).Inpar-ticular,whenpreschoolorschool-ageattachmentclassifica-tionsareused,ambivalentchildrenaremorelikelytoshowexternalizingproblemsthantheirpeers(Cohn1990;Mossetal.2006).Despitethesesignificantresults,theambivalentclassificationhasnotbeenshowntobeariskfactorforbehaviorproblemsinmeta-analyticreviews(Fearonetal.2010;Grohetal.2012;Madiganetal.2012).However,whenexaminingonlypreschoolsamples,ambivalentchil-drenweremorelikelytoshowexternalizingproblems(Fearonetal.2010).

Insummary,areviewoftheliteraturesuggeststhatconsiderableambiguityremainsregardingtheindependentcontributionofmaternalpsychosocialfunctioningandmother-childrelationships,asshownalsoinstudiesexam-iningchild-to-motherattachmentrelationships,tothedevel-opmentofbehaviorproblemsinmiddlechildhood.Thelongitudinalprofilesofchildrenwithdifferentdisorganizedandcontrollingattachmentsubtypesareofparticularinter-est,giventhattheseprofileshavebeenunderinvestigated.Inaddition,ithasbeenwidelyacknowledgedthatexternalobserversmayunderestimatetheprevalenceofinternalizingdisorders(Youngstrometal.2000).Givenchildren’sin-creasingreluctancetoshareinternalizingsymptomswithadults,despitetheincreaseinsuchsymptomsduringmiddlechildhood,self-reportmeasuresbecomeincreas-inglyimportantinthedetectionofbehaviorproblems(Edelbrocketal.1985).Morespecifically,theobjectiveofthisstudyistoexaminethelongitudinallinkbetweenearlyschool-agemeasuresofmaternalpsychosocialdis-tress(age4and6),childattachment,andqualityofmother-childinteractions(age6),andbehaviorproblemsinmiddlechildhood(age8.5)usingamulti-informantdesign.

First,weexpectedthatmaternalpsychosocialdistress,poormother-childinteractionquality,andinsecureattach-mentwouldallsignificantlypredictchildexternalizingandinternalizingproblems,asreportedbyeithermotherorchild,andusingbothcontinuousandclinicalcut-offscores.Second,weexpectedqualityofthemother-childinterac-tionstoactasamediatorintheassociationbetweenmater-nalpsychosocialdistressandchildbehaviorproblem.Finally,wepredictedthatchildrenwithinsecureattachmentstrategies,andparticularlythedisorganizedandcontrollinggroups,wouldshowhigherlevelsofbothmother-reportedandchild-reportedinternalizingandexternalizingbehaviorproblemsthanwouldsecurechildren.

1314MethodParticipants

Participantsinthisstudyare243French-speakingmother-childdyads(122girls)whoarepartofanongoinglongitudinalprojectinvestigatingtheinfluenceofparent-childrelation-shipsondevelopmentaladaptation.Participantsinthisstudywerecombinedfromtwosamplesrecruitedin1988(sample1)and1996(sample2)throughpreschoolsindiversesocio-economicareasofMontreal(sample1:N=125;sample2:N=118).Familiesofbothsampleswererecruitedwhenthechil-drenwereofearlypreschoolageandreturnedfortwofollow-upvisitsapproximately2yearsapart.Therefore,labvisitstookplaceatthreedifferenttimepoints:Time1(Mage=4.0years,SD=0.68),Time2(Mage=6.0years,SD=0.92),andTime3(Mage=8.7years,SD=0.53).Attachmentatearlyschool-agebasedonseparation-reunionbehaviorwasassessedforchildreninbothsamplesatTime2.Participantsofthetwosamplescompletedaslightlydifferentsetofmea-suresofchildfunctioning.Therefore,onlycommonmeasuresfrombothdatacollectionswerecombined.Analysesperformedonbothsamplesrevealedthatfamilyincomeandmaternaleducationwashigherinsample2,χ2(2)=7.52,p<0.05andχ2(1)=15.29,p<0.01,respectively.Maritalstatusandchildgenderandagedidnotdiffer.Datafrombothsamples(notcombined),reportingonTime1,2and3,havebeenpublishedindependentlyinpreviouspapers(Mossetal.2005;Mossetal.2006).Bothsampleswerecombinedonlyforonepreviouspublication(Mossetal.2004)inwhichwereporteddatacollectedatTime1and2.Incontinuitywiththisonepreviouspublication,thecurrentstudyaddsthecombineddatacollectedatTime3.Withthisnewtimepoint,itwaspossibletoincludechildself-reportdata,allowingamulti-informantdesign,andtofollowchildrenintomiddlechild-hood,contributingtoabetterunderstandingoftheimpactofattachmentdisorganizedsubtypesduringthisunderstudiedperiod.Italsoallowedfortheinvestigationofnewresearchquestionsregardingmediatingmodels.

AtTime3,thefinalsampleisquiteheterogeneouswithrespecttobackgroundvariables,with13%offamiliesearningunder$20,000,38%between$20,000and$50,000,and49%earning$50,000ormore.Thirty-sevenpercentofmothershadonlyahighschooldegreeorless.Thirty-sixpercentofsamplechildrenwerelivinginamother-headedsingleparentfamily,whiletherestwerelivingwithbothparents.Thesampleconstitutesamiddle-income,low-riskgroup.MeasuresandProcedure

Motherswereinformedofthegeneralpurposeofthestudy,aswellasthenatureoftheirimplication.Theirwrittenconsentwasobtainedpriortodatacollection.Theprotocol

JAbnormChildPsychol(2013)41:1311–1324

wasapprovedbytheinstitution’sethicscommittee.Mothers’reportsofdepressionandparentalstressaswellasfamilybackgroundinformationwerecollectedatTime1andTime2whenchildrenwere4and6yearsofage.Childseparation-reunionbehaviorandmother-childinteractivequalitywasobservedatTime2duringalaboratoryvisitoccurringwhenchildrenwere6yearsold.Motherandself-reportedbehaviorproblemmeasureswerecollectedatTime3duringthesecondfollow-uplabvisit(age8–9).BeckDepressionInventory(Age4and6)Thisquestion-naireisawell-known21-itemmaternalself-reportmeasurethattapsaffective,cognitive,andphysiologicalsymptomsofdepression(Becketal.1961).Thetotalscorerepresentsboththeprevalenceandseverityofsymptoms;acutoffscoreof10issuggestedformilddepression,19formoderatedepression,and30forseveredepressiveillness.Thevalid-ityoftheBDIiswell-documented(Becketal.1988).ParentalStressInventory(Age4and6)Motherscompletedthe101-itemPSI,amaternalself-reportmeasurefocusingonsourcesofperceivedstressrelatedtotheparentalrole(Abidin1995).Inadditiontoanoverallstressscore,theinstrumentgeneratessubscalescoresinthematernalandchilddomains.ThePSIhasbeenwidelyusedandvalidityandreliabilityhavebeenwell-documented(Abidinetal.1992).

AttachmentProtocol(Age6)TheprocedureadoptedbyMainandCassidy(1988)forschool-agechildreninvolvingtwo45minseparationsandtwo5minreunionswasusedtoassesschildattachmentbehavior.Duringseparationschildrenandmotherscompletedindividualtaskinseparaterooms.Childrenattachmentclassificationwasgivenonthebasisofbehaviorobservedduringbothreunionperiods.Thevalidityofthisprocedureforclassifyingattachmentbehavior,withrespecttosocioemotionaladaptation,hasbeendemonstrated(Cohn1990;Mossetal.2004,2005;Solomonetal.1995).CodersclassifiedreunionbehaviorusingcriteriafromthePreschoolAttachmentClassificationSystem(PACS;CassidyandMarvin1992)forthe5year-oldsandtheMainandCassidysystem(1988)forthe6–7year-olds.Bothsystemsareconceptuallysimilarandusea6-categorycodingscheme.Thesecure(B)patternismanifestedwhenthechildrespondstothemother’sreturninaconfident,relaxed,andopenman-ner.Theinsecure-avoidant(A)patternischaracterizedbythechild’sneutralcoolnesstowardstheparentinclud-ingminimizingphysicalorverbalcontact.Intheinsecure-ambivalent/dependent(C)attachmentpattern,thechildshowsexaggeratedintimacyanddependencywiththeattachmentfigurethroughcute,babyishorangrybehav-ior.AsidentifiedbyMainandCassidy(1988),control-ling-caregiving(Ccare)behaviorismanifestedwhenthechildfocusesonhelpfullyguiding,orientingorcheering

JAbnormChildPsychol(2013)41:1311–1324uptheparent.Acontrolling-punitive(Cpun)childuseshostile,directivebehaviorwiththecaregiver,whichmayincludeverbalthreatsorharshcommands.Childrenareclas-sifiedbehaviorallydisorganized(D)iftheyseemunabletousethecaregiverasasecurebaseforexplorationbutdonotclearlyshowtheAorCpattern.Thesechildrenmaydisplaydisorderedsequencingofmovements,confusion,apprehen-sionoracombinationofotherinsecurepatterns.

Videotapedreunionswerecodedbytwocoderswhowereblindtoparticipantscoresonanyotherstudymeasures.CodersweretrainedbyR.MarvinorE.Mossandachievedreliabilitywiththeseexpertcodersonaseparatesampleoftapes.Discrepancieswereresolvedbyreviewingofthetapesuntilconsensuswasreached.Agreementforthemajorclassifications(A,B,C,D),calculatedon30%ofsamplecases,was88%(kappa=0.81)and80%fortheDsubtypes.Mother-ChildInteraction(Age6)Mothersandchildrenwereinvitedtoa10-minsnacktime.Toysandmagazineswereavailableandnoinstructionsweregiventothedyads.Thequalityofmother-childinteractionwascodedfromthevideotapedsnack-timeinteractions.Anoverallrating(fromhighquality[i.e.,harmonious,reciprocal]topoorquality[i.e.,indifferentorconflictual])andeight7-pointsubscales(Coordination,Communication,PartnerRoles,EmotionalExpression,Responsivity/Sensitivity,Tension,Mood,andEnjoyment)wereusedtocaptureglobalaspectsofparent-childinteractivequality,withhigherscoresconsideredmoreoptimalandscoresfrom1to3representingadysfunctionalpatterns.

Thescaleshavebeenshowntodistinguishthemother-childinteractivepatternsof3-to7-year-oldchildrenwithdifferentattachmentclassificationsandhaveshownconcur-rentandlongitudinalassociationswithbehaviorproblemratingsandhomebehavior(Dubois-ComtoisandMoss2004;Mossetal.2004).Snacktimelabtapeswereevenlydividedamongthreeobservers(blindwithrespecttootherfamilyinformation).Interraterreliability(intraclasscorrela-tions)calculatedon30%ofsamplecasesrangedfrom0.62to0.84forsubscaleratingsandfrom0.78to0.84fortheoverallrating.Becauseaprincipalcomponentsfactoranal-ysisofthe9mother-childinteractionscalesyieldedasinglefactorexplaining86%ofthevariance,onlytheoverallscalescorerepresentingreciprocated,balanced,andopencom-munication(atthehighend)wasincludedindataanalyses.Mother-ReportedBehaviorProblems(Age8–9)Motherscom-pletedtheChildBehaviorChecklist(CBCL;AchenbachandEdelbrock1983),whichconsistsof118itemsscoredona3-pointscale(nottrue,somewhatorsometimestrue,verytrueoroftentrue).TheCBCLgeneratesscoreson9psychopa-thologysubscales;withdrawn,somaticcomplaints,anxious/depressed,socialproblems,thoughtproblems,attention

1315

problems,delinquentbehavior,aggressivebehaviorandotherproblems.Theadditionofthedelinquentbehaviorandaggressivebehaviorsubscalesyieldsanexternalizingscore,whiletheadditionofthewithdrawn,somaticcom-plaints,andanxious/depressedsubscalesyieldsaninternal-izingscore.Clinicalcut-offpointsareestablishedatthe98thpercentile.Theinstrumenthasexcellentpsychometricqualities.

Self-ReportedBehaviorProblems(Age8–9)TheDominiquequestionnaire–Version5(Vallaetal.1994)requiresabout20–30minofadministrationtimeandispresentedtothechildorallybyanexperimenter.Drawingsdepictemotionalandbehavioralsituationsforwhichchildrenareaskedwhethertheyact,feel,think,orareliketheprotagonistDominique(answersareyes/no).TheDominiquequestion-nairegenerates2scales:theadditionoftheattentiondeficithyperactivity,conduct,andoppositionaldisorderssubcalesyieldsanexternalizingscore,whiletheadditionofthedepressive,separationanxiety,overanxious,andsimplephobiadisorderssubscalesyieldsaninternalizingscore.Thepresenceofsuicidalideationisalsotested.Diagnosticcut-offswereestablishedinalarge(N=1,575)representa-tivesampleofQuebecchildrenaged6to11,balancedforageandgender(Vallaetal.1994).Psychometricpropertiesareexcellent(Edelbrocketal.1985).

Results

MissingDataOfthe243participantsforwhomwehaveTime1andTime2measures,20%werenotpartofthefollow-up(n=49).One-wayANOVAsandchi-squareanal-ysesonchildattachment,backgroundvariables,andchildcharacteristics(ageandsex)comparingdyadswhopartici-patedatbothagesfromthosewhodidnot,revealedonlyasignificantdifferenceforchildage.ParticipantswhocamebackatbothtimeswereslightlyolderatTime2(meanage=6.0years,SD=0.9year)comparedwiththosewhodidnotreturnatTime3(meanage=5.7years,SD=0.6year),F(1,242)=5.34,p<0.05.AnanalysisoftheexistingdatasetwithLittle’sMissingCompletelyatRandom(MCAR)Testindi-catedthatthedatawasmissingcompletelyatrandom,χ2(37)=9.99,ns.MissingdatawasthereforereplacedbymeansofExpectationMaximization.Analysesconductedwithandwithoutreplacedmissingdatarevealedsimilarresults.

PreliminaryAnalysesTable1presentsmeansandintercor-relationsbetweencontinuousstudyvariables.Giventhatmaternalstress(PSI)anddepression(BDI)atTime1and2weremoderatelytohighlycorrelated(rsbetween0.44and