What are the associations between parenting, callous–unemotional traits, and antisocial behavior

16 Pages·2013·0.44 MB·English

by  Rebecca Waller

ClinicalPsychologyReview33(2013)593–608
ContentslistsavailableatSciVerseScienceDirect
Clinical Psychology Review
What are the associations between parenting, callous–unemotional
traits, and antisocial behavior in youth? A systematic review of evidence
Rebecca Waller a,⁎, Frances Gardner a, Luke W. Hyde b
aCentreforEvidence-BasedIntervention,DepartmentofSocialPolicyandIntervention,UniversityofOxford,UK
bDepartmentofPsychology,UniversityofMichigan,USA
H I G H L I G H T S
•DimensionsofparentingpredictCUtraitsinprospectivelongitudinalstudies.
•CUtraitsareresponsivetoparentinginparent-focusedinterventionsforAB.
•Theconstructof'CUtraits'asunresponsivetoparentingneedsrevisiting.
•FuturestudiesneedbettermeasurementapproachestoassessCUtraitsandparenting.
•Greaterprecisionisneededinstudydesigntoassessassociations.
a r t i c l e i n f o a b s t r a c t
Articlehistory: A growing body of research has examined callous–unemotional (CU) traits among samples of antisocial
Received24August2012 youth. Debate surrounds the malleability of CU traits and their responsiveness to parenting and
Receivedinrevised18February2013 parent-focusedinterventions.Thisreviewexaminesevidencefromstudiesthathaveinvestigatedvariousre-
Accepted1March2013
lationshipsbetweenparenting,CUtraits,andantisocialbehavior(AB).Studieswerecategorizedaccordingto
Availableonline14March2013 fivedistinctresearchquestionseachaddressingassociationsamongparenting,CUtraits,andABinadifferent
way.TheresultssuggestthatdimensionsofparentingareprospectivelyrelatedtochangesinCUtraits.Sub-
Keywords:
groupsofyouthwithbothhighlevelsofCUtraitsandABalsoappeartohaveexperiencednegativeparenting
Antisocialbehavior
Callous–unemotional practices.However,negativeparentingisnotconsistentlyrelatedtoABincross-sectionalstudiesforyouth
Conductproblems withhighlevelsofCUtraits.Atthesametime,parenting-focusedinterventionsappeareffectiveinreducing
Parenting thelevelofABandCUtraitsinyouth.Thefindingsandimplicationsforfuturestudiesarecriticallydiscussed
astheyposechallengesforcurrentetiologicaltheoriesofAB.
©2013PublishedbyElsevierLtd.
Contents
1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 594
1.1. CUtraits:overviewanddefinitions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 594
1.2. Parenting:theoreticalandempiricallinkswithABandCUtraits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 594
2. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 595
3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 595
3.1. Methodologicallimitationsofincludedstudies(seeTable2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 595
3.2. DoesparentingdirectlypredictlevelofyouthCUtraits?(Fig.1a). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 599
3.2.1. Earlychildhood/preschool(ages3–6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 599
3.2.2. Middle/latechildhood(ages7–12) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 599
3.2.3. Adolescence(ages13–18) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 600
3.2.4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 600
Abbreviations:AB,antisocialbehavior;APQ,AlabamaParentingQuestionnaire;APSD,AntisocialProcessScreeningDevice;CD,conductdisorder;CP,conductproblems;CU,
callous–unemotional;ODD,OppositionalDefiantDisorder;RCT,randomizedcontrolledtrial.
⁎ Correspondingauthorat:CentreforEvidence-BasedIntervention,DepartmentofSocialPolicyandIntervention,UniversityofOxford,32WellingtonSquare,Oxford,OX1-2ER,
UK.Tel.:[phone].
E-mailaddress:[email protected](R.Waller).
0272-7358/$–seefrontmatter©2013PublishedbyElsevierLtd.
http://dx.doi.org/10.1016/j.cpr.2013.03.001
594 R.Walleretal./ClinicalPsychologyReview33(2013)593–608
3.3. DoesparentingpredictyouthABatdifferentlevelsofCUtraits?(Fig.1b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 600
3.3.1. Earlychildhood/preschool(ages3–6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 600
3.3.2. Middle/latechildhood(ages7–12) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 601
3.3.3. Adolescence(ages13–18) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 601
3.3.4. Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 601
3.4. DoesparentingdifferbetweensubgroupsofyouthcategorizedaccordingtotheirlevelofABandCUtraits?(Fig.1c). . . . . . . . . . 602
3.4.1. Earlychildhood/preschool(ages3–6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 602
3.4.2. Middle/latechildhood(ages7–12) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 602
3.4.3. Adolescence(ages13–18) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603
3.4.4. Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603
3.5. DoparentinginterventionsdirectlyreducethelevelofyouthCUtraits?(Fig.1d). . . . . . . . . . . . . . . . . . . . . . . . . . . 603
3.5.1. Earlychildhood/preschool(ages3–6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603
3.5.2. Middle/latechildhood(ages7–12) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603
3.5.3. Adolescence(ages13–18) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604
3.5.4. Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604
3.6. DoCUtraitspredictormoderateeffectivenessofparentinginterventionsforAB?(Fig.1e) . . . . . . . . . . . . . . . . . . . . . . 604
3.6.1. Earlychildhood/preschool(ages3–6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604
3.6.2. Middle/latechildhood(ages7–12) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604
3.6.3. Adolescence(ages13–18) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605
3.6.4. Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605
4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605
4.1. Integrationacrossstudies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605
4.2. Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 606
4.3. Conclusionsandfuturedirections. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 606
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 607
1.Introduction psychopaths,includingreducedresponsivitytoothers'distress(Blair,
Colledge, Murray, & Mitchell, 2001) and insensitivity to punishment
Antisocialbehavior(AB)causesharmtoindividuals,families,and (Blair,Colledge,&Mitchell,2001).CUtraitsarealsoheritable(Viding,
communities, and is costly to society (Scott, Knapp, Henderson, & Blair,Moffitt,&Plomin,2005),associatedwithdifferencesinneuralre-
Maughan, 2001). Developing well-evidenced etiological theories and activity(e.g.,Marshetal.,2008),andappearmoderatelystable(e.g.,
identifying intervention targets are key research priorities. Develop- Dadds,Fraser,Frost,&Hawes,2005).
mentalpsychopathologyresearchhasconsistentlyhighlightedhetero- However,studieshavetypicallyassessedpreadolescentoradoles-
geneity in the cause, development, treatment responsiveness, and cent samples of males using cross-sectional designs. These caveats
long-termoutcomesbetweensubgroupsofantisocialyouth.Inthelast limitthegeneralizabilityoffindingsandmakeitdifficulttodrawcon-
15years,therehasbeenafocusonthepresenceofcallous–unemotional clusionsaboutcausality.StudiesinvestigatingthestabilityofCUtraits
(CU)traitsamongasubgroupofantisocialyouth.Asignificantbodyof havealsotendedtoassessadolescentmalesovershorttimeintervals,
researchhasexaminedmeasurementofCUtraits,associatedcognitive andreporthigherstabilitywhenthesameinformantisusedacross
andsocioemotionalcharacteristics,andtheneurobiologicalandgenetic assessments(Andershed,2010).Inaddition,whenstudiesinvestigate
profilesofantisocialyouthwithCUtraits(Frick&White,2008).Further- mean-levelversusindividual-levelstability,orbehaviortrajectories,
more,ithasbeenproposedthatCUtraitsbeincludedasaspecifierof there are significant fluctuations in the level of CU traits (see
conduct disorder (CD) in the forthcoming edition of the Diagnostic Fontaine, McCrory, Boivin, Moffitt, & Viding, 2011; Frick, Kimonis,
and Statistical Manual of Mental Disorders (DSM-5; Frick & Moffit, Dandreaux, & Farell, 2003). Indeed, the question of how malleable
2010), making them an important clinical target. A key debate has CUtraitsareisakeyfocusofthisreview.Atthesametime,research
alsofocusedonthemalleabilityofCUtraits.Studieshaveinvestigated intoCUtraitsatveryyoungagescouldhelptoidentifychildrenatrisk
whetherparentingpracticesarerelatedtothedevelopmentofCUtraits, ofsevereAB,andprovideinsightintorisk/protectivefactorsatatime
whetherparentinginterventionsforABarelesseffectiveforyouthwho whenpersonalitystructureisstilldevelopingandmaybemostame-
havehighlevelsofCUtraits,andwhetherCUtraitsimprovefollowing nabletointervention.
parentinginterventions.Thisreviewseekstosynthesizeandevaluate
findingsfromobservationalandinterventionstudiesthathaveinvesti- 1.2.Parenting:theoreticalandempiricallinkswithABandCUtraits
gatedassociationsbetweenparenting,CUtraits,andABinyouth.
Researchovermanydecadeshashighlightedtheimportanceofpar-
1.1.CUtraits:overviewanddefinitions entingtoABdevelopment,includingcoerciveparent–childinteractions
(Patterson, 1982), poor parental supervision (Loeber, Farrington,
In adults, psychopathy refers to a complex syndrome, which in- Stouthamer-Loeber,&VanKammen,1998),andlowpositiveparent–
cludesinterpersonal,affective,andlifestyletraits,andidentifiesmore child engagement (Gardner, Ward, Burton, & Wilson, 2003). A key
severeandviolentoffenders(Hare,1991).Inrecentyears,itsdown- question,however,iswhetherparentingisrelatedtothedevelopment
wardextensiontoyouthhasbeendominatedbyafocusoninterperson- ofCUtraits.Inaddition,whiletherehavebeenmanyinvestigationsinto
al(e.g.,deceitfulnessandmanipulativeness)andaffective(e.g.,shallow the effectiveness of intervention programs for AB in general, less is
affectandlackofguilt)dimensions,whichhavemostcommonlybeen knownaboutpreventionandtreatmentwhenyouthshowalackofem-
labeledasCUtraits(Frick,2004;Frick,O'Brien,Wootton,&McBurnett, pathyandlowaffectiveresponsivity(Frick,2001).
1994).Thisfocusissupportedbyresearch,whichsuggeststhatCUtraits Ononehand,childrenwithCUtraitsappearlesssusceptibletopa-
alsodesignateahigh-riskgroupofantisocialyouth.Forexample,CU rental socialization and discipline efforts (e.g., Oxford, Cavell, &
traitsareassociatedwithparticularlyseverepatternsofAB(seeFrick Hughes,2003).Specifically,CUtraitsarehypothesizedtoproduceanat-
&White,2008).ExperimentaldatasuggestthatyouthwithCUtraits tenuatedresponsetopunishmentcuesorthedistressofothersfollowing
show a neurocognitive and socioemotional profile similar to adult behavioraltransgressions.Failuretomaketheseassociationsappearsto
R.Walleretal./ClinicalPsychologyReview33(2013)593–608 595
disruptthedevelopmentofempathyandmoralsocialization,therebyin- Australia. To clarify associations between parenting, CU traits, and
creasingthelikelihoodofaggressivebehavior(Blair,Peschardt,Budhani, youth AB, five research questions were identified among studies
&Pine,2006;Kochanska,1997).Ontheotherhand,thereisevidence (Fig. 1a–e): (a) Does parenting directly predict level of youth CU
thatparentalharshnessandlowwarmtharerelatedtoCUtraits(Frick traits?(b)DoesparentingpredictyouthABatdifferentlevelsofCU
etal.,2003;Pardini,Lochman,&Powell,2007).Harsh punishmentis traits?(c)Doesparentingdifferbetweensubgroupsofyouthcatego-
thoughttoelicithighlevelsofarousal,makingitdifficultforchildren rizedaccordingtotheirlevelofABandCUtraits?(d)Doparentingin-
to internalize parental messages about prosocial behavior (Pardini et terventions directly reduce the level of youth CU traits? (e) Do CU
al.,2007).Incontrast,parentalwarmthandresponsivenessmaywork traitsmoderateorpredicttheeffectivenessofparentinginterventions
againstthedevelopmentofABbypromotingempathyandprosociality, foryouthAB?
particularlyinchildrenwithfearlesstemperaments(Kochanska,1997). Giventhedistinctresearchquestionsaddressedacrossstudies,range
However, developing a precise understanding of how parenting, CU of analytic techniques, and heterogeneous measurement approaches,
traits,andABarerelatedinyouthhasbeenhamperedbystudiestesting meta-analysiswasnotfeasible.Studieswerethereforereviewedina
markedlydifferentresearchquestionsviadifferentdesigns,invarying narrativesynthesisaccordingto:(a)researchquestion;and(b)ageof
samples,andusingarangeofassessmentmethods.Thissystematicre- youthbasedonmeanageofsampleandgivenretainedstudieshadan
viewaimstosynthesizetheresultsofstudiesthathaveinvestigatedvar- upper age range of 18years (early childhood/preschool, 3–6years;
iousassociationsbetweenparenting,CUtraits,andAB.Specificallythe middle/latechildhood,7–12years;adolescence,13–18years).Several
reviewexaminesseveralquestions,includingwhetherparentingisre- studiesexaminedmorethanoneresearchquestion,andthereforeap-
latedtoCUtraits,whetherparentinginterventionsforABareeffective pearinmultiplesectionsofthereview.Asummaryofthemainfindings
foryouthwithCUtraits,andwhethertherearedevelopmentaldiffer- andkeylimitationsforeachstudyispresentedinTable2.Studieswere
encesintheextentortypeofassociationsbetweenparenting,CUtraits, systematically and critically appraised for methodological limitations
andAB. accordingto:(a)samplesize;(b)typeofsample(e.g.,communityvs.
clinic-referred; male vs. female); (c) measurement of CU traits; and
2.Methods (d)measurementofparenting.
Thereviewwasbasedonasystematicsearchofeightdatabases,in- 3.Results
cludingMEDLINE,PsycINFO,andDissertationsandThesesAbstracts.
Thesearchstrategycombinedtermstoidentifystudiesinvestigating 3.1.Methodologicallimitationsofincludedstudies(seeTable2)
CUtraitsandparentingamongyouth:(adolescen*ORboy*ORchild*
ORgirl*ORinfant*ORjuvenile*ORpreadolescen*ORpre-adolescen* Withinlongitudinalstudies,samplesizestendedtobelarge(8stud-
ORpreschool*ORpre-school*ORschoolchild*ORtoddler*ORteen* ieswithN>500),whereaswithincross-sectionalstudies,samplesizes
ORyoungORyouth)AND(callous*ORpsychopathyORpsychopathic weresmaller(8studieswithNb200).Inseveralcross-sectionalstud-
ORpsychopathORsociopath*ORunemotional)AND(caregiver*OR ies, sample sizes were very small (Nb100). Treatment studies also
care-giver*ORfamil*ORfather*ORmaternalORmother*ORparent* tendedtohavesmallsamplesizes(7studies,Nb200).Theriskofre-
ORpaternal).Nodate,publication,orlanguagerestrictionswereim- duced generalizability of findings due to small samples, particularly
posed.Abstractsofidentifiedarticleswerescreenedforthefollowing forcross-sectionalandtreatmentstudies,needstobeconsideredalong-
inclusioncriteria:(a)presentationofdatatestingcross-sectionalor sideresults.Withinthe12longitudinalstudies,5examinedcommunity
longitudinalassociationsbetweenparenting,CUtraits,andABusing sampleswithequalnumbersofmalesandfemales,whichwerealso
validatedorpreviouslypublishedmeasuresofparentingandCUtraits; reported as reflecting the ethnicity of the general population. The
and(b)assessmentofyouthwithanupperagerangeof18yearsold, remaining seven studies assessed at-risk or aggressive youth, and
from community, school, clinic, or research settings. Treatment tended to comprise majority Caucasian or African-American partici-
studiesmeetingthesecriteriawereretainedif:(a)treatmentfocused pants. One longitudinal study assessed only females (Kroneman,
on changing parenting behavior or parent–child interactions; and Hipwell, Loeber, Koot, & Pardini, 2011) and another, only males
(b)CUtraitsweretestedasanoutcomeormoderatorofeffectiveness. (Pardini&Loeber,2008).Ingeneral,becauseofsizeandtypeofsample,
Norestrictionswereplacedondesign,exceptthatstudiesrelyingon longitudinal studies appear to have greater generalizability. Within
youthretrospectivereportsofparentingwereexcluded. cross-sectionalstudies,fivewereclinic-referredoraggressivesamples,
Thesearchidentified2606records.First,titlesandabstractsofall and were either all or majority male. Necessarily, treatment studies
identifiedrecordswerescreened.Studieswereretainedforfurtherin- assessedclinic-referredoradjudicatedsamples,andthreeintervention
spectioniftheycorrespondedtotheaimsofthereview.Fromthere, studieswereconductedusinghigh-riskyouthwithincommunityset-
the full texts of 64 potentially relevant studies were examined to tings.Nevertheless,withoneexception(Hydeetal.,2013),intervention
assess whether they met the inclusion criteria of the review. After studiesassessedmajority/exclusivelymalesamples.
inspecting these 64 papers, 34 were removed (typically, these did Acrossthe30studies,24usedtheAntisocialProcessScreeningDe-
notassessparentingorCUtraits).Severalstudiesinvestigatingmod- vice (APSD; Frick & Hare, 2002) or its predecessor, the Psychopathy
eration of the effectiveness of non-parenting interventions by CU ScreeningDevice(PSD),toassessCUtraits.TheAPSDisa20-itemmea-
traits(e.g.,Waschbusch,Carrey,Willoughby,King,&Andrade,2007) sure,whichincludesa6-itemCUtraitsubscale.Ithasbeenvalidated
or treatment of incarcerated psychopathic youth (e.g., Caldwell, across different formats, including parent and teacher rating and
Skeem, Salekin, & Van Ryoboek, 2006) were also identified but not self-report scales. Nevertheless, there are various well-documented
retained.Twostudiesthatreliedonretrospectivereportsofparenting drawbacks of the CU trait subscale of the APSD, including the small
by incarcerated adolescents (Edens, Skopp, & Cahill, 2008; Fritz, numberofitems(n=6),poorinternalconsistencyofsubscales,and
Ruchkin,Kaposov,&Klinteberg,2008)wereexcludedbecauseofthe limitedrangeofresponseoptions(n=3),whichreducesthevariance
difficultyofobtainingreliablereportsofparentingfromthistypeof ofscoresobtained(Kotler&McMahon,2010).Despitetheselimitations,
sample.Thefinalpoolof30studiescomprised26differentsamples, thefactthatsomanystudiesusedtheAPSDtoassessCUtraitsenables
and includes research published between 1997 and 2013, and one greatercomparabilityofresults.
unpublisheddissertation. Of the six studies that did not use the APSD, two assessed
Table1summarizesstudycharacteristics.Of30studies,12were ‘deceitful-callous behavior’ in the same sample of preschool children
longitudinal, 10 cross-sectional, and 8 investigated parenting inter- (Hyde et al., 2013; Waller, Gardner, Hyde, Shaw, Dishion, & Wilson,
ventions. All but three studies were carried out in the US, UK, or 2012),oneassessed‘interpersonal-callousness’inadolescents(Pardini
596 R.Walleretal./ClinicalPsychologyReview33(2013)593–608
Table1
Characteristicsofincludedstudies.
Study Country Sample Typeofsample %female Agerange Ethnicity Measureofparenting MeasureofCUtraits
size (years)
Longitudinalstudies
Barkeretal.(2011) UK 6673 Representative 43 2,4&13 Mixed,majorityCaucasian Questionnaire(P) 6-itemquestionnaire,
CUtraits(P)
Fontaineetal.(2011) UK 9578 Representative 53 4,7,&12 PredominantlyCaucasian Questionnaire(P) 3CUitems,APSD;
4prosocialitems,SDQ(T)
Fricketal.(2003) USA 98 Aggressive 47 8–12 Mixed,majorityCaucasian APQ&questionnaire 6CUitems,APSD(PT&Y)
(P&Y)
Hawesetal.(2011) Aus 1008 Representative 47 3–10 PredominantlyCaucasian APQ(P) 3CUitems,APSD;
4prosocialitems,SDQ(P)
Kimonisetal.(2004) USA 98 Aggressive 47 8–12 Mixed,majorityCaucasian APQ(P&Y) 6CUitems,APSD(PT)
Kronemanetal.(2011) USA 1233 High-risk 100 7–8 Mixed,majority Questionnaire(P) 4CUitems,APSD(PT)
African-American
Larssonetal.(2008) UK 4430 Representative 47 3–4 PredominantlyCaucasian Questionnaire(P) 3CUitems,APSD;
4prosocialitems,SDQ(T)
Muñozetal.(2011) USA 98 Aggressive 47 12–16 Mixed,majorityCaucasian APQ(P) 6CUitems,APSD(Y)
PardiniandLoeber(2008) USA 506 High-risk 0 13–14 Mixed,majority Questionnaire(PY) 8items,ICquestionnaire(P)
African-American
Pardinietal.(2007) USA 120 Aggressive 41 9–12 Mixed,majority APQ&questionnaire 6CUitems,APSD(PT)
African-American (P&Y)
Vidingetal.(2009) UK 4508 Representative 54 7,12 PredominantlyCaucasian Questionnaire(P) 3CUitems,APSD;
4prosocialitems,SDQ(P&T)
Walleretal.(2012) USA 731 High-risk 49 2,3,4 Mixed,majorityCaucasian Observed& 5items,DCbehavior
questionnaire(P) questionnaire(P)
Cross-sectionalstudies
Enebrinketal.(2005) Swe 41 Clinic-referred 0 6–13 PredominantlyCaucasian Interviews&case 6CUitems,APSD(P)
records(P&A)
FalkandLee(2011) USA 208 Clinic-referred 30 6–9 Mixed,majorityCaucasian APQ(P) 6CUitems,APSD(P)
Hipwelletal.(2007) USA 990 High-risk 100 7–8 Mixed,majority Questionnaire(P) 4CUitems,APSD(PT)
African-American
KoglinandPetermann Lux 177 Representative 50 5–7 Mixed,majorityCaucasian APQ(P) 6CUitems,APSD(P)
(2008)
Loneyetal.(2007) USA 83 Representative 53 7–13 Mixed,majorityCaucasian APQ(P) 6CUitems,APSD(PT)
Oxfordetal.(2003) USA 199 Aggressive 35 8–10 Mixed,majorityCaucasian APQ(P) 6CUitems,APSD(PT)
Pasalichetal.(2011) Aus 95 Clinic-referred 0 4–12 PredominantlyCaucasian Observed&speech 3CUitems,APSD;
samplescoding(O&P) 4prosocialitems,SDQ(PTY)
Vitaccoetal.(2003) USA 136 Representative 100 10–15 Hispanic APQ(Y) 6CUitems,APSD(Y)
Woottonetal.(1997) USA 166 Clinic-referred 24 6–13 PredominantlyCaucasian APQ(P) 6CUitems,PSD(PT)
Yehetal.(2011) USA 1210 Representative 52 9–10 Mixed,majorityCaucasian Parentaffect CPS(P)
questionnaire(Y)
Treatment/interventionstudies
HawesandDadds(2005) Aus 49 Clinic-referred 0 4–8 nr Observed& 3CUitems,APSD;
questionnaire(O&P) 4prosocialitems,SDQ(P)
HawesandDadds(2007) Aus 49 Clinic-referred 0 4–8 nr Observed& 3CUitems,APSD;
questionnaire(O&P) 4prosocialitems,SDQ(P)
Hydeetal.(2013) USA 731 High-risk 49 2,3,4,5 Mixed,majorityCaucasian Observed& 5items,DCbehavior
questionnaire(P) questionnaire(P)
KolkoandPardini(2010) USA 177 Clinic-referred 19 6–11 Mixed,majorityCaucasian na 6CUitems,APSD(T)
Kolkoetal.(2009) USA 139 Clinic-referred 15 6–11 Mixed,majorityCaucasian na 6CUitems,APSD(T)
McDonaldetal.(2011) USA 66 High-risk nr 4–9 Mixed,majorityCaucasian Questionnaire(P) 16-itemtotalPSD(P)
SomechandElizur(2012) Israel 209 High-risk 20 2–5 PredominantlyIsraeli APQ-Revised(P) 3CUitems,APSD;
8items,ICU(P)
White(2010)and USA 134 Clinic-referred 28 11–17 Mixed,majority APQ(P&Y) ICU(Y)
Whiteetal.(2012) African-American
Note.Forinformantonmeasures:P,parent;Y,youth;T,teacher;PT,combinationofparentandteacher;PC,combinationofparentandchild;O,observed;A,alternativeapproach.
APSD,AntisocialProcessScreeningDevice;APQ,AlabamaParentingQuestionnaire;CPS,ChildPsychopathyScale;DC,deceitful-callousbehavior;IC,InterpersonalCallousness;ICU,
InventoryofCallous–UnemotionalTraits;PCL-YV,PsychopathyChecklist-YouthVersion;PSD,PsychopathyScreeningDevice.Forlongitudinalstudies,agerangereportedforfirst
assessment,orforageatoutcome(ifrelevant).
&Loeber,2008;Pardini,Obradovic,&Loeber,2006),oneusedamodified measureCUtraits.Finally,White(2010)1assessedCUtraitsamongad-
versionoftheChildPsychopathyScale(CPS;Lynam&Gudonis,2005)to judicatedadolescentsusingtheInventoryofCallous–Unemotionaltraits
assess psychopathy in middle childhood (Yeh, Chen, Raine, Baker, & (ICU;Frick,2004).SomechandElizur(2012)alsouseditemsfromthe
Jacobson, 2011), and one used a CU trait scale similar to the APSD preschoolICUcombinedwithAPSDitems.
(Barker, Oliver, Viding, Salekin, & Maughan, 2011). Several studies
(e.g.,Hawes&Dadds,2005;Viding,Fontaine,Oliver,&Plomin,2009)
usedthreeoftheAPSDCUtraititems,andcombinedthemwithfour
(negativelycorrelating)itemsoftheProsocialBehaviorsubscaleofthe 1 Apeer-reviewed/publishedversionisnowavailable(Whiteetal.,2012).Thepub-
StrengthsandDifficultiesQuestionnaire(SDQ;Goodman,1997).Many lishedversionofthepaperdoesnotincludeanalysesusingparentingmeasures,which
wereincludedintheoriginaldissertation(White,2010).However,thepublishedpa-
ofthesealternativesaretherefore‘home-grown’measurescomprising
perincludes useful analytic approachesforexaminingchangeinCUtraits andAB
questionnaire items within studies that did not originally set out to scores.Assuch,botharereferenced.
R.Walleretal./ClinicalPsychologyReview33(2013)593–608 597
a) b)
c) d)
e)
f)
g)
Fig.1.Diagrammaticrepresentationsofstudydesignsincludedinthereview(a–e)andproposedalternativetreatmentdesigns(f&g).a)Doesparentingdirectlypredictlevelof
youthCUtraits?b)DoesparentingpredictantisocialbehaviorregardlessofthelevelofyouthCUtraits?c)Doesparentingdifferbetweensubgroupsofyouthcategorizedaccording
tolevelortrajectoryofantisocialbehavior(AB)andCUtraits?d)DoparentinginterventionsdirectlyreducethelevelofyouthCUtraits?e)DoCUtraitspredictormoderate
effectivenessofparentinginterventionsforyouthAB?f)DoparentinginterventionssimultaneouslyreducelevelofyouthCUtraitsandAB?g)Dointerventionsthattargetspecific
dimensionsofparentinguniquelyreducelevelofyouthCUtraitsversusAB?
Acrossall30studies,23usedparentreporttoassessCUtraits.Of somestudies.Howeverwhilethereareadvantagestothisapproach,
these23studies,13reliedsolelyonparentreportandtheremaining it also leads to loss of scale- and item-level information. Likewise,
10combinedparentratingswithteacherorchildratings(orboth). while cut-off scores are used in various studies, the validity of a
Commonly,thisapproachinvolvessummingratingsatanitemlevel ‘high’versus‘low’distinctionforCUtraitsamongyouthisyettobe
or creating a best estimate score, which combines ratings by using established.Toretainscale-anditem-levelinformation,futurestudies
thehigherscorefromreportersforeachitem.Inanalternativeap- could also create latent variable scores, which combine parent-,
proach, scores were obtained by calculating the proportion of re- teacher-,andself-reports,andcouldevenincorporateofficialreports,
porterswhoclassifiedthechildashighonCUtraits(Pasalich,Dadds, including court records (see Trentacosta, Hyde, Shaw, & Cheong,
Hawes, & Brennan, 2011). The use of the best estimate approach 2009).Finally,sevenofthe30reviewedstudiesprovidednodataon
helps incorporate multiple informants, provides a cross-context as- theinternalconsistencyoftheirCUtraits'measure.Ofthe23studies
sessment of behavior, may increase validity, and is thus favored in thatdidprovideCronbach'salphas,10werereportedasbeingb.70,
598 R.Walleretal./ClinicalPsychologyReview33(2013)593–608
Table2
Resultsandmajorrisksofbiasofincludedstudiespresentedaccordingtoresearchquestion/design(seeFig.1).
Study Keyfindingsrelevanttoaimsofreview Mainmethodologicallimitations
a)DoesparentingdirectlypredictlevelsofyouthCUtraits?
Barkeretal.(2011)a L Harshparentingatage4predictedboys'CUtraitsatage13. Measuresallparentreport;noalphareported
forCUtraitmeasure
Fricketal.(2003)a L Youthandparentreportsofnegativeparentingpartially Smallsamplesize;selectedtooverrepresent
correlatedwithCUtraits4yearslater. youthwithCP
Hawesetal.(2011) L Parentalpositivereinforcement,poormonitoring/ Measuresallparentreport
supervision,involvementpredictedCUtraits
Loneyetal.(2007) CS Parentingdysfunctionmediatedassociationbetween Relianceonparentreport;noalphasreported
maternalpsychopathyandCUtraits
McDonaldetal.(2011)a L Parentpsychologicalaggressionandinconsistent LowalphaforCUtraitmeasure(.34);
parentingrelatedtopsychopathicfeatures measuresallparent-report
PardiniandLoeber(2008) L Poorparent–childcommunicationpredictedhigher CUtraitsmeasurefromarchivedparent-reported
initialandfinallevelsofCUtraits items;malesample
Pardinietal.(2007)a L Parent-reportedpunishmentandchild-reported Lowalphaformeasureofcorporalpunishment(.29)
parentalinvolvementpredictedCUtraits
Vidingetal.(2009) L NegativeparentaldisciplinepredictedCUtraitsbut Lowalphasformeasuresofnegativeparental
notMZtwindifferencesinCUtraits disciplineandCUtraits
Vitaccoetal.(2003) CS Poormonitoringandinconsistentdisciplinenotrelated Allfemale,Hispanicsample;youthreportonly;
toCUtraits noalphasreported
Walleretal.(2012) L Observedandparent-reportedharshness(notobserved LowalphaforCUtraitmeasureatages2(.57)
positiveparenting)predictedCUtraits. and3(.64)
b)DoesparentingpredictyouthABatdifferentlevelsofCUtraits?
FalkandLee(2011) CS In1/8modelstested,lowerpositiveparentingpredicted Relianceonparentreport
CDsymptomsatlowCUtraits.
Hipwelletal.(2007) CS CPassociatedwithharshpunishmentandlowparental 4itemsonlyforCUtraitmeasure;lowalpha(.60);
warmthatlowlevelsofCUtraits allfemalesample
Hydeetal.(2013)a L CUtraits-parentinginteractiondidnotpredictgrowthin Low-moderatealphaforCUtraitmeasureat
CPfromages2–4. age3(.64)
KoglinandPetermann(2008) CS Inconsistentdisciplineassociatedwithboys'aggressive Measuresallparentreport;noalphasreported
behavioratlowlevelsofCUtraits.
Kronemanetal.(2011) L Lowparentalwarmthpredictedhigherinitiallevelsof Measuresallcomprisedparentreport;lowalpha
CD/ODDandfasterdecreasinglevelsofCD/ODDover formeasurementofCUtraits(.60);femalesample
5yearsathighlevelsofCUtraits.
Muñozetal.(2011) L ForlowCUtraitsgroup,higherparentalcontrolpredicted Smallsample;largeagerange;selectedtoover
knowledge,andhighersolicitationpredictedcontroland representyouthwithCP;only1yearfollow-up
lessknowledgeledtoincreasedparentalcontrol
Pardinietal.(2007)a L CUtraits-parentinginteractionsdidnotpredictAB1yearlater Lowalphaformeasureofcorporalpunishment(.29)
Pasalichetal.(2011) CS Maternalandpaternalobservedcoercionrelatedtoboys' Small,clinic-referredsamplewithlargeagerange
CPatlowlevelsofCUtraits.Maternalwarmthrelatedto
boys'CPathighormeanlevelsofCUtraits
Woottonetal.(1997) CS In3/6interactions,ineffectiveandlowpositive Small,clinic-referredsample;largeagerange;
parentingwererelatedtoCPatlowCUtraits. parentreport
Yehetal.(2011) CS Lowerpositiveandhighernegativeparentalaffectpredicted Lowalphasformeasurementofnegativeparental
reactiveaggressionatlowCUtraits.Highernegativeparental affect(male,.53;female,.60)
affectpredictedproactiveaggressionathighormeanCUtraits.
c)DoesparentingdifferbetweensubgroupsofyouthcategorizedaccordingtotheirlevelofABandCUtraits?
Barkeretal.(2011)a L AB+CU+versusAB−CU−groupexperiencedharsher Parentreportforallmeasures;noalphareported
parentingandlowerwarmth.AB+CU+versus forCUtraitmeasure
AB+CU−grouphadhighermaternalpsychopathology,
negativefeelingsandharsherparenting
Enebrinketal.(2005) CS AB+CU+versusAB+CU−groupexperiencedpoorer Verysmall,clinic-referredsampleofmales
homecircumstancesandmorefamilystress
Fontaineetal.(2011) L AB+CU+versusAB−CU−groupexperiencedmore Population-basedsampleoftwins;verysmall%
negativeparentalfeelingsandharshdiscipline.AB+CU ofsampleinsomejointtrajectorygroups
increasinghadmorefamilyandparentingriskfactors
Fricketal.(2003)a L StableCU+youthhadlowerlevelsofyouth-reported Verysmallsubgroups;largeagerange;over-selected
positiveparentingandhigherCP sampleforCP
Kimonisetal.(2004) L AB+CU+andAB−CU+versusAB−CU−and Smallsample,largeagerange;over-selectedsample
AB+CU−groupexperiencedlowerparent-reportedmonitoring/ forCP;relianceonparentreportformeasures
supervisionandyouth-reportedparentalinvolvement
Larssonetal.(2008) L AB+CU+andAB+CU−groupshadhigherlevelsofnegative Population-basedsampleoftwins;harshparental
parentalfeelingsandharshparentaldisciplineatages3,4, disciplinehadlowalphaatages3(.59),4(.56)
and7.ControllingforearlierCPmadethisnon-significant. and7(.56).
d)DoparentinginterventionshaveadirecteffectonthelevelofyouthCUtraits
HawesandDadds(2007) T Pre-andpost-treatmentchangesinCUtraits. Nocontrolgroup;verysmallsample;maleonly;
allparentreported
McDonaldetal.(2011)a I ReductionsinCUtraitsintreatmentgroupandmediated LowalphaforCUtraits(.34);smallsample;
throughimprovedparenting relianceonparent-report
Kolkoetal.(2009) T ReductionsinCUtraitsshowninbothtreatmentarms Nocontrolgroup;multipletreatmentcomponents;
across3-yearfollow-up malesonly
SomechandElizur(2012) T ReductionsinCUtraitsintreatmentgroup,maintained Attrition;measuresallparentreport;male
at1yearfollow-up dominatedsample
R.Walleretal./ClinicalPsychologyReview33(2013)593–608 599
Table2(continued)
Study Keyfindingsrelevanttoaimsofreview Mainmethodologicallimitations
e)DoCUtraitsmoderateorpredicttheeffectivenessofparentinginterventionsforyouthAB?
HawesandDadds(2005) T CUtraitsassociatedwithincreasedlikelihoodofODD Nocontrolgroup;verysmallsample;maleonly;
diagnosisaftertreatment. allparentreported
Hydeetal.(2013)a I Age3CUtraitsdidnotmoderateinterventioneffectiveness Manychildrendidnothaveexistingbehavior
problems
KolkoandPardini(2010) T TeacherreportsofCUtraitsdidnotpredicttreatment Nocontrolgroup;majoritymalesample;
outcomebutparentreportsof‘hurtfulness’did manytreatmentcomponents
White(2010)andWhiteetal.(2012) T FFTeffectiveandhighCUtraitshadbetteroutcomes, Nocontrolgroup;highattrition;lowinternal
whichrelatedtochangesinparenting consistencyformeasures
Note.Fordesign,L,longitudinalstudy;CS,cross-sectionalstudy.APSD,AntisocialProcessScreeningDevice;CD,conductdisorder;CP,conductproblems;DC,deceitful-callous;IC,
interpersonal-callousness;FFT,FunctionalFamilyTherapy;ODD,OppositionalDefiantDisorder;PSD,PsychopathyScreeningDevice.
a Studyappearsintwosectionsasanalyticstrategyaddressestworesearchquestions.
and,insomecases,aslowas.40,whichisathreattothevalidityof 3.2.1.Earlychildhood/preschool(ages3–6)
measuresandhighlightstheneedforcontinuedinvestigationofthe TwostudiesinvestigatedthelongitudinalpredictionofCUtraits
constructofCUtraitsamongyouthsamples. by dimensions of parenting in preschool samples. First, Hawes,
Inthemeasurementofparenting,13ofthe30studiesassesseddi- Dadds,Frost,andHasking(2011),inarepresentativesample(aged
mensions of parenting using the Alabama Parenting Questionnaire 3–10years; N=1008), found that parental positive reinforcement
(APQ;Shelton,Frick,&Wootton,1996)andin12ofthese13studies, predictedCUtraitsoneyearlater.Therewereinteractionsbetween
parentreportontheAPQwasusedexclusively.Twotreatmentstud- positive reinforcement and gender, poor monitoring/supervision
iesofthesamesample(Kolko&Pardini,2010;Kolkoetal.,2009)did andage,andbetweenparentalinvolvementandsex.Second,Waller
notmeasureparenting,butareretainedinthereviewbecausepar- etal.(2012)foundthatCUtraitsatages3and4werepredictedby
entingwasthefocusoftreatmentmodules.Amongstudiesthatdid observedandparent-reportedharshness(atages2and3),control-
notusetheAPQ,parentreportonaquestionnairewasstillthemost lingforage2CUtraitsanddemographicfactorsinalarge,highrisk
commonmethod.However,althoughcommonlyemployedandrela- sample(N=731).
tivelyinexpensive,parentreportissubjecttothewell-knownthreats
tovalidityassociatedwithself-reportmethods,includingsocialde- 3.2.2.Middle/latechildhood(ages7–12)
sirabilityeffectsordifficultiesinterpretingthemeaningofitemsre- SixstudiesinvestigatedthedirectpredictionofCUtraitsbyparent-
lating to parenting constructs (e.g., time-out or proactiveness) inginmiddlechildhood.First,Vitacco,Neumann,Ramos,andRoberts
(Morsbach&Prinz,2006).Theselimitationsneedtobeconsidered (2003)foundthatyouthreportsofpoorparentalmonitoringandincon-
alongside findings of studies that relied solely on parent-reported sistentdisciplinewerecross-sectionallyrelatedtonarcissismandim-
measures. Depending on the research question, direct observation pulsivity, but not to CU traits, in a community sample of Hispanic
of parenting is a potentially stronger alternative, or complement, females(N=136).Second,Fricketal.(2003)investigatedpredictors
toparent-reportedparenting,takingadvantageofassessingnatural- ofCUtraitstabilityoverfourannualassessments(N=98).Selection
ly occurring parenting behavior, using relatively unbiased ob- from a community sample via stratified sampling ensured enough
servers. In five studies, parenting was assessed via observation, youthwithhighCUtraits,althoughthismayhavemeantthatantisocial
whichstrengthenstheconclusionsthatcanbedrawnabouttheas- youth were overrepresented. CU traits at the final assessment point
sociationbetweenparentingandCUtraits,especiallyifusedtocor- werepartiallycorrelatedwithearliernegativeparenting(parentand
roborate the results obtained using parent-reported measures. youth report), controlling for earlier CU traits. Third, Pardini et al.
Observationalmethods,however,arealsonotimmunefromthreats (2007) investigated predictors of CU traits and AB in an aggressive
to validity, including observer reactivity by parents or inadequate sample(9–12yearsold;N=120)overayear.Controllingforearlier
sampling of behavior (Gardner, 2000). As such, it may be helpful AB,parent-reportedcorporalpunishmentandchild-reportedparental
toconsiderthedimensionofparentingbeingassessed,theageand warmth/involvementpredictedCU traits oneyearlater.Therewasa
typeofsample,andbyextension,whichmethodismostappropriate marginally significant interaction suggesting that children with low
toprovideavalidassessmentofparenting(i.e.,whethertheparent- anxiety who reported low parental warmth showed increases in CU
ing dimension being assessed is readily observable or better traits one year later. Fourth, Loney, Huntenburg, Counts-Allan, and
assessed using youth- or parent-report). At the same time, in 11 Schmeelk(2007)foundthatdysfunctionalparentingpracticesfullyme-
studies,nodatawasprovidedabouttheinternalconsistencyofthe diatedtheassociationbetweenmaternalpsychopathictraitsandchild
parenting measure, and in 14 studies, parenting measures had al- CU traits in a representative sample (N=83; ages 7-14years old).
phasofb.70,whichunderminestheirvalidity(seeTable2).Finally, Fifth,ina longitudinal studyof monozygotic twin pairs(N=4508),
in21studies,parentreportwasusedsolelyorincombinationwith Vidingetal.(2009)foundthatage7parent-reportednegativedisci-
otherreportstoconstructthemeasureofparentingandCUtraits.In plinepredictedparent-andteacher-reportedCUtraitsatage12,after
onlysixstudieswastheinformantforparentingandCUtraitsdiffer- controllingforearlierCUtraits.However,age7negativedisciplinedid
ent.Ifparentsareaskedtoreflectontheirbehavior,includingim- notpredictmonozygotictwindifferencesinCUtraitsatage12,control-
plementation of different discipline strategies, and then evaluate ling for age 7 CU traits. The authors speculated that associations
affective/interpersonal characteristics of their child, it is unclear reflectedgene–environmentcorrelations(i.e.,genesandenvironment
howratingsforoneaffecttheother,makingitdifficulttointerpret providedbyparentsbutfurtherevokedbyachildwithCUtraits).As
studiesthatonlyusedparent-reportedmeasures. such,negativeparentaldisciplinecouldbeanon-sharedenvironmental
riskfactorforABdevelopmentbutnotforCUtraitsduringthisagepe-
3.2.DoesparentingdirectlypredictlevelofyouthCUtraits?(Fig.1a) riod (see Vidinget al., 2005). Finally, McDonald, Dodson,Rosenfield,
and Jouriles (2011) assessed the association between parenting and
Studiesaddressingthisresearchquestiondirectlytestedtheasso- CUtraitsinthecontextofarandomizedcontrolledtrial(RCT)ofanin-
ciationbetweenacontinuousmeasureofparentingandcontinuous terventionforCPinchildrenaged4–9yearsold(N=66),recruited
outcomeofCUtraits. from domestic violence shelters. Assessments took place at baseline,
600 R.Walleretal./ClinicalPsychologyReview33(2013)593–608
and4,8,12,16,and20monthspost-baseline.Twoparentingvariables dimensions of parenting and the development of CU traits.
(psychologicalaggressionandinconsistency)wererelatedtoincreases While Viding and colleagues themselves present evidence for such
in psychopathic features over time (total PSD score and a CU-like anassociationinwhattheydescribeasa‘phenotypicanalysis’,nega-
subscale) and mediated reductions in psychopathic features in the tivedisciplineonlypredictedmonozygotictwindifferencesinAB,not
treatmentgroup. inCUtraits.Thispatternofresultshighlightstheneedforstudiesto
take into account passive correlations (i.e., correlations between
3.2.3.Adolescence(ages13–18) child behavior and the parenting environment because of a shared
TwostudiesinvestigatedassociationsbetweenparentingandCU genetically transmitted liability) and evocative correlations (i.e.,
traits in adolescence. First, Pardini and Loeber (2008) investigated child'straitsevokepoorparentingresponses)whenconsideringas-
predictorsofCUtraitstabilityoversevenassessmentsinahighrisk, sociationsbetweenparentandchildbehavior.Indeed,thisconclusion
malesample(N=506).Parentingwasassessedviafivetime1mea- issupportedbythehighheritabilityestimatesobtainedfortheCPof
suresusingsummedchildandparentreports,includingpoorsupervi- childrenwithhigh concurrent levelsofCUtraits(e.g.,Vidingetal.,
sion,physicalpunishment,andpoorparent–childcommunication.All 2005).
parentingvariableswererelatedtohigherinitiallevelsofCUtraitsin While the majority of the studies in this section attempted to
bivariateanalyses.Inafinalmultivariatemodelhowever,onlyhigher controlforearlierchildCUtraitsorAB,studiesdidnotconsistently
CP, ADHD, and poor parent–child communication predicted higher control for one or both, and few studies considered in detail the
initialCUtraits,andonlyhigherCPandpoorparent–childcommuni- effects of emerging CU traits versus AB on parenting behavior at
cationpredictedCUtraitsatthefinalassessmentpoint.Noparenting differentages.Interestingly,onestudydidinvestigatechild-driven
predictors were associated with the slope of CU traits. Second, effectsonparenting.Hawesetal. (2011)testedwhether CUtraits
Barkeretal.(2011)assessedpredictorsofCUtraitsatage13within directly predicted parenting over a one-year period, controlling
thecontextoflarge,ongoingpopulation-basedstudyofatwincohort for earlier AB. Higher levels of CU traits predicted inconsistent
(N=6673).Maternalreportsofharshparentingatage4wereassoci- parental discipline acrossages, decreasedparental involvementin
atedwithhigherlevelsofCUtraitsatage13.Inasignificantindirect older boys, and increased corporal punishment among older chil-
pathway, prenatal risk predicted boys' fearlessness at age 2, which dren.Theprediction ofparentingdimensionsbyCUtraitsshowed
was associated with higher levels of harsh parenting at age 4, and larger effect sizes than that by AB in general. As such, the results
higherlevelsofCUtraitsatage13. highlight the role of CU traits in conferring greater risk for the
development of AB by uniquely shaping parenting. Future studies
3.2.4.Summary that assess whether parenting directly predicts CU traits would
Acrossallstudies,arelativelyconsistentpictureemergesacrossa therefore benefit from simultaneously considering the effects of
rangeofdevelopmentalstages,suggestingthatdimensionsofpar- emergingCUtraitsonparenting.Hawesetal.(2011)provideause-
enting are associated with CU traits when the relationship is fulmodelforhowtotestthisquestion.However,amorepowerful
assesseddirectly.Oftheninestudiesproviding evidencethat par- analysis could involve testing cascade models in which CU traits
entingisrelatedtoCUtraits,eightwerelongitudinal,andintesting andparentingpredicteachotheracrossmultipletimepoints,there-
forpredictionbyparenting,controlledforearlierCUtraitsorAB.As byenablingaclearerpicturetobebuiltupofthekindsofinterac-
such, there is evidence that parenting is prospectively associated tions occurring in families with a youth at risk of developing, or
with CU traits, over and above earlier child-driven effects. In six already showing, high levels of CU traits. It is also worth noting
studies,negativedimensionsofparentingpredictedyouthCUtraits. thatabetterunderstandingofparent–childinteractionsisvitalfor
Specifically,harshparentingexperiencedinearlychildhood(Barker intervention,evenifriskforthemisunderpinnedbysharedgenetic
et al., 2011; Waller et al., 2012), negative discipline (Viding et al., liability.Finally,interactionsbetweenaparentandyouthwithhigh
2009),corporalpunishment(Pardinietal.,2007)andpsychological CUtraitsmaydifferwithage,althoughfuturestudiesareneededto
aggressionandinconsistentdiscipline(McDonaldetal.,2011)expe- developapreciseunderstandingofassociationsatdifferentagesof
riencedinmiddlechildhood,andpoorparent–childcommunication youthtoinformintervention.
inadolescence(Pardini&Loeber,2008)allpredictedhigherlevels
ofCUtraits.Thereisalsoevidencethatpositiveparentingpredicts
3.3. Does parenting predict youth AB at different levels of CU traits?
decreases in CU traits. Positive reinforcement in early childhood
(Fig.1b)
(Hawesetal.,2011),child-reportedpositiveparenting(Fricketal.,
2003),andchild-reportedparentalwarmth(Pardinietal.,2007)in
Studies addressing this question investigate if associations be-
middlechildhoodwereassociatedwithlowerCUtraits.Inaddition,
tween continuous measures of parenting and AB are moderated by
interaction tests suggest that parental involvement is particularly
youthCUtraits.Specifically,aninteractiontermisaddedtoregres-
important to decreases in CU traits in boys, that supervision/
sionmodels(parenting×CUtraits).Ifsignificant,theassociationbe-
monitoringandpositivereinforcementaremoreimportantingirls
tween parenting and AB is tested at high versus low levels of CU
(Hawes et al., 2011), and that parental warmth has a more pro-
traits.
nouncedeffectondecreasesinCUtraitsforchildrenwithlowanxi-
ety(Pardinietal.,2007).
Inlightoftheevidencepresentedabove,itischallengingtointer- 3.3.1.Earlychildhood/preschool(ages3–6)
pretfindingsoftheonestudythatdidnotfindevidenceforadirect Twostudiesassessedmoderationoftheassociationbetweenpar-
association between parenting and CU traits (Vitacco et al., 2003). enting and AB by CU traits in preschool samples. First, Koglin and
Interestingly, this was one of only two cross-sectional studies, and Petermann (2008) found a trend towards significance (pb.10) for
it relied on child reports of both parenting and CU traits among the interaction between low parental involvement and CU traits
Hispanicfemales.Itisdifficulttogeneralizethesefindingstocommu- predictingaggressivebehaviorinarepresentativesampleofKinder-
nityorat-riskmixedgenderandmulti-ethnicsamples.Furthermore, gartenchildren(N=177)assessedcross-sectionally.Inaddition,in-
it is difficult to be confident about the validity of measures when consistent discipline was related to aggression, but only for boys
relying solely on child reports of parenting, especially when the (n=89)withlowlevelsofCUtraits.Second,Hydeetal.(2013)test-
outcome is deficient affective and socioemotional experience. It is edwhetherage3CUtraitsmoderatedtheassociationbetweenage3
alsointerestingtoconsiderthefindingsofVidingetal.(2009)along- observed positive parentingand growthin child CP from ages 2–4,
side the evidence for a prospective association between negative andfoundthatitdidnot.
R.Walleretal./ClinicalPsychologyReview33(2013)593–608 601
3.3.2.Middle/latechildhood(ages7–12) oppositepatternofresultsforproactiveaggressionsuggeststhatchil-
Eightstudies(sixcross-sectional)investigatedmoderationofthe drenwithhighlevelsofpsychopathictraitsmaybemoreatriskof
association between parenting and AB by CU traits in middle/late learning from negative displays of parental affect they experience,
childhood. First, Wootton, Frick, Shelton, and Silverthorn (1997) although results need replication in other samples. In particular,
investigatedwhetheradichotomousCUtraitmeasuremoderatedas- resultsneedreplicatinggiventhatotherstudieshavefoundthatpos-
sociations between parenting and CP among clinic-referred males itiveparentalaffectmaybemoreimportanttoCPamongyouthwith
(N=166).CUtraitswerealwaysassociatedwithCP.Interactionsbe- highlevelsofCUtraits.Indeed,itisunclearwhyparentalnegativeaf-
tweenCUtraitsandineffectiveparenting,andCUtraitsandpositive fect might be related to proactive aggression for youth with high
parentingweresignificant.Foreach,associationsbetweenparenting levelsofCUtraits.Futurestudiesarethereforeneededtoexaminedif-
andCPwerestrongerforboyswithlowlevelsofCUtraits.However, ferential associations between dimensions of parenting (including
itisdifficulttointerprettheresults,asonlythreeofsixmodelstested parentalaffect)andreactiveversusproactiveaggressionatdifferent
included significant interaction terms. Second, Oxford et al. (2003) levelsofCUtraits.
testedwhetherCUtraitsmoderatedassociationsbetweenineffective TwostudiesinvestigatedwhetherCUtraitsmoderatedlongitudi-
parenting and CP in aggressive children (N=199), attempting to nal associations between parenting and AB. First, Pardini et al.
replicateandextendthefindingsofWoottonetal.FourCPoutcomes (2007)investigatedpredictorsofCUtraitsandABinasampleofag-
weretested(twodifferentteacherreportsofexternalizingbehavior, gressivechildrenoveroneyear(ages9–12years;N=120).Asone
parent-reportedexternalizingbehavior,andpeer-nominatedaggres- of several analyses (see earlier), Pardini et al. tested the prediction
sion)usingdichotomous(highvs.low)andcontinuousCUtraitmea- ofCPbyinteractionsbetweenvariousparentingdimensionsandCU
sures. CU traits were always related to CP. The interaction term traits. No interaction terms were significant. Second, Kroneman
betweenCUtraits(dichotomous)andineffectiveparentingwasonly et al. (2011) tested whether CU traits interacted with parenting to
significant in predicting peer-nominated aggression score (one of predict five-year CD/ODD symptom trajectories in girls (N=1233,
four models). The interaction term between CU traits (continuous) ages7–8)frommiddlechildhoodtoearlyadolescence,following-up
and ineffective parenting was a significant predictor of teacher- on the cross-sectional study of Hipwell et al. (2007). Low parental
reported externalizing and peer-nominated aggression scores (two warmth was more strongly associated with higher initial CD/ODD
offourmodels).Despitethefactthatmoremodelsthannotfeatured symptoms and faster decreasing CD/ODD symptoms for girls with
non-significantinteractionterms,theauthorsarguedthattheresults high levels of CU traits. However, by year five, the interaction be-
provided partial support for CU traits moderating associations be- tweenlowparentalwarmthandCUtraitswasnolongersignificant.
tweenparentingandCP. Nevertheless, parental warmth may be important in protecting
Third,Hipwelletal.(2007)investigatedthemoderatingeffectsofCU againstthedevelopmentofABforgirlswhoshowhighCUtraits,par-
traits on associations between CP and parenting in girls (7–8years; ticularlyduringmiddlechildhood.
N=990).Inregressionanalysis,controllingfordemographic factors
andCP,CUtraitswerenotdirectlyrelatedtoharshpunishmentorlow 3.3.3.Adolescence(ages13–18)
warmth (design was different to other studies as parenting was the Onestudymeetingtheinclusioncriteriainvestigatedthisquestion
outcome). However, the interaction between CU traits and CP was inadolescence.Muñoz,Pakalniskiene,andFrick(2011)testedbidirec-
significantlyrelatedtoharshparentingandlowparentalwarmth.The tionalrelationsbetweendimensionsofparenting(control,knowledge,
magnitude of the association between CP and parenting decreased andsolicitation)andABoverayear,usingcross-laggedpanelmodels
with increasing levels of CU traits (although remained significant in (N=98).Ofrelevancetothereview,multi-groupmoderatoranalysis
eachcase).Fourth,FalkandLee(2011)investigatedwhetherCUtraits (highvs.lowCUtraits)ofcross-laggedmodelswasconducted.Higher
moderated associations between parenting and AB in children with levelsofparentalcontrolpredictedincreasesinknowledge,andhigher
and without ADHD (6–9years; N=208). For children with low/ solicitation predicted increased control in the low CU trait group.
meanlevelsofCUtraits,lowerpositiveparentingwasassociatedwith Having less knowledge led to increased parental control for youth
parent-reported(butnotteacher-reported)CDsymptoms,controlling withlowCUtraits,whereaslessknowledgeledtodecreasedcontrol
forADHD(oneoffourmodels).Therewere,however,nosignificantin- for youthwith high CU traits. This pattern of associations highlights
teractionsbetweennegativeparentingandCUtraitsforparentorteach- theimportanceofyouthcharacteristics(e.g.,willingnesstolie)tofuture
erreportedODDorCDsymptoms(threeoffourmodels). parenting,whichmaycontributetoincreasingABovertime.
Fifth,Pasalichetal.(2011)investigatedwhetherCUtraitsmoder-
atedassociationsbetweenobservedparentalcoercionversusparental 3.3.4.Summary
warmth(codedfromspeechsamples)andCPinclinic-referredboys IthasbeenhypothesizedthatyouthwithhighlevelsofCUtraitsare
(3–10years;N=95).Formothersandfathers,coercionwasrelated lessinfluencedbynegativedimensionsofparenting(includingharsh
toCPinboyswithlowCUtraits.Incontrast,maternalwarmthwasas- discipline and coercion) because of punishment insensitivity (Blair,
sociatedwith fewerCP at high/mean levelsof CU traits,suggesting ColledgeandMitchell,2001),reducedresponsivitytonegativestimuli
that warmth could be a specific buffer against the development of (Blair, Colledge, Murray, et al., 2001), and physiological hypoarousal
ABforboyswithhighlevelsofCUtraits.Finally,inalarge,represen- (e.g., Marsh et al.,2008). Despite appearing tobe a well-established
tativesample(N=1210),Yehetal.(2011)foundthatnegativeand findingintheliterature,thereviewsuggeststhatevidenceforCUtraits
positiveparentalaffectinteractedwithyouthpsychopathyscoresto moderating associationsbetween dimensions ofparenting and AB is
predict reactive and proactive aggression. Low levels of positive mixed.Forexample,offivelongitudinalstudies,threefoundevidence
parentalaffectandhighlevelsofnegativeparentalaffectwereassoci- of moderation by CU traits and two did not. Of the seven cross-
atedwithhigherlevelsofreactiveaggression,butonlyatlowlevelsof sectionalstudies,fivereportedsomemoderation(butnotinallmodels
psychopathictraits.Incontrast,higherlevelsofnegativeparentalaf- tested)andonereportedmoderationindifferentdirections(associa-
fect were associated with more proactive aggression, but only at tionbetweenparentingandCPwasathighorlowlevelsofCUtraits
mean or high levels of psychopathic traits. These interactions were dependingonthedimensionofparentingorABassessed).Thereview
similarwhethertotalpsychopathyscoreoraffectsubscalescore(sim- therefore firstchallengesthehighlycitednotioninthefieldthatCU
ilartoCUtraitscale)wasused(K.C.Jacobson;personalcommunica- traits develop independently of parenting when this study design is
tion, March 2012). The moderating effect of psychopathic traits on adopted.Forexample,inthetwomostcommonlycitedpapersinvesti-
associations between parenting and AB is similar to the findings of gatingthequestionofmoderationbyCUtraits(Woottonetal.,313cita-
other studies for the outcome of reactive aggression. However, the tions;Oxfordetal.,103citations,November2012),neitheraddresses
602 R.Walleretal./ClinicalPsychologyReview33(2013)593–608
questionsrelatingtothedevelopmentofCUtraitsperse;bothstudies parenting may be important concomitants. For example, a parent
arecross-sectionalandhaveABastheoutcome.Furthermore,inboth whohasfosteredawarmrelationshipduringearly/middlechildhood
studies,interaction termsbetweenCUtraitsandparentingaremore couldbe more effectivein monitoring duringadolescence,and it is
frequentlynon-significant. uncleartowhatextentathird,unobservedvariable,suchasashared
Second,methodologicalcaveatsforstudiesthathaveinvestigated geneticliabilityforlowwarmth,maybetterexplaintheassociations.
thequestionofCUtraitsasacross-sectionalmoderatorareparticularly Atthesametime,thesharperincreaseinself-reporteddelinquency
relevanttoanyconclusionsthatcanbedrawn.Allfivecross-sectional of youth with high CU traits reported by Muñoz et al. indicates a
studiesthatfoundevidenceformoderationbyCUtraitsreliedonparent needforabetterunderstandingofthestrategiesparentscouldemploy
reportsofparentingandparentreportasamajorpartoftheirmeasure tomanageAB.
of CU traits (e.g., Koglin & Petermann, 2008; Oxford et al., 2003; Finally,aless consistentpictureemergesfromstudiesthat have
Woottonetal.,1997).Asalreadyoutlined,itisnotclearhowaparent's investigated positive dimensions of parenting. Two studies suggest
ratingoftheirownparentingstrategies(includingeffectivediscipline, that lowpositive parentingisrelatedto CPforyouthwithhighCU
warmth,orpositiveparent–childinteractions)mightimpactontheir traits(Kronemanetal.,2011;Pasalichetal.,2011).Itisnoteworthy
perceptionoftheirchild'saffectiveandsocioemotionalcharacteristics thatasimilarpatternofresultswasfoundinthesetwostudiesthat
(and vice versa). Furthermore, several studies are limited by having separately assessed boys and girls. The results also suggest that, in
lowalphasfortheirmeasuresofCUtraits(Falk&Lee,2011;Hipwell middlechildhoodatleast,childrenwithhighlevelsofCUtraitsmay
etal.,2007;Kronemanetal.,2011)orparenting(Pardinietal.,2007; beespeciallyresponsivetoaparent–childrelationshipcharacterized
Yeh et al., 2011). It is also difficult to generalize from studies that by positive affect and warmth, which may serve to promote their
assessed small, clinic-based samples (Pasalich et al., 2011; Wootton emotional responding and internalization of values (Pasalich et al.,
et al., 1997). Methodologically stronger studies had large samples 2011) and protect against the development of AB. In contrast, Yeh
(Hipwelletal.,2007;Hydeetal.,2013)andassessedparentingversus etal.(2011)foundthatlowerpositiveparentalaffectwasrelatedto
CU traits using different methods (Hyde et al., 2013; Pasalich et al., reactive aggression in children with low levels of CU traits. Lower
2011;Yehetal.,2011)tominimizesharedmethodvariance. levelsofpositiveparentingwerealsorelatedtoCPonlyatlowlevels
Taking into account strengths and limitations of studies, and ofCUtraitsintwoclinic-referredsamples(Falk&Lee,2011;Wootton
focusingonlyonmodelsthatincludedsignificantinteractions,there etal.,1997),althoughthemethodologicalissuesoftheselattertwo
is some evidence that negative dimensions of parenting are cross- studies make the results difficult to interpret. Finally, two studies
sectionallyrelatedtoABinyouthwithlowbutnothighlevelsofCU thatassessedwhetherCUtraitsmoderatedalongitudinalassociation
traits.Thispatternemergedforinconsistentdisciplineinpreschoolers betweenpositiveparentingandCPdidnotreportsignificantinterac-
(Koglin & Petermann, 2008), and negative affect (Yeh et al., 2011), tions(Hydeetal.,2013;Pardinietal.,2007).Furtherempiricalwork
harshness(Hipwelletal.,2007),ineffectiveparenting(Oxfordetal., isthereforeneededtounderstandassociationsbetweenpositivedi-
2003; Wootton et al., 1997) and coercion (Pasalich et al., 2011) in mensionsofparentingandCPatvaryinglevelsofCUtraits.Aswith
middle-childhood.Thispatternoffindingsfoundinstudies,compris- negativeparenting,futurestudieswouldbenefitfromincreasedpre-
ing community, high risk, aggressive, and clinic-referred samples cision in operationalization of parenting, and from making specific
of different ages, suggests that similar mechanisms may underlie predictionsabouttheimportanceofdifferentparentingpracticesat
cross-sectional associations between parenting, CU traits, and AB differentages.
acrossyouth.
However,recentstudiescomplicatethepicture.First,Yehandcol-
3.4. Does parenting differ between subgroups of youth categorized
leaguesfoundthatnegativeparentalaffectwasassociatedwithreac-
accordingtotheirlevelofABandCUtraits?(Fig.1c)
tiveaggressionatlowlevelsofCUtraits,butproactiveaggressionat
high levels of CU traits. This finding suggests that future studies
Studiesaddressingthisquestiontestavariationofthemoderator
might benefit from separating different forms of aggression when
questionoutlinedabove.Studiescreategroupsusingcutoffscoreson
measuringAB.Atthesametime,theoverlapbetweenconceptualiza- CUtraitsandABquestionnaires(althoughbydefinition,thelevelof
tionsandmeasurementofproactiveaggressionandCUtraitsneeds
ABmaybehighacrosscertainsamples).Themostcommongroups
furtherconsideration.However,aparticularstrengthofthisstudyis
created are a high AB and high CU trait group (AB+CU+), a high
the different reporters for proactive aggression (youth self-report) AB and low CU trait group (AB+CU−), and a group with low AB
versusCUtraits(parent-report),reducingtheriskofsharedmethod andCUtraits(control;AB−CU−).Studiestestforsignificantdiffer-
variance.Theresultsalsosuggestthattheremaybeimportantetiolog-
encesinparenting(assessedprospectivelyorcross-sectionallyinre-
icaldifferencesin the developmentof proactiveversusreactiveag-
lation to CU traits/AB) between groups at one time point or across
gressionthatmayberelatedtobothparentingandthepresenceof,
timepoints(i.e.,basedonstabilityofCUtraits/AB).
oroverlapwith,CUtraits.Forexample,throughsociallearningpro-
cesses,childrenwithhighlevelsofCUtraitsmaylearntovalueaggres-
3.4.1.Earlychildhood/preschool(ages3–6)
sivestrategiestoobtainrewardfollowingexposuretoaggressiverole
Nostudieshaveinvestigatedthisquestioninearlychildhood/pre-
models.
Second,whileKronemanetal.(2011)didnotfindevidencethat schoolsamples(althoughseeBarkeretal.,2011).
harsh parenting predicted AB differently in girls with low versus
highlevelsofCUtraits,theyspeculatedthatthedevelopmentalperiod 3.4.2.Middle/latechildhood(ages7–12)
inwhichharshparentinginfluencesbehaviorisearlierthanthatcap- Four studies have assessed parenting between groups of youth
turedbytheageoftheirsample.Thispointhighlightsthatfewofthe classified according to AB/CU traits in middle childhood. Enebrink,
includedstudiesmadepredictionsaboutwhichdimensionsofnega- Andershed,andLangstrom(2005)cross-sectionallytestedassociations
tiveparentingmaybespecificallyrelevantatdifferentages.Forexam- between parenting factors and CU trait groupmembership in clinic-
ple,parentalharshnesscouldbemoreimportanttoABdevelopment referredboys(N=41;ABhighacrosssample).Parentingwasassessed
inyoungchildren,whereaspoorparentalmonitoringmaybeamore via interview and evaluation of case records. The AB+CU+ group
salient dimension to investigate among adolescents with different (n=13)wasmorelikelytohaveexperiencedpoorhouseholdcircum-
levels of CU traits. Indeed, Muñoz et al. (2011) found that among stancesandhighfamilystress(whichremainedsignificantaftercon-
adolescentswithhighlevelsofCUtraits,parentswithlessknowledge trollingforODD/CDsymptoms),butdidnotdifferonothermeasures
decreased their control over time. However, other dimensions of (including little caregiver continuity and ineffective parenting).