Onset of antisocial behavior, affiliation with deviant friends, and childhood maladjustment

19 Pages·2007·0.29 MB·English

VU Research Portal
Onset of antisocial behavior, affiliation with deviant friends, and childhood
maladjustment: A test of the childhood- and adolescent-onset models
van Lier, P.A.C.; Wanner, B.; Vitaro, F.
published in
Development and Psychopathology
2007
DOI (link to publisher)
10.1017/S0954579407070095
document version
Publisher's PDF, also known as Version of record
Link to publication in VU Research Portal
citation for published version (APA)
van Lier, P. A. C., Wanner, B., & Vitaro, F. (2007). Onset of antisocial behavior, affiliation with deviant friends,
and childhood maladjustment: A test of the childhood- and adolescent-onset models. Development and
Psychopathology, 19(1), 167-185. https://doi.org/10.1017/S0954579407070095
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal ?
Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
E-mail address:
[email protected]
Download date: 15. Feb. 2023
DevelopmentandPsychopathology19~2007!,167–185
Copyright©2007CambridgeUniversityPress
PrintedintheUnitedStatesofAmerica
DOI:10.10170S0954579407070095
Onset of antisocial behavior, affiliation
with deviant friends, and childhood
maladjustment: A test of the childhood-
and adolescent-onset models
POLA.C.VANLIER,a BRIGITTEWANNER,b andFRANKVITAROb
aVrijeUniversiteitAmsterdam;and bUniversityofMontreal
Abstract
Predictorsandconcurrentcorrelatesofchildhood-onsetandadolescent-onsetantisocialbehaviorwerestudiedina
sampleof165boysand151girls,followedfromage6toage15.Anintegratedgeneralgrowthmixturemodelwas
usedtodeterminethenumberandshapeofdevelopmentaltrajectoriesofantisocialbehaviorexhibitedbyboysand
girls.Associationsofthesetrajectorieswithtrajectoriesofattention-deficit0hyperactivity~ADH!problemsand
deviantpeeraffiliationwereexamined.Achildhood-onset,anadolescent-onset,andalowantisocialbehavior
trajectorywereidentified.Aminorityofthesample~11%!,mostlymales,followedthechildhood-onsettrajectory.
ThistrajectorywaspredictedbypriormembershipinthehighADHtrajectoryinchildhood.Theadolescent-onset
trajectory~46%!wasassociatedwithincreasesinfriends’antisocialbehaviorbutnotwithADHproblems.Most
females~60%!followedthelowantisocialtrajectory.Asubstantialproportionoffemales,however,followedthe
childhood~5%!andadolescent-onsettrajectories~35%!.Themale–femaleratiosinthechildhoodand
adolescent-onsettrajectoriesweresimilar.Theresultslargelysupportedtheoriesthatdistinguishbetweenchildhood
andadolescentonsetsofantisocialbehavior,buttheydidnotsuggestthatboysandgirlsdifferintheageofonsetof
antisocialbehavior.
Recent theories of the development of anti- way ofASB. Other children will first exhibit
social behavior emphasize the importance of ASB at a later age ~i.e., during adolescence!.
theageofonsetofantisocialbehavior~ASB; Thepathwayofantisocialbehavioramongthe
Loeber & Stouthamer Loeber, 1998; Moffitt, adolescent-onset youth is thought to be tran-
1993;Patterson,DeBaryshe,&Ramsey,1989; sient. Their levels of ASB diminish by early
Patterson&Yoerger,1993,1997!.According adulthood. Childhood onset of antisocial be-
to these theories, some children will exhibit havior is relatively rare and persistent, and is
ASB early in life and, subsequently, are very considered pathological. In contrast, adoles-
likely to follow a life course-persistent path- centonsetisbelievedtobecommonandeven
normative, especially among males ~Moffitt,
1993!. Risk factors for childhood-onsetASB
This study was supported by grants from the Conseil
include early neuropsychological problems,
QuébécoisdelaRechercheSocialeandtheCanadaSocial
suchasverbalIQ,andadifficulttemperament
Sciences Research Council. We thank Ted Barker and
Sara Pedersen for their helpful comments and sugges- that is often reflected in early behavior prob-
tionsonthismanuscript. lems.Thesebehaviorproblemsmaybefurther
Addresscorrespondenceandreprintrequeststo:Pol exacerbated by a high-risk social environ-
A.C.vanLier,DepartmentofDevelopmentalPsychol-
ment. On the other hand, social and environ-
ogy, Vrije UniversiteitAmsterdam, Van der Boechorst-
mental factors have been hypothesized to
straat1,Amsterdam1081BT,TheNetherlands;E-mail:
[email protected]. primarily promote the adolescent-onset path
167
168 P.A.C.vanLier,B.Wanner,andF.Vitaro
of ASB development ~Loeber & Stouthamer tive in boys ~Moffitt, 1993!. In a later study,
Loeber,1998;Moffitt,1993;Patterson&Yoer- Fergusson and Horwood ~2002! included in-
ger, 1997!. Children exhibiting adolescent- formation from childhood, and again identi-
onset ASB are believed to be motivated to fied a chronic path in addition to multiple
engage inASB as a form of rebellion against adolescent-onset paths. Likewise, Schaeffer,
adults or through their association with devi- Petras, Ialongo, Poduska, and Kellam ~2003!
ant peers.The purpose of this study is to test identified a group of boys with aggressive0
theimportanceofdistinguishingamongyouth disruptive behavior that increased with age
withearlyandlateonsetofASBby~a!empir- ~from6to13years!,inadditiontothechron-
icallyidentifyingtrajectoriesofASB,and~b! ically high group. However, the “increasers”
studying whether the identified developmen- group again comprised only 7% of the total
tal trajectories are associated with indices of sample. In an attempt to expand the theory
both neuropsychological and social environ- into adulthood, White, Bates, and Buyske
mentalfactorsinexpectedways. ~2001!identifiedtrajectoriesofdelinquentbe-
Over the past decade, Moffitt and col- haviorfromage12to31yearsandidentified
leagues~Moffitt&Caspi,2001;Moffitt,Caspi, a persistent delinquency group ~7%! and an
Dickson, Silva, & Stanton, 1996; Moffitt, adolescent limited group ~33%!. In contrast,
Caspi,Harrington,&Milne,2002!haveused Chung, Hill, Hawkins, Gilchrist, and Nagin
ageneralmeasureofantisocialbehavior,which ~2002! found no evidence for an adolescent-
includesphysicalaggression,vandalism,theft, onset path, although this study started at the
anddruguse,toidentifyindividualswithchild- relativelylateageof13years.Inaddition,itis
hood and adolescent-onset forms of ASB. importanttonotethatstudiesthatfocusedon
Childhood-onsetASBhasbeenassociatedwith morespecificformsofantisocialbehavior~i.e.,
life course-persistent antisocial behavior and physical aggression, oppositional behavior,
othernegativeoutcomes~Moffittetal.,1996, propertyviolations!generallydidnotfindboth
2002!. The identification of these two paths childhood- and adolescent-onset paths ~e.g.,
of ASB onset were, however, based on pre- Bongers, Koot, Van der Ende, & Verhulst,
determined cutoff values and guided by the 2004; Brame, Nagin, & Tremblay, 2001;
theoretical work in this area. Newly avail- Broidyetal.,2003;Nagin&Tremblay,1999,
ablemethodsallowresearcherstodistinguish 2001!.
among groups of individuals on different de- Theories that stress the risk of childhood
velopmental pathways through the empirical onsetofASBversusadolescentonsetsuggest
identification of developmental trajectories that these trajectories of ASB development
~e.g.,Muthén&Shedden,1999;Nagin,1999!. areassociatedwithindividual-levelcharacter-
Studiesthathaveemployedthesenewermeth- isticsandsocialenvironmentalfactors.Asde-
odsusingmeasuresofgeneralizedaggression scribed above, the childhood-onset path is
similartothoseemployedbyMoffittandcol- thought to emerge from neuropsychological
leagues,andhavefoundsomesupportforthe deficits manifested as subtle cognitive defi-
distinctionbetweenchildhoodandadolescent cits, such as difficult temperament and0or
onset of ASB. For example, Fergusson, hyperactivity ~Moffitt, 1993!. Loeber and
Horwood,andNagin~2000!usedmixturemod- StouthamerLoeber~1998!arguethatthepres-
elingtoidentifya“chronicoffenders”andan enceorabsenceofcomorbidattention-deficit0
“adolescent-onset offenders” group in addi- hyperactivity disorder ~ADHD! distinguishes
tion to a “nonoffenders” and a “moderate of- theearly-onsetfromlate-onsetchildren.ADHD
fenders” group of youth. In accordance with isbelievedtobeprospectivelylinkedtochild-
the theory, the chronic offenders group com- hoodonsetthroughthreeprocesses,including
prisedonly6%ofthetotalsample.Fergusson poorcognitiveandacademiccapabilities,op-
et al. ~2000! also determined, however, that positionalbehavior,andearlyandaccelerated
theadolescent-onsetgroupcomprisedonly8% developmentofaggressivebehaviorsandcon-
ofthesample,afindingthatconflictswiththe ductproblems.Itshouldbenotedthoughthat
theoreticalcontentionthatthispathisnorma- some recent studies have demonstrated that
Childhoodversusadolescentonsetofantisocialbehavior 169
thetrajectoriesofantisocialbehaviorarealso social behavior are linked to affiliation with
predictedbyearlychildhoodphysicalaggres- deviant peers. No indices of early behavioral
sion~NICHDEarlyChildCareResearchNet- problemswereincorporatedinthisstudy.
work,2004;Tremblayetal.,2004!.Therisky Itisuncleartowhatextentthedistinctions
child characteristics are thought to be main- amongthetypesofASBtrajectoriesandamong
tainedandexacerbatedbysocialenvironmen- the predictors of childhood and adolescent-
tal factors, such as poor parenting styles, onsetantisocialbehaviortrajectoriesholdfor
poverty,poorrelationswithpeers,andtheaf- girls. Most of the research and theories cited
filiationwithandlearningfromotherdeviant above were validated with samples of boys
peers ~Dishion, Patterson, & Griesler, 1994; only. Silverthorn and Frick ~1999! suggested
Elliott, Huizinga, & Ageton, 1985; Moffitt, that only one type of onset accurately de-
1993; Patterson, Dishion, & Yoerger, 2000; scribed the development of antisocial behav-
Snyder, Prichard, Schrepferman, Patrick, & ioringirls.Girlsweredescribedasexclusively
Stoolmiller, 2004; van Lier, Vitaro, Wanner, followingadelayed-onsettrajectoryinwhich
Vuijk, & Crijnen, 2005!. Support for this as- the development of antisocial behavior ap-
sertion has been identified in the empirical pears in adolescence. Partial support for
literature. The “life-course persisters” identi- this was found by Moffitt and Caspi ~2001!
fiedbyMoffittandCaspi~2001!werecharac- who reported a 10:1 ~boys to girls! ratio in
terizednotonlybypersistentlyhighlevelsof childhood-onset path versus a 1.5:1 ~boys to
aggressive behavior, but also by early neuro- girls! ratio for the late-onset, adolescent-
cognitive and behavior problems, especially limited path. Similar findings were reported
hyperactivityandfighting,andbyexposureto byFergussonandHorwood~2002!,whodem-
inadequateparenting. onstratedthatwhenfemalesexhibitantisocial
Adolescent-onsetASB,incontrasttochild- behavior,theyaremuchmorelikelythanmales
onset ASB, is hypothesized to stem from tofollowanadolescent-onsetpath.However,
environmental, rather than organic, factors a small number of females ~2%! did follow
~Moffitt, 1993; Patterson & Yoerger, 1997!. thechildhood-onsetpath.
Specifically,affiliationwithdeviantpeersdur- Althoughthestudiesdescribedabovehave
ing adolescence is thought to contribute to provided a solid foundation for work in this
adolescentASB.Ithasbeenshownthatyoung area, the extent to which the age of onset of
childrenwillnotaffiliatewithearly-onsetanti- antisocial behavior reflects an important di-
socialchildrenand,instead,ignoreandreject mensiononwhichtrajectoriesofASBcanbe
them~Coie,Dodge,&Kupersmidt,1990!.Dur- distinguishedhasnotbeendirectlytestedbya
ingadolescence,however,asantisocialbehav- model that incorporates all the critical ele-
iorbecomesmorenormative,moreadolescents mentsofthedevelopmentaltaxonomy.Todi-
model the behavior of their antisocial peers, rectlytestthetwotypesofonsetpatterns,three
resulting in an increase inASB among youth different indices are needed. First, repeated
in this age group. Support for the distinc- assessmentofantisocialbehaviorfromchild-
tion between predictors of the childhood and hoodintoadolescenceisrequiredfortheem-
adolescent-onset paths was reported by Mof- pirical identification of the developmental
fittandCaspi~2001!.Theyfoundthatthe“ad- trajectories.AccordingtoDSM-IV~American
olescentlimited”youthtendedtoaffiliatewith PsychiatricAssociation,1994!onecharacter-
antisocial peers. Recently, Vitaro, Brendgen, istic of the early-onset type of conduct dis-
andWanner~2005!studiedpatternsofaffilia- order is antisocial behavior beginning prior
tionwithdeviantfriendsinlatechildhoodand to or at the age of 10. Ratings of antisocial
early adolescence. Early and late affiliation behavior starting at around this age would
trajectories were identified. In addition, most therefore be required. Second, indices of
children were found to parallel the level of learned behavioral or personal dispositions
deviancy among the peers with whom they during childhood are needed to test whether
affiliated.Thislatterfindingsuggeststhatboth they discriminate between childhood-onset
the childhood and adolescent onset of anti- and adolescent-onset cases. Given the hy-
170 P.A.C.vanLier,B.Wanner,andF.Vitaro
pothesized influence of attention-deficit0 childrenarelikelytoaffiliatewithfriendswho
hyperactivity~ADH!problemsonachildhood- are high on antisocial behavior, we expected
onset trajectory, information on the de- that these children already exhibit elevated
velopmental course of such problems in levels of antisocial behavior prior to their af-
childhood would be needed. Third, informa- filiationwithdeviantfriends.Wealsohypoth-
tion on the concomitant process of friends’ esizedthatthechildhood-onsetpathwouldbe
antisocialbehaviorwouldbenecessarytotest predictedbyhighlevelsofADHproblemsin
whether the course of antisocial behavior childhood,whereastheadolescent-onsetpath
among adolescent-onset children is directly would not be predicted by childhood ADH
associatedwithaffiliationwithdeviantfriends. problems.
Although deviant peers may exacerbate the
deviantbehaviorofchildhood-onsetchildren,
Method
the high level of antisocial behavior among
childhood-onset children should not be ex-
Participants
plained solely by deviant friends’influences.
This study attempts to resolve the contra- Participantsinthepresentstudywere316Cau-
dictory findings in past research by attempt- casianchildrensurveyedyearlyfromage6to
ingtoestablishtheexistenceofchildhoodand age 15 ~52% boys!. All children lived in a
adolescent-onset trajectories of ASB and es- small city in northwestern Quebec, Canada.
tablishingdifferentialassociationsofthesetra- They attended five elementary schools up to
jectorieswithearlyADHproblemsandfriends’ age 12, after which they transferred to one
deviancy.Specifically,weaddressthefollow- large high school ~except for those who were
ing questions: how many developmental tra- retained!. The schools were relatively homo-
jectories of self-reported antisocial behavior geneousinsize,numberofclasses,character-
can be identified in a sample of French- istics of the children in the classes, and
Canadian youths surveyed across middle characteristics of the neighborhood. In addi-
childhoodandadolescence?Howmanydevel- tion,allchildrenattendedregularclassrooms,
opmental trajectories of ADH behavior are although 12% of the sample was in class-
evident in the same sample surveyed from rooms with younger classmates because of
kindergarten to middle childhood? What is grade retention. In each year of data collec-
the developmental course of friends’ anti- tion, at least 90% of children in the targeted
social behavior of the children in each of the classrooms participated in the study. Those
developmentaltrajectoriesofantisocialbehav- whodidnotparticipatedidnotreceiveparen-
ior?Arethedevelopmentaltrajectoriesofanti- talpermissionorwereabsentfromschoolon
socialbehaviorassociatedwiththepreceding thedayofdatacollection.
developmental trajectories ofADH problems Datausedinthisstudycomesfromalarger
and with the simultaneous development of studyof445children.Twocriteriawereused
friends’ antisocial behavior? We expected to for inclusion in the present study. First, anti-
identifythreetrajectoriesofASB,includinga socialbehaviorscoresinchildhoodhadtobe
childhood-onset, an adolescent-onset, and a present to ensure that the early-onset trajec-
lowantisocialbehaviortrajectory.Wealsohy- torycouldbedisentangledfromanadolescent-
pothesizedthatparticipationinthechildhood- onset trajectory. Second, teacher-rated ADH
onset trajectory would be limited to a small problem scores from kindergarten to age 10
numberofsubjectsandwouldbeparticularly had to be available. In total, 129 participants
rareamongfemales.Incontrast,anadolescent- wereexcludedbyvirtueofthesetwocriteria,
onset path was expected to characterize the leadingtothefinalsampleof316participants
ASBdevelopmentofasubstantialproportion ~165 boys, 151 girls!. No differences were
of males. We further hypothesized that the present in gender distribution, grade reten-
levelofantisocialbehaviorofchildrenonthe tion,orantisocialbehaviorscoresbetweenin-
adolescent-onset path mirrors that of friends’ cluded and excluded children. Participants’
antisocialbehavior.Althoughchildhood-onset socioeconomic status was lower compared to
Childhoodversusadolescentonsetofantisocialbehavior 171
participantsofarepresentativesampleofsame- tingfire!,theft~10items,e.g.,takenandkept
ageQuebecchildrenaccordingtotheBlishen, something in a store without paying, steals
Carroll,andMoore~1987!occupationalpres- something from parents, steals an amount of
tige scale ~means for the years they partici- CAD 10 or more from school, buying, using
pated in the studies: M(cid:1)41.92, SD(cid:1)12.77 orsellingsomethingthathasbeenstolen,steal-
vs. M(cid:1)46.02, SD(cid:1)13.18!. Seventeen per- ing by breaking and entering!, or alcohol–
cent of the parents completed high school, drug use ~3 items: used marijuana, binge
43%hadposthighschooleducation,and40% drinking,drinksalcohol!.The25itemsofthe
hadlessthanhighschooleducation,and75% SRDQwereembeddedinvariousotheritems
if the participant’s families were intact at the pertainingtoschool,hobbies,socialrelation-
startofthestudy,whenchildrenwere6years ships, and parent relationships. LeBlanc and
old. Parental permission was obtained each McDuff ~1991! reported satisfactory internal
yearforallparticipants. consistency,test–retestreliability,andconver-
gent,discriminant,andpredictivevalidityfor
early adolescents of both genders. With re-
Procedure
spect to convergent validity, these authors
Each spring, participants spent 2 hr of class- found higher correlations among the sub-
roomtime,dividedbya20-minbreak,answer- scalesoftheSRDQ~i.e.,violence,vandalism,
ing questionnaires. After the children were theft, substance use; r;.35! than with other
informedaboutthepurposeofthestudy,they scales tapping into family relationships and
were told that all of their answers would be school~r;.20!.Asfordiscriminantandpre-
confidential and that they did not have to an- dictivevalidity,theSRDQscalessignificantly
sweranyofthequestionsiftheydidnotwant distinguished groups of adjudicated and non-
to.Thechildrenwereencouragedtokeeptheir adjudicatedjuveniles.Italsosignificantlydis-
answersconfidentialandnottotalkwithclass- criminated adolescents who subsequently
mates about their answers. Trained research received a criminal record by age 30 years
assistantsadministeredandcollectedtheques- from those who remained without a crime
tionnaires. Teachers were asked to leave the record ~see also Loeber & LeBlanc, 1990!.
classroom during the assessment time to em- Other researchers have also documented the
phasize that participants’ answers would not validity of self-reported measures of delin-
be revealed to their teachers. Teachers also quency ~Hindelang, Hirschi, & Weiss, 1981;
completedquestionnairesduringthisperiod. Klein,1989!.Forthetotalantisocialbehavior
scale, which was created by summing the in-
dividual item scores, Cronbach a values var-
Measures
ied from .83 to .92 from age 10 through age
Participants’ antisocial behaviors were as- 15 years. Fifty-seven percent were missing
sessed by means of the Self-Reported Delin- oneormoreassessments.
quency Questionnaire ~SRDQ; LeBlanc & Reciprocalfriends’antisocialbehaviorwas
Fréchette, 1989!.At each time of assessment assessed from age 10 through age 13 years.
from age 10 through age 15 ~i.e., six times!, Participantswereaskedtonominateuptofour
theparticipantsreportedhowfrequently~0(cid:1) best friends in the classroom. Classroom size
never, 1 (cid:1) rarely, 2 (cid:1) sometimes, or 3 (cid:1) variedfrom20to27participants,withanav-
often!theyhadbeeninvolvedoverthepast12 erageof24.5participants.Friendshipnomina-
months in physical violence ~6 items, e.g., tion was restricted to the classroom at all
threatentobeatsomeone,usedaweapon,beat assessment times because classroom compo-
someone without a reason, engaged in a fist- sitionremainedstablethroughouttheyearand
fight, threw rocks or other objects at some- participants spent all of their time within the
one!, vandalism ~6 items, e.g., broken or sameclassroom.Afriendshipwasconsidered
destroyed something that did not belong to reciprocalonlywhenthenominatedfriend,in
you, demolishing school equipment, having turn,alsonominatedtheparticipantasoneof
broken down parts of a car deliberately, set- hisorherfourbestfriends~Bukowski&Hoza,
172 P.A.C.vanLier,B.Wanner,andF.Vitaro
1989!. Because the friends also filled out all mental trajectories of ADH problems were
questionnaires,includingtheSelf-ReportedDe- analyzed using trajectory analyses in a gen-
linquencyQuestionnaire,itwaspossibletoas- eral mixture model ~GMM; Muthén & Shed-
sess the reciprocal friends’ engagement in den, 1999; Nagin, 1999!. The objective of
antisocialbehavior.Theuseoffriends’ownre- GMMistofindthesmallestnumberofclasses
ports of their antisocial behaviors avoids the ofindividualswithsimilardevelopmentaltra-
problem of shared source variance with re- jectories of antisocial behavior. GMM esti-
specttotheparticipant’sownantisocialbehav- matesmeangrowthcurves,thatis,initialstatus
ior,whichisself-reported~Thornberry&Krohn, ~intercept!andchange~slopes!,foreachclass
1997!.Whenaparticipanthadmorethanone andcapturesindividualvariationaroundthese
mutualfriendinagivenyear,thefriends’total growthcurvesbytheestimationoffactorvari-
antisocial behavior scores were averaged to ances for each class. Several indications may
gainamorecompletepictureofthecharacter- beusedindecidingontheoptimalnumberof
isticsofaparticipant’sselectedfriendshipnet- classes~Muthén&Muthén,2000!.Thefirstis
work.Tostudywhetherthisprocedureimpacted the Baysian information criterion ~BIC; Kass
thefriends’antisocialbehaviorscores,theav- &Raftery,1993;Schwartz,1978!.LowerBIC
eragedscoreswerecomparedtothefirstrecip- values indicate improvement of the model
rocalbestfriend’sscores.Nodifferencesinmean when compared to the model with one fewer
antisocial behavior scores were found. Eigh- class. The second is the usefulness of the
teenchildren~6%!didnothaveamutualfriend classes, which can be determined by com-
onanyofthefourassessments.Boyshad,on paring the developmental trajectories, the
average, between 1.22 ~SD(cid:1)1.10! and 1.61 number of children in each class, and the as-
~SD(cid:1)1.19!reciprocatedfriendshipsandgirls sociationsamongtheclassesandthepredictor
had1.27~SD(cid:1)1.20!to1.77~SD(cid:1)1.30!re- variables. The third is the stability of the so-
ciprocatedfriendshipsacrossassessments.Of lutions. To test this, models were repeatedly
thosefriendships,only3.03–6.67%fortheboys estimatedwithrandomperturbationsofstart-
and3.31–7.95%forthegirlswerewithachild ing values, generated by the program. Stable
oftheothergender.Onlyoneboyandonegirl models achieve similar solutions despite dif-
hadtwofriendsoftheoppositegenderatone ferentstartingvalues.
timeofmeasurement. In the second stage of the analyses, the
ADH problems from age 6 to age 10 were developmental trajectories ofADH problems
ratedbytheteachersusingtheSocialBehav- were included in the model. The trajectory
ior Questionnaire ~SBQ; Loeber, Tremblay, classes of ADH problems were regressed on
Gagnon, & Charlebois, 1989!. The SBQ is a thetrajectoryclassesofantisocialbehaviorto
32-itembehavior-ratingquestionnaireinwhich test whether ADH problem trajectories were
teachers indicated whether items did not ap- predictive of later developmental trajectories
ply ~0!, applied sometimes ~1!, or applied of- ofantisocialbehavior.Inthethirdstageofthe
ten ~2! to a target child.The inattention scale analyses, the developmental course of recip-
comprisedtwoitems.Thehyperactivityprob- rocal friends’ antisocial behavior was added
lems scale was comprised of four items. The to the model. The development of friends’
two scales were summed to create an ADH antisocial behavior was estimated using the
problems scale ~ADH problems!. Of the 316 averagescoresofthereciprocalfriendsofthe
children, 30% were missing one or more as- childrenineachoftheantisocialbehaviorde-
sessments. Cronbach’s a values ranged from velopmentaltrajectoriesidentifiedinthefirst
.81to.89 fromage6throughage10years. stageoftheanalyses.Figure1givesanover-
view of the observed and latent variables of
this third and final model. The estimated pa-
Statisticalapproach
rametersofthismodelare~a!latentclassmem-
Theanalyseswereperformedinthreestages. bershipprobabilities~i.e.,theprobabilitythat
First, the developmental trajectories of self- each individual belongs to each of the anti-
reportedantisocialbehaviorandthedevelop- socialbehaviorandADHproblemstrajectory
Childhoodversusadolescentonsetofantisocialbehavior 173
Figure1. Observed and latent variables to test the theories childhood versus adolescent onset of
antisocialbehavior.
classes!; ~b! the means and variances of the friends’ASB from age 10 to age 13. In line
growth factors ~intercept and slopes! of the with the expected age-crime curve, ASB in-
antisocialbehaviortrajectories,themeansand creased with age for both own and friends’
variancesofthegrowthfactorsoftheirrecip- ASB.
rocal friends’ antisocial behavior, and the
means and variances of the growth factors of
Determiningtrajectoriesofantisocial
the ADH problems trajectories; and ~c! the
behaviorandADHproblems
multinomial regression coefficient ~i.e., the
odds ratio!, which tested whether the anti- To find the optimal number of trajectory
socialbehaviortrajectorieswerepredictedby classes, the variances and covariances of the
theADHproblemstrajectories.Malesandfe- continuous latent variables and the variance
males were not analyzed separately, because of the indicators were set to zero ~Muthén &
of the sample size, but the parameter esti- Muthén, 2000!. The BIC indicated that up to
mateswereadjustedfor~male!genderinthose fourdevelopmentaltrajectoriesofself-reported
trajectoriesthatcontainedasufficientnumber ASBcouldbefitted~BICthreeclasses(cid:1)8863,
of males and females.All analyses were per- fourclasses(cid:1)8788!.Movingtofive~BIC(cid:1)
formed with Mplus 3.0 ~Muthén & Muthén, 8721! trajectory classes or more resulted in
1998–2004!. All presented results are based trajectory classes comprising very few sub-
onthefinalmodel. jects~,3%ofthesample!,andtounstableor
to nonconverging solutions due to empty tra-
jectory classes. To study whether the three
Results
andfourclasssolutioncouldbeimprovedthe
indicator variances were freed for the overall
Descriptivedata
model ~three class BIC (cid:1) 8791, four class
Figure 2 illustrates the development of self- BIC(cid:1)8685!.Then,thevariancesofthecon-
reported ASB from age 10 to age 15 and tinuouslatentvariableswerefreed~threeclass
174 P.A.C.vanLier,B.Wanner,andF.Vitaro
Figure2.Meanlevelsofself-reportedantisocialbehaviorandfriends’self-reportedantisocialbehavior
fromchildhoodintoadolescence.
BIC (cid:1) 8647, four class BIC (cid:1) 8600!. How- developmentaltrajectoriesofASBandfriends’
ever, this four class solution had one class ASB were estimated simultaneously. To test
comprising less than 3%, and a second com- whether missing assessments impacted the
prising less than 5% of the sample.When al- model ofADH problems, the model was run
lowing for class-specific indicator variances, again for children with complete data only.
model fit further improved in the three class Parameterestimatesweresimilartothoseiden-
solution ~BIC(cid:1)7922!, but resulted in unsta- tifiedbythemodelthatincludedallavailable
ble solutions for the four class model. The data. The model including all available data
developmentofself-reportedantisocialbehav- wasthereforeusedforthepresentationofthe
ior was therefore best described by three de- results.
velopmental trajectories. To test whether
missingassessmentsvaluesimpactedtheesti-
mationoftheASBtrajectories,themodelwas Developmentaltrajectoriesofantisocial
runagainforchildrenwithcompletedataonly. behavior:Shape,prevalence,andgender
The parameter estimates were similar to the
modelthatusedallavailabledata. Figure3presentsthedevelopmentaltrajecto-
ThesameapproachwasusedforADHprob- ries of self-reported antisocial behavior from
lems.Uptofivetrajectoryclasseswerefitted. age10toage15.Elevenpercentofallpartici-
The two class ~BIC (cid:1) 5344! and three class pants followed a chronically high trajectory
~BIC (cid:1) 5296! solution were stable and had thatwassignificantlyhigherthantheremain-
meaningful trajectory classes. The four and ing trajectories from childhood into adoles-
fiveclasssolutionshadimprovedBICvalues cence. This trajectory is best described as
~four class BIC (cid:1) 5258, five class BIC (cid:1) childhood onset of ASB. The average scores
5225! compared to the three class solution. ontotalASBandonthefourscales~physical
However, the extra classes were always sub- violence, vandalism, theft, and alcohol–drug
populations comprised of children with low use!thatcomprisethetotalscalearegivenin
ADHproblems~i.e.,noadditionalmeaningful Table 1. The table shows that the childhood-
devianttrajectorieswereidentified!.Wethere- onset group had consistently high values of
forechosethethreeclasssolutionastheopti- each type ofASB.The gender distribution of
malmodel. thedevelopmentaltrajectoriesispresentedin
Inthefinalmodel,thethreedevelopmental Table2.Although75%ofthechildrenonthe
trajectories of ADH problems and the three childhood-onset trajectory were male, 6% of
Childhoodversusadolescentonsetofantisocialbehavior 175
Figure3. Developmentaltrajectoriesof~—!self-reportedantisocialbehaviorand~{{{!friends’self-
reportedantisocialbehaviorfromchildhoodintoadolescence.
all females were members of this trajectory rameter estimates were adjusted for gender,
group. and the shape of the adolescent-onset trajec-
To validate the level of deviancy of this torywasfoundtobesimilarforboysandgirls
childhood-onset trajectory group, the ASB ~seeTable2!.Again,theASBscoresofthese
scores at age 13 were compared to the ASB children were compared with theASB scores
scoresofearlyadolescentsfromalargerepre- ofboyswhohadbeenarrested.Attestshowed
sentative sample ~N (cid:1) 2000! who had prob- thattheadolescent-onsetyouthshadlowerASB
lems with the police and were convicted of scores at age 13 than boys who had been
illegal acts ~N (cid:1)185!. These data were col- arrested.
lectedfromtherepresentativesamplethrough- Theremaining43%ofthesample,ofwhom
out the province of Quebec during the same 67% were females, followed a trajectory of
periodthatdatawerecollectedfromthestud- lowASB in childhood that slightly increased
iedsample.AttestshowedthattheASBscores throughadolescence.Overall,thelevelofanti-
of the two groups did not differ significantly social behavior in this trajectory class was
~childhood-onset trajectory group, M(cid:1)12.5, lowerthanthelevelintheothertwotrajectory
SD(cid:1)10.5; representative sample, M(cid:1)10.6, classes.Thus,thisclassisreferredtoasalow
SD(cid:1)7.2; only boys were compared because antisocialbehaviortrajectoryclass.Sixtyper-
ofthesmallnumberofgirlsinthechildhood- centofallfemalesinthesamplefollowedthis
onsettrajectorygroup!. trajectory.Asignificantrelationbetweengen-
Nearlyhalf~46%!oftheparticipants~65% der and the intercept of this trajectory indi-
males! followed a trajectory characterized by cated that males who traveled along this
low initial levels of ASB in childhood that trajectory had higher levels of antisocial be-
increasedthroughoutadolescence.Thistrajec- haviorthanfemales~seeTable2!.
tory is best described as an adolescent onset Themale–femaleratiodifferedamongthe
ofantisocialbehaviorclass.Forthisclass,the three trajectory classes, x2 ~2, N (cid:1) 316! (cid:1)
increase in antisocial behavior is mainly due 37.5, p , .01. However, the male–female ra-
toincreasesintheftandalcohol–druguse~see tioofthechildhood-onsettrajectoryclasswas
Table 1!. Fifty-seven percent of all males in not statistically different from the male–
the sample followed this developmental tra- female ratio of the adolescent-onset trajec-
jectory.Thirty-fourpercentofthefemaleswere tory,x2~1,N(cid:1)316!(cid:1)1.3,p..05,indicating
in this trajectory class ~see Table 2!. The pa- thattheoverallinmaletofemaleratiodiffer-