Family School

download Family School

of 9

Transcript of Family School

  • 7/27/2019 Family School

    1/9

    Developmental Psychology1991 ,Vol.27, No.1,172-180 Copyright 1991 by the American Psychological Associaiion, Inc.0O12-1649/91/S3.O0

    Family, School, and Behavioral Antecedents to EarlyAdolescent Involvement With Antisocial PeersT. J. Dishion, G. R. Patterson, M . Stoolmiller, and M. L. Skinner

    Oregon Social Learning CenterEugene, OregonThis study focuses on the prediction of early adolescent involvement with antisocial peers fromboys'experiences in school, family, and behavior at age 10. Two hundred and six boys and theirfamilieswereassessed at school, interviewed, observedin thehome , and the n followedup at age12.Poor parental discipline and monitoring practices, peer rejection, and academic failure at age 10wereprognos tic of involvement with antisocialpeers at age12.Wealso found considerable c ontinu-ity between the boys' antisocial be havior and contact w ith antisocial peers atage 10and subsequentcontact at age12.Afterwecontrolled for such continuity, only academic failure and peer rejectionremained as significant predictors. These data indicate a need to study the ecological context ofdeviant peer networks in m iddle childhoo d.

    The p ast decade of large-scale survey studies showed an im -pressive link between involvement with antisocial peers andvarious adjustment problems endemic to adolescent popula-tions, such as substance use (Dishion, Reid,&Patterson, 1988;Elliott, Huizinga,&Ageton,1985;Huba& Bentler, 1982,1983;Kandel, 1973), delinquency (Elliott et al, 1985; Patterson &Dishion, 1985), and school dropo ut. H aving delinquent friendsin adolescence has also been found to be associated with ma in-tenance of delinquent behavior into adulthood (West &Farrington, 1977). What are poorly understood, however, areearlier childhood social experiences that contribute to chil-dren's gradually selecting deviant peer relationships by the tim ethey reach adolescence.Thepresent research considersthesepa-rate and multivariate influence of parenting practices, thechild's antisocial behavior, academic skills, and peer relationsmeasured at age 10, and children's risk for later exposure toantisocial pee rs at age 12.

    The inclusion of the family as a risk factor for exposure toantisocial peers is supported by a body of delinquency and

    This research was supported in part by research awards to G. R.Patterson from the National Institute of Mental Health, Section onStudies in Antisocial and Violent Behavior, from the N ational InstituteofAlcoholAbuse (Grant MH 37940), and from the National Instituteof Child and Human Development (Grant H D 22679), as well as by agrant awarded toT.J. Dishion by the National Institute ofDrugAbuse(Grant DA 05304).This research was possible because of the carefully collected dataprovided by the skilled O regon Youth Study staff supervised by Debo-rah Capaldi. Katie Douglas and Carol Kimball are especially appre-ciated for the careful preparation of tables and the editing of drafts ofthe manuscript.Wethank David Andrews forhiscritique of an earlierversion of this article. In addition, we acknowledge R obert Lady,Charles Stevens, and Robert H amm ond for their continued collabora-tion in providing access to schools for data collection.

    Correspondence con cerning this article should be addressed to T.J.Dishion, Oregon Social Learning Center, 207 East 5th Avenue, Suite202,Eugene, Oregon97401.

    aggression research showing a strong predictive relationshipbetween child-rearing environments in middle childhood anddelinquent behavior in adolescence (Loeber&Dishion, 1983).A social interactional model has been developed by Pattersonand colleagues (e.g., Patterson, 1982; Patterson, Reid, & Di-shion, in press) that provides an exp lanation for this predictiverelation as wellas anetwork of hypotheses regarding correlatedsocial sequelae of disrupted families. In this model, the firststage of a developmental process underlying child antisocialbehavior begins with maladaptive parent-c hild interaction pat-terns that provide payoffs to children for coercive and antiso-cial behavior. The more extreme these parent-child exchangepattern s, the more likely the child's antisocial disposition spillsover to other settings such as the school. In school settings,children's antisocial behavior interferes w ith the learning pro-cess by virtue of reduced time on-task in learning assignmentsand of being disliked by peers. Failure in school and in theconventional peer groupisconsidered (Pattersonetal.,in press)the second stage in the developm ental process. The third stageis that the failing, disliked, a nd antisocial child quite naturallyselects social settings that maximize social reinforcement. It issuspected thatthesedeviant social contacts may be increased inschools that track children with low academic skills by pu ttingthem into the same classroom environmen ts. By virtue of sys-temic influences and the child's available peer network, peer-group settings are established that may actually encourage thechild's antisocial behavior or model and shape new forms ofproblem behavior.

    Many of the links between thechild'santisocial behavior andfailure in school and with peers have been well established. Forexample, high levels of such behavior have been shown todisrupt peer relations, leading to pervasive and persistent peerrejection (Coie & Kupersmidt, 1983; Dodge, 1983). Dishion(1990) examined a model that showed both academic failureand antisocial behavior to account for variance in peer rejec-tion in 10-year-old boys. In this model, the effect of parentdiscipline practicesonpeer rejection was found to be m ediatedby boys academic performance and antisocial behavior.172

  • 7/27/2019 Family School

    2/9

    INVOLVEMENT WITH ANTISOCIAL PEERS 173Being rejected by the conventional peer group seems to berelated to a concomitant process of selective affiliation withother children who are rejected. Putallaz and Gottman (1979)noted that there was a strong tendency for the popular andunpopular elementary school children in their study to asso-ciate in separate groups, although unpopular children tendedto nominate popular childrenasfriends. L add(1983)examinedthe social interaction pa tterns of pop ular, average, and rejectedchildren in school and found t hat rejected children played withother rejected children but were unlikely to receive friendshipnominations from theirplaymates.On the other hand, Cairns,Cairns, Neckerman, Gest and Gariepy (1988) found that ag-gressive children, although often rated as disliked, were per-ceived by their peers and themselves as members of the mainclassroom social clusters. This study also revealed a high intra-class correlation coefficient between reciprocated best friendson ratings of aggression.At this stage of research, it is unclear whether similarity inaggressive behavior between friends u nderlies their mutual at-traction (e.g., Kand el, 1986) or is an outc ome of peer socializa-tion. There has been research focusing on the latter concern

    that children's deviancy status actually increases as a functionof friendship associations. Research by Cillessen (1989) indi-cated that the detrimental effect of associating with other re-jected children may emergeasearly asfirstgrade. In this study,rates of antisocial behavior for children were much higher whenthey played in small groups (i.e., triads) including only otherrejected children. H owever, when rejected children were placedin small groups including children with positive sociometricstatus,their rates of problem behavior were within norm al lev-els.Along these lines, Coie, Dodge, and Christopoulus (1989)also found that 50% of the aggression observed in their playgroups occurred in 20% of the peer dyads. Moreover, the chil-dren in the20%aggressive dyads were not particularly aggres-sive when interacting in other dyadic arrangements.Exposure to peer antisocial behavior has been shown to takea sharp increase in middle adolescence. These observed in-creases in exposure to antisocial peers in adolescence is asso-ciated with relatively rapid increases in problem behavior (El-liott & Menard, 1988). One factor that may account for thisincrease is the child's unsupervised contact w ith peers. For ex-ample, Patterson and Stouthamer-Loeber (1984) revealed thatparents' reports of the child being unsupervised are muchhigher for boys in the 10th grade tha n for those in t he 4th an d7th grades.In addition, a study by Steinberg (1986) showed that beingwith peers in places that lacked adult supervision or structuremade children more susceptible to pressure from peers to en-

    gage in problem behavior. Snyder, Dishion, and Patterson(1986) usedacross-sectional sample to exam ine the joint influ-ence of social skills deficits and parental monitoring on thechild'sassociation with antisocial pee rs in the4th,7th, andI Othgrades. These analyses revealed parental m onitoring tobeespe-cially important in the 7th and I Othgrades, accounting forunique variance in deviant peers along with a measure of theboy's social skills deficits. In this cross-sectional study, socialskills deficits were thought to represent the effect of peer rejec-tion, which was not measured.The present study extends this research in two ways. First,

    theSnyder et al. (1986)study was cross-sectional (planning sam-ple) and therefore did not address the hypothesis that parentalmonitoring and peer relations are prognostic of early adoles-cent involvement with antisocial peers. In the Oregon YouthStudy (OYS) longitudinal analyses that follow, the parentingpractices, academic skills, antisocial behavior, and peer rela-tions construc ts were assessed at age 10 and exam ined in rela-tion to involvement with antisocial peers at age 12.Two sets of hypotheses were tested. First, construct scoresrepresenting theboys'experiences in the family an d school andtheir behavioral adjustment at age10were correlated with con-tact with antisocial peers at age 12, Second, the stage modeldiscussed by Patterson et al. (in press) was examined using amultiple regressionapproach.Inthisanalysis, explanatory vari-ables focusing on the influence of p arenting, peer relations, andacademic skills were first entered into the multiple regressionpredicting age 12 associations with antisocial peers. The boys'earlier antisocial behavior and involvement with antisocialpeers was then added. T his approach to e ntering variables intoa multiple regression was used to address the multivariate hy-potheses that (a) both parent and school experiences impactchildren's involvement with antisocial peers and(b)school fail-ure and peer rejection uniquely account for individual differ-ences in subsequent involvement with deviant peers.

    We adopted a multimethod definition for each of the con-struct scores used in this research. The m ethod used for devel-oping these con struct scores, as well as the subject acquisitionand assessment procedures, are described belowMethods

    SampleThe sample com prised two coho rts of 102 and 104 boys and theirfamilies recruited in the 1983-84 and 1984-85 school years in the

    initial assessmentwaveof theOYS(Capaldi&Patterson,1988;Patter-son et al., in press). Only boys were recruited because the study inwhich our data were obtained was investigating the family precursorsto adolescent delinquency and because the incidence of such behavioris much higher in boys than ingirls(Elliott el al., 1985). T he study wasbeing conducted in a community w ith a population of 150,000 to200,000; three major school districts participated inthisresearchproj-ect. When the data in the following analyses were collected, the boyswere 9 to10years old and in fourth grade during the first year.Boys and their families were recruited through the school system.Ten elementary schools with the highest density of neighborhood de-linquency (i.e., juvenile arrest records in the school district) were se-lected from the43public elementary schools in the study community.For each cohort, the sampling order was randomly selected among 10schools. Fam ilies were considered ineligible for thestud y forthefollow-

    ing reasons: (a) The family planned to move from the area before thefirst assessment wave; (b) the family did not speak English; or(c)thefamily had m oved from the area before the research team had achanceto solicit their participation. T hirteen and seven percent of the poten-tial study participan ts in Cohorts1 and 2, respectively, were ineligiblefor one or more ofthesereasons.Of all families eligible for the study, 74.4% agreed to participate.Capaldi and Patterson (1988) compared the participant boys with thenonparticipants (rated anonymously) and found thatthere were noreli-able differences between the boys on the prim ary c linical scales of theteacher version of the Child Behavior Checklist (Edelbrock &Achen-bach, 1984). When com pared with national norm s (Patterson etal.,in

  • 7/27/2019 Family School

    3/9

    174 DISHION, PATTERSON, STOOLMriXER, AND SKINNERTable1Demographic Characteristics of Study Cohorts

    CharacteristicFamily socioeconomic s tatus8Lower (Categories1 & 2)Working (Category 3)Middle (Categories 4 & 5)Employment statusUnemployedFamily income$0-4,999$5,000-9.999$10,000-14,999$15,000-19,999$20,000-24,999$25,000-29,999$30,000-39,999$40,000+Family structureSingle p arentNumber of children1-2 children3-4 children

    5 or m ore childrenMean age of motherMean age of father

    Cohort 150 %26%23 %34%11 %22%15%14%16%8%13%3%31 %52 %37 %12%3436

    Cohort 245 %29 %26 %30%16%20%17%14%14%10%6%3%39 %58 %32 %11 %3437

    'Se e Hollingshead (1975).

    press), families in both cohorts were of lower socioeconomic status,withasomewhat higher percentage of unemployed p arents than wouldbe expected in a representative sample. Both c ohorts were predomi-nantly White (99%). The demographic characteristics of the two co-horts are provided in Table 1. Of the 206 families assessed when theboys were 9-10 years old (Wave 1), 201 were reassessed at age 11-12(Wave3);the retention rate was97%.Assessment Phases

    Theboysand their fam ilies in theOYS wereassessed in three phasesat age10.The hom e phase consisted of three observations(1 hr each),six telephone interviews conducted at approximately3-dayintervals,and questionnaires com pleted by family m embers at the end ofeachobservation session. The school phase consisted of teacher ratings,records oftheresults of standardized achievement tests administeredin the schools, and peer nom inations, all when the boys were10yearsold. The interview phase consisted of the boyandhisfamily co ming tothe research center for a structured interview, questionnaires, and avideotaped family problem-solving task. At age 12, a similar assess-mentwascompleted, butonlymea sures servingasindicators of antiso-cial peer involvement were included.

    Construct ScoresIn the present study, five constructs from the boys' assessments atage10were included in the analysis: social preference and sociometricstatus (peer relations), peer antisocial behavior, parental monitoring,observed parenting, an d child academ ic performance. At age 12, amultiagent and multimethod construct measure of peer antisocial be-havior was expanded by adding m ore items to teachers', parents', andboys'reports.The general strategy for developing m easures of constructs can bedescribed as a process of elimination. First, an a priori list of items

    from a particular instrument (e.g., interview, questionnaire, etc.) wasformulated intoa scale andtested for internal consistencyusingCron-bach's alpha (Cronbach, 1951). Items w ith an item-to-total c orrelationoflessthan .20 were eliminated from the scale. Scales from differentmeasures were then analyzed within a principle component analysis.To confirm the hypothesis that the scales representing a constructassessed one underlying dimension, a single-factor solution was used.Scales withastandardized loading of less than .30wereculledasindi-cators for a given construct. Construct indicators were generated inthis way by first using the data in Cohort1and subsequently replicat-ing them in Cohort 2. The specific contents and reliabilities of theindicators for each construct are described below. Retest stability wasassessed on a small randomly selected subsample of 20families whocompleted assessment measuresdevelopedat theOregonSocialLearn-ing Center.Sociometric status(age 10). The scoring approach de scribed byCoie, Dodge, and Coppo telli (1982) was used for the social im pact andsocial preference scores. The two nomination items used were kidswho you like as friends and kids who you don't want to be friendswith. These two items were stand ardize d by classroom. The socialpreference score consisted of each boy's score on the liked as friendsitem and his score on the don't want to be friends with item. Thesocial impact score represented the sum of nominations on both ofthese items, reflecting how often each boy was nominated on bothpositive and negative nomination items and, therefore, his saliencewithin the peer group. The following five sociometric groups wereformulated by using the scoring criteria described by Coie et ah: (a)rejected(high social im pact and negative social preference scores); (b)controversial(high social impact and ambivalent social preferencescores);(c)neglected(low social im pact an d negative social preferencescores); (d)popular (highsocial impa ct and positive social preferencescores); and (e)average(all other boys). The average classification isconsistent with research by Coie and Dodge (1988) in which childrendefined in previous research as having und eterm ined sociom etricstatus (Coie et a I., 1982) were combined with the average group. Therewere 30 rejected boys, 14 controversial boys, 11 neglected boys, 27popular boys, and 124 average boys (CohortsIand 2).Child antisocialbehavior (age 10). The child antisocial behaviorscore at age 10 and age 12 reflects the child's disposition to engage inantisocial behavior across settings as rated by parents, teachers, andthe childhimself.Acentral assumption underlying the measurementofthisconstruc t is that cross-setting c onsistency reflects more severeconduct problems than situational antisocial behavior (Loeber&Di-shion, 1984). The pare nts' report c onsisted of 12 items from the ChildBehaviorChecklist1(CBC; Achenbach&Edelbrock.1983;e.g.,arguesalot, disobe dient at hom e, lies, steals, etc.), an expanded set (29 items)ofsimilar items from the Overt-Clandestine Antisocial Questionnairedeveloped at the Oregon Social Learning Center, and 7items from theparents' telephone interview assessed over six daily interviews. Theteachers'1report consisted of18items from the teache r version of theCBC (Achenbach & E delbrock, 1986; e.g., argues a lot, disrup ts class,lies,steals,etc.).The children's report consisted of similar items takenfrom their telephone interviews(13 items)andthestructured interviewat the research center (9 items). Items that were dropped from thescales generally included low-base-rate behaviors for 10-12-year-old

    1The CBC scales recommended for the parent and teacher versionswere not used in developing constructs for the OYS. Problem swiththestandard CBC scales for this research are that (a) items representingone behavior are often included in m ore than on e scale and (b) itemslike hangs out with others who get into trouble isrepresentedontheDelinquency Scale, introducing a confound in our efforts to examinethe relationship between deviant peer involvement and delinquent be-havior.

  • 7/27/2019 Family School

    4/9

    INVOLVEMENT WITH ANTISOCIAL PEERS 175boys, suchasdrug use or vand alism. Across the two cohorts(Ar=206),the correlation between the teachers' and the parents' antisocial scoreswas reasonably high (r = .39). The correlations between the children'sreport and the parents' (r = .19) and teachers'(r =.14) reports werelower. The children's reports of their antisocial behavior obviously re-duced the overall convergent validity ofthisconstruct.However, givena modest level of retest stability of the children's reports (r = .46,n =20),this indicator was retained a s an estimate of antisocial behavioroutside ofthepurview of caretaking adults.Childacademic achievement (age 10), The Wide Range Achieve-ment Test (WRAT)Readingsubtest,standard score, and standardizedtests of achievement used in the public schools were combined withteacher CBC ratings of achievement to reflect a general score of aca-demic skill. The WRA T Reading subtest, adm inistered during the in-terview session at OSLC, was used as one indicator of academicachievement. The stanine scores of the standardized public schoolachievement b atteries, administered in the schools, were used as an-other indicator of academic achievement. The correlation between theteacher rating and the standardized achievement tests was very high(r =.70,N= 206), and the correlation oftheWRAT Reading scale tothese two indicators was also high (r = .51 and .55, respectively,N=206) .

    Observed parent discipline (age10). This scorewas an indirect mea-sure of the parent's skill in discipline issues, in which high scores re-flect the absence of negative discipline practices. Three indicatorsfrom the home observation sessions were used for this construct: twoderived directly from the home observations and one from the ob-server impressions. The observer impression score reflects impres-sions that the parents were effective, fair, consistent, and even-handedin their discipline practices. Item analyses were completed separatelyfor mothers and fathers. Six descriptors of the mother's even-handedand consistent discipline survived the item analysis and yielded alphacoefficients of .77 and .75 for Cohorts 1 and 2, respectively. Elevenobserver impressions of the fathers' even-handed and consistent disci-pline survived th e item analysis, yielding alpha coefficients of.82and.74 for Cohorts1 and2,respectively. The mother and father observerimpressionscales werecombinedtoformaparentscalefor two-pare ntfamilies. The retest stability of this parent scale was found to be .680i - 20).Two-parent discipline scores were derived from the behavioral ob-servations in the families' hom es using the Family Process Code (Di-shion et al., 1983). The parent nattering score measures the parent'snoncontingent aversiveness with the child, represented by the likeli-hood that the parent was coded as negative while interacting with thechild regardless of the child's behavior. Eight content codes were de-fined on a rational basis to describe this disciplinary tactic(negative

    verbal,coerceambiguous, refuse,negativenonverbal, noncomply, com-mandwith negativevalence, comm andambiguous with negativevalence,andphysical interact with negativevalence).The retest stability of theParent Nattering indicator was found to be .36 over a 2- to 3-monthinterval (n = 20).The second parent-to-child interactional measure was the parent

    punishment density score. This score reflects the density(i.e.,relativefrequency) of parent-to-ch ild nega tive exchanges in relation to positiveparent-child interactions. The rate per minute of all parent-to-childpositive content codes (positiveverbal, endearment,request, requestambiguous, agree,positivenonverbal,touch,hold,andcomply)wassub-tracted from the rate ofnegativecodes(negativeverbal, verbalattack,coerce, coerceambiguous, refuse, negativenonverbal, physicalaggres-sive, physicalinteract,physicalattack,andnoncomply).Parentsscoringhigh on the punishment density score were frequently and predomi-nantly negative. In contrast, the Parent Nattering score does not varyaccording to the parent's rate of interaction (i.e., 5/10 = 500/1000) oradjust for thepositive aspectsofaparent'sbehavior.The retest stability

    of the parent punishm ent density scorewasfound tobe .45(n =20,2-to3-monthretest interval).The observer impression scale was negatively correlated with theParent Nattering and Parent Punishment Density scores (-.47 and.38,respectively). The intercorrelation between parent nattering andpunishment density was moderately high(r =.49). The nattering andpunishment scoreswerereversed inscale tocombine with the observerimpressions oftheparent's positive discipline practices.

    Parent monitoring (age10). This score assessed the parent's supervi-sion of the child in relation to estab lishing clear guidelines of conductand monitoring his daily activities. The construct score was based onthe following:1. The parent's and the child's interviewer's global ratings on theextent to whichthechild was well supervisedbyparents. One item wasincluded on each interviewer impression inventory, How well is thischild monitored? The retest stability of the composite of these twoitems was.51(n =19, 2- to3-monthretest interval).2. The child'sreport of house rules inthestructured interview at theresearchcenter.The six items on this sc ale reflected the c hild's percep-tion of his parents' rules concerning telling p arents when he w ill behome, leavinganote about where he is going, checking in after school,whether there is someone hom e after school, knowing how to reachparents when they are out, and talking to parents about daily plans.The internal consistency ofthis scalewaslow(.59and.49forCohorts 1and 2, respectively). Despite the low internal consistency, the reteststability of the m easure was satisfactory over a 2- to3-monthintervalC68,n=20). Because ofthe reasonable stability ofthisvariable, it wasretained as an indicator of parental m onitoring.3. In the parent telephone interview, the primar y caretak er (usuallythe mother)wasasked about the num ber of hours the parent spent withthe son in the previous 24hr.Th eparent'sresponseto thisquestion wasaggregated over six telephone calls. Retest stab ility is not available forthis indicator.The intercorrelations among the three indicators for the monitoringconstructweremoderate, with the interviewer impressions correlatingwiththe child'sinterview report on rules (r =.36,N=206)and with theparent's telephone report on m onitoring(r =.23,N=206). The inter-

    correlation between the child's and the parent's report was low andpositive(r= .15,N= 206).Family context (age 10). Parents were asked their occupation andeducation in the interview. In addition, parents wereasked the numberof children living at home at the time ofth e interview. The number ofchildren in the family was divided by the number of parents in thehome to provide a child-to-parent ratio as a third contextual descrip-tion for the family.Peer antisocial behavior (age10). This construct atage 10consistedofthe parents', the teachers', and the children's reports on peer antiso-cial behavior. The child interview indicator included the followingitems concerning how many of the child's friends had (a)cheated onschool tests, (b) ruined or dam aged other people's things on purpose,(c)stolen something worthlessthanfivedollars,(d)hitorthreatened tohit someone w ithout a reason, (e) broken into someplace like a car or

    buildingto stealsomething,(f) stolen something worth more than $50,or(g) m ade suggestions of illegal activity. Theboysansweredtheaboveitems on a5-pointscaleranging from1 (noneof them) to5 (allof them).These items describing peer antisocial activities atWave 1and Wave 3yielded alphas of.89and .80, respectively.The parent report consisted of an item from the Child BehaviorChecklist (Achenbach & Edelbrock, 1981), which asks how often thechild hangsout withothers whoget into trouble, and ofasimilaritemfrom a parent report questionnaire developed by the researchstaff,which asks how often the child hangs out with children who steal.The retest stability for this parent report questionnairewasfound to bequite stable over a 1- to2-monthinterval (r=.72,n= 20,p

  • 7/27/2019 Family School

    5/9

    176 DISHION, PATTERSON, STOOLMILLER, AND SKINNERThe Child Behavior Checklist completed by teachers included anitem on how often th e child hangs out with children who get intotrouble, which served as the third indicator for this construct.Peer antisocial behavior (age12). A similar measurement strategywas used to assess this construct atage12,although m ore measures ofinvolvement with antisocial peers were available to buttress the mea-surement of this construct- The expanded list of items ofeach boy'sreport of antisocial peer ac tivity was a s follows. H is friends(a)get into

    manyfights,(b)don't like schoolwork,(c)get into troublealot,(d)don'tget along with adults, or(e)are kids who get into trouble and who aredaring. Ananalysisof these items revealed satisfactory internal consis-tency for both cohorts (alpha = .75,Cohort 1, and .84, Cohort 2).Items added to the teacher report of hangs around with others whoget into trouble at age 12 included (a) How often does this studen tassociate with kids who misbehave at school? (b) Does the studentassociate with kids who steal or vandalize? and (c) Does the studentassociate with kids who get into fights? These four items from theteacher ratings producedascale that showed very high intern al consis-tency across the two coho rts (alpha = .90, CohortsIand 2).Three items comprised the parent report indicator of peer antisocialbehavior, including hangs around w ith others who get into trouble(parent CBC), Does he hang out with kids who steal? and Has hebeen w ith a friend or group of friends who were setting fires? Themean intercorrelation a mong the scales of the peer antisocial behaviorconstruct at age12was .40 for Coho rt1and .42 for Cohort 2.The child report of contact correlated moderately w ith teacher re-ports(/=.34,N=206) and parent reports(r= .30,N=206). Teacherreports of peer antisocial behavior correlated moderately with parentreports( r = .31,N=206).Analysis strategy: Hypotheses were tested using two strategies.First, the bivariate correlations among individual construct scoreswere analyzed separately by cohort to determine the zero-order rela-tion between each construct and later involvement with antisocialpeers. Second, the cohorts were combined to increase the accuracy ofthe contribution of each construct in a multivariale regression equa-tion including age 10 predictors of age 12 involvement with antisocialpeers. Predictor variables were entered (i.e., forced-entry multiple re-gression) in threesets.Initially, theage 10explanatoryvariables(parentdiscipline, monitoring, peer relations, and academic skills) were en-teredtodeterm ine the relative contribution of familyand schoolexperi-ences. Next, the child's antisocial behavior at age 10 was added to themultiple regression to determine the extent that family and schooladjustment predicted later involvement with deviant peers in additionto early problem behavior. The age 10 measure of association withdeviant peers was then added to the m ultiple regression to determinewhich oftheexplanatory measures competed with the stability of thechild's involvement with antisocial peers from ages 10 to 12. We as-sumed that the age 10 predictors that accounted for significant vari-ance over and above the stability of contact with antisocial peers weremore likely to be sociaj experience s that incre ased th e child's risk foradeviant trajectory, consistent with the stage hypothesis described byPatterson et al. (in press).

    ResultsMean Level of Contact With Antisocial Peers

    Three measures of the boys' involvement with antisocialpeers were collected at both ages10and 12 and showed m ixedtrends in terms of mean levels across the two ages. Children'sreports (seven items consistent across assessments) of theirfriends' antisocial behavior decreased from a mean level of 1.64at age10to 1.52 at age12,a trend that was statistically reliable,as determined by a repeated-measures analysis of variance

    Table 2Correlations BetweenChildBehavioran dSchool AdjustmentatAg e10andPeer Antisocial Behavior atAge 12Peer antisocial behavior at age 12

    Age 10 construct scores Cohort 1 (N=102) Cohort 2(N= 104)Social preference score

    (peer relations) -. 40 *Academ ic skills -.38**Antisocial behavior .54** . 4 3. 3 4.59

    *p< . 01 . **p

  • 7/27/2019 Family School

    6/9

    INVOLVEMENT WITH ANTISOCIAL PEERS 177Table 3Peer Antisocial BehavioratAg e12byPeer SociometricStatusat Age10

    Peersociometricstatus(age10)RejectedNeglectedControversialAveragePopular

    JV3212812228

    M0.49*0.120.07-0.07-0.23*

    SD0.820.520.810.660.48

    Minimum- 0 . 65-0.62-0.84-0.94-0.94

    Maximum2.800.971.251.860.88

    *Aplanned-comparison 7 testcontrasting sociometric subgroup withaveragestatuswassignificant atp

  • 7/27/2019 Family School

    7/9

    178 DISHION, PATTERSON, STOOLMILLER, A N D SKINNERTable 5Multiple Regressionof Age10 Constructs on Peer Antisocial Behavior atAge 12

    Age 10 constructsPeer antisocialChild antisocialSocial preferenceAcademic skillParent monitoringParent discipline

    RAdjusted R2FtestP

    Ecological(family and school)constructs

    - . 2 5 * * *- .19**-.20**- .16*.5 4.2720.06.001

    Child antisocial +ecologicalconstructs

    .38***- .14*- . 1 5 *- . 1 3 *- . 06

    .62.3724.47.001

    Age 10peerantisocial +childantisocial+ ecologicalconstructs.22**.26***- . 15 *- . 15 *- . 10- . 03.6 4.4 023.08.001

    / < . 0 5 . * / ? < . 0 1 . * * * / > < . 0 01 .

    parent discipline and monitoring practices and involvementwith antisocial peers at age 12 becomes nonsignificant whencom pared with the stability of the boys' peer network and withearly problem behavior. Peer relations (i.e., social preference)and academic skills, however, remained as statistically viablepredictors. This third multivariate equation accounted for41 %of the variance in peer antisocial behavior at age 12}One issue in comparing the relative impact ofalist of inde-penden t variables on a dependen t variable in a multiple regres-sion analysisisthe level of intercorrelation among th e indepen -dent variables (see Table 6). In general, all age 10 constructscores were moderately intercorrelated. Parental disciplineskill and monitoring practices were negatively correlated withpeer antisocial behavior at an equal magnitude for ages 10a nd12;there was no difference in predictive validity over the twoyears. As documented in previous research with these boys,discipline skill and monitoring were negatively correlated withthe boys' antisocial behavior at age 10 (Patterson, 1986). Thecorrelation between age 10 peer antisocial behavior and thechild's own antisocial b ehaviorwasalso extremely high(f=.59,df= 204). This level of correlation seems to indicate that de-viant peers are important to understanding children's antiso-cial behavior prior to adolescence.

    DiscussionInterest in the deviant peer group has been largely restrictedto middle adolescence and beyond. The present research pro-vides some perspective on two issues related to contact with

    such peers. First, these results suggest that the role of the peergroup may be important developmentally earlier than adoles-cence. In fact, there is very little change in mean levels of peerantisocial activity as reported by the child, parent, or teacher,although only a limited amount of data were available in thisstudy at ages10and12.Given the potential im portance of thepeer group in middle childhood in establishing developmentaltrajectories, research on children's social development maybenefit from improving strategies for measurement of antiso-cial peer involvement in middle childhood. Two important re-lated issues are the age in which stable friendship networks

    emerge and the extent to which these networks can be differen-tiated on the basis of problem behavior.Second, the results of this study provide some be ginningevi-dence as to the ecological factors that might influence thechild's early selection of dev iant pee r contexts, implicating th eimpo rtance of both the boy's experiences in the family a nd th eschool context. Parental m onitoring and discipline practices inmiddle childhood were found to be significantly correlatedwith involvement with antisocial peers at ages10and 12. Con-sistent with th e stage model of Patterson etal.(in press) for th edevelopment of chronic antisocial behavior, only academic fail-ure and poor peer relations accounted for unique variance inthe peer antisocial behavior construct at age 12 when we con-trolled forthestability of such associations and for earlier prob-lem behavior.One possible explanation for the relation between academicfailure and involvement with deviant peers is the tendency ofschools to group children of commensurate academic skillsinto the same classroom. It is possible that, as antisocial chil-dren become increasingly deficit in academic skills, they alsofind themselves in classroom environments comprising chil-dren with similar behavioral, social, and academic profiles(Kellam, 1990). In these classroom settings, long-term friend-ships may emerge that support problem behavior and discour-age academic engagement, to the frustration of well-meaningadults.The link between poor peer relations and involvement withantisocial peers deserves more discussion. In a review of theliterature on the prediction of future adjustment problems

    from early peer relations, Parker and A sher(1987)summ arizedtheir findings by stating that there are no data supporting theidea of a causal model over an incidental model oftherole ofpeer rejection in social adjustment. The incidental modelclaims that peer rejection is simply an outcome of the child's

    2Al lpossible interactions among the entiresetof independentvari-ables were examined todetermine ifany interactions uniquelyac-countedforvarianceinpeer antisocial behavior at age12.Th ese inter-actions were near zeroan dnonsignificant.

  • 7/27/2019 Family School

    8/9

    INVOLVEMENT WITH ANTISOCIAL PEERS 179Table 6CorrelationsAmong Age10 Constructs

    Construct 11. Observed discipline2. Parent monitoring3. Peer antisocial behavior(age 10)4. Academic skills5. Social preference6. Antisocial behavior

    .20**-.33**

    .33**.34**-.40**

    - .34**

    . 2 4.25*- . 3 2

    - .25**-.25**.59**.33**-.33** . 4 6

    />

  • 7/27/2019 Family School

    9/9

    180 DISHION, PATTERSON, STOOLMILLER, AND SKINNERan danalysis issuesfor field settings. Boston, MA: Houghton Mif-flin.Cronbach, L.S.{1951). Coefficient alpha and the internal structure oftests.Psychometrika,16, 297-334.Dishion, T. J. (1990). The family ecology of boys' peer relations inmiddle childhood.ChildDevelopment, 61 ,874-892.Dishion,T.J., Gardner, K., Patterson, G. R., Reid, J.B.,Spyrou, S., &Thibodeaux, S. (1983),The FamilyProcessCode: A multidimen-sionalsystem forobservingfamily interaction. Unpublished techni-cal report, O regon Social Lea rning Center, Eugene, OR.Dishion,T.I, R eid, J.B.,& Patterson, G. R. (1988). Empirical guide-lines for a family intervention for adolescent drug use. JournalofChemical DependencyTreatment, 1(2),181-216.Dodge, K. A. (1983). Behavioral antecedents: A peer social status.ChildDevelopment, 54 ,1386-1399.Domjan, M , & Burkhard, B. (1986).77?*-principles oflearninga ndbehavior(2nded.).Monterey, CA: Brooks/Cole.Edelbrock, C , & Achenbach, T. M. (1984). The teacher version of thechild behavior profile: I.Boysaged6-11 .JournalofConsulting andClinicalPsychology, 52 ,207-217.Elliott, D. S., Huizinga, D., & Ageton, S. S. (1985). Explainingdelin-quency and druguse.Beverly Hills, CA: Sage.Elliott,D.S., & M enard,S.(1988).Delinquent behavior and delinquentpeers: Temporala nddevelopmentalpatterns. Unpublished manu-script, Institute of Behavioral Science, Department of Sociology,University of Colorado, Boulder.Gottm an, J. M. (1983). How children becom e friends.Monographsofthe SocietyforResearch in ChildDevelopment,48(3,SerialNo.201).Hollingshead, A. B. (1975).Four-factorindex ofsocialstatus. Unpub-lished m anuscript.YaleU niversity, New Haven, CT.Huba, G. J., & Bentler, P. M. (1982). A developmental theory of druguse: Derivation a nd assessment ofacausal modeling approach.Life-SpanDevelopmentandBehavior, 4, 147-203.Hub a, G. J.,&Bentler,P.M . (1983). Causalmodelsof the developmentof law abidance a nd its relationship to psychosoc ial factors and druguse. In W S. Laufer & J. M. Day (Eds.),Personality theory, moraldevelopment, an dcriminal behavior(pp.165-215). Lexington, MA:Lexington Books.

    Kandel, D. B. (1973). Adolescent marijuana use: Role of parents andpeers.Science, 181,1067-1070.Kandel, D. B. (1986). Process of peer influence on adolescence. InR.K.Silbereisen (Ed.),Development as action in context(pp. 33-52).Berlin, Federal Republic of G erm any: Springer-Verlag.

    Kellam,S.(1990). Deve lopmental epidem iological framework for fam-ily research on depression and aggression. In G. R. Patterson (Ed.),Depression an daggressionin familyinteraction(pp. 11-48). Hills-dale, NJ: Erlbaum.Lad d, G. W (1983). Social networks of popular, average, and rejectedchildren in school settings.Merrill-PalmerQuarterly, 29 ,283-307.Loeber, R., & Dishion, T. J. (1983). Early predictors of male delin-quency: A review.PsychologicalBulletin, 94, 68-98.Loeber, R., .Dishion, T. J. (1984). Boys who fight at home and inschool: Family conditions influencing cross-setting consistency.JournalofConsultingandClinicalPsychology, 52 , 759-768.Panella,D.,&Henggeler,S.W(1 986). Peer interactions of conduct-dis-ordered, anxious-withdrawn, and well-adjusted black adolescents.JournalofAbnormal ChildPsychology, 14 ,1-11.Parker, J. G, & Asher, S. R. (1987). Peer relations and later personaladjustment: Are low-accepted children at risk? PsychologicalBulle-tin,102, 357-389.Patterson,G.R . (1982).Coercive family pro cess.Eugene, OR: Castalia.Patterson,G.R. (1986). Performancemodelsforantisocialboys.Ameri-canPsychologist, 41 , 432-444.Patterson, G. R., Dishion,T.J. (1985). Contributions of families andpeers to delinquency.Criminology, 23,63-79.Patterson, G. R., Reid,J.B.,&Dishion,T.J. (in press).Antisocialboys.Eugene, OR: C astalia.Patterson, G. R.,&Stouthamer-Loeber, M. (1984). The correlation offamily management practices and delinquency.ChildDevelopment,55,1299-1307.Putallaz, M., & Gottman,XM. (1979). Social skills and group accep-tance. In S. R. Asher J. M. Gottman (Eds.),T hedevelopmento fchildren'sfriendships.Cambridge, MA: Cambridge University Press.Snyder, X, D ishion,T.J., & Patterson, G. R. (1986). Determinants andconsequences of associating with deviant peers.Journalo fEarlyAdolescence, 6,29-43.Steinberg, L. (19 86). Latchkey children an d susceptibility to peer pres-sure: An ecological analysis.Developmental Psychology 22 ,433-439.West, D. X, & Farrington, D.P.(1977).The delinquent wayof ife. Lon-don: Heinemann.

    Received October 25,1 988Revision received Xune6,1990Accepted June 25,1 99 0