Transcript

Parenting Predictors of Early Conduct Problemsin Urban, High-Risk Boys

GAIL A. WASSERMAN, PH.D., LAURIE S. MILLER, PH.D., ELIZABETH PINNER, M.A.,

AND BEATRIZ JARAMILLO, D.P.H.

ABSTRACT

Objective: As part of a larger, prospective study, the authors examined concurrent and prospective relations among

parenting and child antisocial behavior in inner-city boys at high risk for delinquent behavior. Method: One hundred

twenty-six younger brothers (aged 6 to 10 years) of convicted delinquents in New York City and their parents were

assessed; 15 months later 112 boys were reassessed. Demographics, parenting, and child diagnosis were examined

as they relate to child externalizing behavior problems. Hierarchical multiple regression analyses predicted changes in

Externalizing scores from year I parenting. Results: At years I and II, 22% and 27% of boys, respectively, scored above

the clinical cutoff for Externalizing. Controlling for earlier Externalizing, each of three domains of parenting still made

significant independent contributions to later Externalizing scores, explaining 17% of the variance. Altogether this model

explained 51% of the variance in year II Externalizing scores. Conclusions: Data support a cumulative risk model,

whereby each of several adverse parenting factors further compounds the likelihood of child conduct problems. J. Am.

Acad. Child Ado/esc. Psychiatry, 1996,35(9):1227-1236. Key Words: conduct problems, parenting, antisocial behavior,

family factors.

Recent formulations of the development of childhoodconduct problems have described two distinct develop­mental patterns (e.g., Moffitt, 1993): early-onset life­persistent and adolescent-limited. Early starters displaybehavior problems prior to age 10 years, are frequentlyphysically aggressive, and tend to have disturbed peerrelations. Compared with those with the adolescent­limited pattern, early starters are more likely to havepersistent conduct problems, to develop antisocial per­sonality disorder in adulthood (American PsychiatricAssociation, 1994), and to be impaired in multiplefunctional domains. Given the high risk for negativeoutcome in such children, understanding the precursorsof early-onset conduct problems is of great importance.

Accepted March 28, 1996.From the Department ofChild Psychiatry, Columbia University/New York

State Psychiatric Institute, New York.

This research was supported by a grant from the Lowenstein Foundation to

the Lowenstein Center for the Study and Prevention ofDisruptive Disorders(Gail A. Wasserman, Principal Investigator).

Reprint requeststo Dr. Wasserman, Columbia University, NYSPI Unit 78,722 West 168th Street, New York, NY 10032.

0890-8567/96/3509-1227$03.00/0©1996 by the American Academyof Child and Adolescent Psychiatry.

Transactional models of early-onset conduct prob­lems suggest that the juxtaposition of a difficult child(e.g., impulsive, irritable) with an adversefamily context(e.g., incompetent parenting) initiates risk for a persis­tent pattern of oppositional and aggressive behavior(e.g., Cicchetti and Richters, 1993; Conduct ProblemsPrevention Research Group, 1992; Moffitt, 1993).While skillful parenting may overcome early childhooddifficulties, family environments characterized by nega­tive interactions will likely maintain or worsen childbehavior problems.

The relation between family environment and child­hood conduct problems has been demonstrated pre­viously (e.g., Henggeler, 1989; Loeber and Dishion,1983; Loeber and Stoutharner-Loeber, 1986; Snyderand Patterson, 1987). For example, families of childrenwith conduct problems are characterized by inconsistentdisciplinary practices and unintentional parent rein­forcement (McCord, 1979; Patterson and Stouthamer­Loeber, 1984). Such family environmental factors haveparticular influence on juvenile antisocial behavior (Ly­ons et aI., 1995; McGuffin and Gottesman, 1985).

Three specific parenting practices emerge from theexisting literature as important for the onset of early

]. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 35:9, SEPTEMBER 1996 1227

WASSERMAN ET AL.

conduct problems (see Loeber and Stourhamer-Loeber,1986, for a review). One such important aspect is"parent-child conflict," which encompasses harsh pun­ishment, fighting and arguing between parent andchild, and overt dislike of the child by the parent.Relations between parents and children with conductproblems are often characterized by harsh physicalpunishment (Andrew, 1981; Eron et al., 1991).

A second important domain is "monitoring," whichinvolves knowing where, how, and with whom thechild spends time. Children with conduct problemstend to be more poorly monitored (Farrington, 1983;Patterson et al., 1992).

"Positive involvement" is a third domain that in­cludes support for socially competent behavior (Pat­terson et al., 1992), emotional support, and expressionsof interest in the child. While involvement providedin the home is important for academic achievement(Bradley et al., 1989; Luster and Dubow, 1992; Richterand Grieve, 1991), recent studies have demonstratedthat a cognitively stimulating and emotionally support­ive home environment also predicts decreased antisocialbehavior (e.g., Dubow and Ippolito, 1994).

While very few studies have examined the relativeimportance of different aspects of family functioningover time, one recent study (Reiss et al., 1995) under­scores the contribution of each of these components toantisocial behavior. Unfortunately, that study examinedneither the relative contribution of these componentsto antisocial behavior nor the prospective associationsbetween parenting and child outcome.

In another series of prospective studies, poor moni­toring and inept discipline made significant independ­ent contributions, of equal magnitude, to children'santisocial behavior (Forgatch, 1991; Patterson et al.,1992). Similarly, Sampson and Laub (1993) demon­strated that erratic and harsh discipline, low supervision,and weak parent-child attachment each predicted self­parent-teacher-reported delinquency in older children.

Following a sample of children at elevated risk forearly-onset conduct problems facilitates the study ofemerging conduct problems. Investigations that mea­sure parenting and child behavior concurrently are lessinformative since components of parenting are as likelyto be reactions to child conduct problems as they areto be contributors to child misbehavior. A prospectiveinvestigation of a high-risk cohort, beginning beforethe onset of serious behavioral disturb-ance, would

better elucidate the specific contributors to child behav­ior problems.

To expand our understanding of the multiple familyinfluences on early-onset child conduct problems, weexamined a sample of boys at high risk for antisocialbehavior because of family history, gender, and urbanresidence. Urban residence (Garbarino et al., 1992;Tolan and Henry, 1994; Walter et al., 1995), malegender (American Psychiatric Association, 1994), andhaving a sibling with a history of delinquency (Joneset al., 1980; Patterson, 1984; Rowe and Herstand,1986; Rowe et al., 1992; Twito and Stewart, 1982)are well-documented risk factors for antisocial behavior.We examined the prospective associations among pa­rental involvement, monitoring, and parent-child con­flict, as they relate to later child conduct problems.

METHOD

Subjects

Subjects were 126 youths at high risk for conduct problemsand their families. Risk status was defined by urban residence,male gender, and the presence of a sibling (half or full) with arecent juvenile court conviction. We first screened the court recordsof every male delinquent (N = 1,614) found guilty in Manhattanor Bronx Family Court from November 1991 until October 1993.Cases dismissed in court or for which the juvenile was found "notguilty" are "sealed" by law, and therefore they were not examined.Families of female delinquents adjudicated during that time framewere not included because of their lower frequency (n = 110).

New York City Family Court probation reports were screenedto identify families of delinquents with a younger brother 6 to 10years old who resided in Manhattan, Brooklyn, or the Bronx, NewYork. Information on gender, age, and residence of the delinquent'ssiblings was taken from the probation report. Probation reportswere unavailable for 272 cases. Of the 1,342 with reports, 1,163were ineligible because (1) there was no younger brother in thetargeted age range (n = 1,055); (2) the family resided outside thedesignated area (n = 81); or (3) the younger sibling had a substantialphysical or mental disability (n = 4), was in foster care (n = 16),or did not reside with the delinquent (n = 7) at the time of arrest.

Of 179 eligible families, 28 were unreachable and 42 refused.In families with more than one age-appropriate son, all were askedto participate. One hundred nine families with 126 eligible boysconsented. Eligible families (n = 179) did not differ from ineligibleones (n = 1,163) on the age of delinquent youth, erhnicity, boroughin which the delinquent committed the crime, and severity ofcriminal charge; participating (n = 109) and eligible, nonparticipat­ing families (n = 70) did not differ on these measures. Onaverage, 7.1 months elapsed between the delinquent's arrest andour home interview.

The delinquent older brother was, on average, 14.8 years old(ranging from 10 to 17 years), and the biological mother was theprimary caregiver in 98 of the 109 participating families. In theremaining families, the primary caregivers were the biologicalfathersin two, the grandmother in eight, and the adoptive mother in

1228 ]. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 35:9, SEPTEMBER 1996

one. When the biological mother was not the primary caregiver,the primary caregiver was interviewed about the biological mother,as well as about herself or himself. Below, "moth er" refers tobiological mothers, and "caregiver" or "parent" refers to theprimary caregiver.

Procedures

Study procedures were approved by the New York State Psychiat­ric Institute/Columbia University Institutional Review Board. Eligi­ble families were sent a letter describing the study. Families whodid not wish to be contacted were asked to return a "do notcontact" postcard. Families not declining further contact weretelephoned or visited at home 2 weeks later. Study procedureswere explained, consent was obtained, and a home interview wasscheduled. Interviewers were college graduates with prior experienceconducting standardized interviews with children and families, whowere familiar with the communities in which study families resided.

During the 3- to 4-hour year I home interview, caregiversreported on their own functioning, parenting practices, and theirchildren's functioning, including psychiatric diagnostic status . Ob­servations were made of the quality of parenting. Latino familieswere interviewed by bilingual staff, and questions were read aloudfor parents not proficient in reading. Information about parentingpractices was collected from children by direct interview.

Approximately 15 months later, 112 boys in 98 families weresuccessfully reassessed (child behavior and diagnostic information)in a clinic visit. This article reports on demographic characteristics,parenting, behavioral, and diagnostic information collected at yearsI and II.

Instruments and Measures

Instruments unavailable in Spanish were translated to Spanishand back-translated into English to ensure accuracy and readability.

Demographic Information. Demographic information includedchild's age, family composition, parental education, ethniciry, andoccupational status (Stevens and Cho, 1985).

Parenting Practices. Three instruments assessed parenting: theHome Observation for Measurement of the Environment (HOME)scale, the Conflict Tactics Scale (CTS) , and components of thePittsburgh Youth Study Survey (PYSS).

HOME. The elementary-age version of the HOME (Caldwelland Bradley, 1984) indexed the quality of the home environment.The semistructured instrument combines observational and inter­view information on aspects of the home, such as parental warmthand acceptance toward the child and physical punishment. Items arescored as present/absent. The instrument has excellent psychometricproperties (Bradley et al., 1989) . While the entire instrument wasadministered, the subscales tapping Emotional Climate, EmotionalResponsiviry, and Encouragement of Maturity were selected on ana priori basis for this report because they are based largely onobservations of the emotional quality of the parent-child relation­ship. These were summed into a single scale, referred to below asEmotional Support.

CTS. We adapted the CTS (Straus , 1979 ; Straus et al., 1980 )to measure the extent to which parents used conflict during disputeswith their children. The CTS was originally developed for assessingconflict between adult domestic partners and was standardized ona large, stratified sample (Straus et al., 1980) . The CTS generatesfour subscales: Reasoning, Verbal Aggression, Violence, and LethalViolence. The Verbal Aggression and Violence subscales werecombined into a single scale, which we termed "Fighting." We

PREDICTORS OF CONDUCT PROBLEMS

did not consider the other twO scales in this report because LethalViolence was extremely rare, and Reason ing has not been foundto be associated with child behavior problems.

PYSS. Caregivers and children were asked questions from thePYSS (Loeber er al., 1991b), including measures of Punishment,Rejection, Monitoring, and Communication. Components of thissurvey have been associated with antisocial behavior in youngchildren (Loeber et al., 1991 b).

Constructs. Selected scales from the above instruments were usedto create three parenting constructs: Involvement, Parent-ChildConflict, and Monitoring. Table 1 presents the components ofthese constructs, their psychometric properties, and sample items .

For each construct, two informants provided information: parentand observer (Involvement), or parent and child (Parent-ChildConflict and Monitoring). For two child report measures, Punish­ment and Monitoring, data were unavailable for eight childrenwho did not understand the questions. The average age of excludedchildren was 7.5, but 8.9 for included children, suggesting thattheir age made comprehension difficult.

ChildSymptomatic Behavior. Parents completed the Child Behav­ior Checklist (CBCL/4-18) (Achenb ach, 1991), a 113-item check­list that has adequate reliability and stability (e.g., Achenbach,1991). It generates eight subscales, a Total scale, and two broad­band scales, Externalizing and Internalizing. The Externalizing scaleis the primary outcome measure in the present study.

Caregivers were also administered the D iagnostic InterviewSchedule for Children (DISC-P, 2.3) (National Institute of MentalHealth, 1995; Shaffer et al., 1996). This structu red interview isthe most extensively tested of all the child and adolescent diagnosticint erviews. Test-retest and interrater reliability of DISC diagnosesare in the "good" range (Jensen et al., 1995 ).

Statistical Analyses

Frequencies were obtained for demographic characteristics, be­havior problems, and diagnostic status . Correlations examinedassociations among demographic features, parenting practices , andchild behavior problems to confirm our a priori formulation ofconstruct composition.

Hierarchical linear regression analyses evaluated the relative con­tribution of each year I parenting practice to subsequent conductproblems. The percentage of variance explained at each step wascalculated, controlling for effects of prior steps. In these analyses,the three domains of parenting practices were entered, one at atime , with the variables in each domain entered together. Becausewe had no a priori assumptions about the relative weight of thethree parenting constructs, entry order was varied systematically.

Examination of the direction of parent-child effects is compli­cated because at any given time, for example , disruptive childrenare likely to elicit increased punishment from their parents. Tocontrol for aspects of year I parenting that might have been reactiveto child year I behavior, we next controlled for year I Externalizingscores, examining the impact of parenting on changes in Externaliz­ing scores between years I and II.

To test whether characteristics of the 16 families with morethan one stu dy child influenced our results , all regression analyseswere repeated, randomly removing one brother from analyses, forall families having more than one son in the study. In one case,where three brothers were in our sample, we removed two brothersfrom analyses. This procedure did not alter our findings.

Missing Data. Missing data were rare, occurring primarily onyoung children's reports. If 50% or more of the items from a scalewere available, we estimated missing data from the mean ofavailable

]. AM . ACAD. CHILD ADOLESC. PSYCHIATRY. 35:9. SEPTEMBER 1996 1229

WASSERMAN ET AL.

TABLE 1Parenting Constructs

Constructs/Components SourceNo. ofItems a Sample Items

InvolvementEmotional Support

Communication

Parent-Child ConflictRejection

Punishment

Punishment

Fighting

HOME: Emotional Climate, 25Emotional Responsivity, En­couragement of Maturity(mother and observer report)

PYSS: S/I (mother report) 4

PYSS: S/I (mother report)

PYSS: Discipline scale (D!) 3(mother report)

PYSS: DI (child report) 3

CTS: Verbal Aggression, Violence 14scales (mother report)

.76 "Has your son done anything that you areespecially proud of this week?"

.80 "When was the last time you talked withyour son about what he had actually doneduring the day?"

NA "In general, is time spent with your sonenjoyable?"

.45 "When your son does something that he isnot allowed to do, do you slap or spankhim with something?"

.52' "If you do something that you are not al­lowed to do or that your parents don'tlike, does your mother/father slap or spankyou or hit you with something?"

.79 "To settle a dispute, how often have youkicked, bit or hit with a fist?"

MonitoringMonitoring

Monitoring

PYSS: S/I (mother report)

PYSS: S/I (child report)

9

8

.59 "When your son is out, do you know whattime he will be home?"

.64 "When you are out, do your parentis) knowwhat time you will be home?"

Note: HOME = Home Observation for Measurement of the Environment; PYSS = Pittsburgh Youth Study Survey;S/I = Supervision/Involvement; DI = Discipline scale; CTS = Conflict Tactics Scale; NA = not applicable.

items within that scale for that subject. When fewer than 50% ofitems from a subscale were available, that scale was dropped forthat subject. In these cases, analyses used pairwise deletion of caseshaving missing data on any given variable.

RESULTS

Sample Characteristics

Demographic InfOrmation. At first assessment, 54%of subjects were African-American, 44% Hispanic, and2% Caucasian, reflecting the New York City adjudi­cated population from which siblings were drawn. Atfirst assessment, children were, on average, 8.9 yearsold and in third grade. Caregivers had completed, onaverage, 11 yearsofschool. Twelve percent ofcaregivershad never worked for pay. Of those who had workedfor pay, the average most recent occupational statuswas comparable to that of a child care worker.

Sixty-seven percent of children had not resided withthe same adult family constellation continuously sincebirth. Twenty-one percent currently resided with both

their biological parents. At first interview, 37% ofcaregivers had lived with a partner for at least theprevious 6 months.

At year II, data were available for 112 boys in 98families. Families seen in year II did not differ fromthose lost to follow-up on maternal ethnicity, maternaleducation, maternal employment, family receipt ofpublic assistance, child age, or year I Externalizingscores. Mothers of successfully reassessed children weresignificantly more likely to be living with a spouse/partner in year I (X2 [1] = 6.40, P < .05).

Diagnostic Status. At year I, on the DISC-P (DSM­III-R), 25% (31 boys) of the sample met criteria fora disruptive disorder (oppositional defiant disorder[ODD], conduct disorder [CD], or attention-deficithyperactivity disorder [ADHDD, 31% (39 boys) foran anxiety disorder, 2% (2 boys) for an affectivedisorder, and 24% (30 boys) for another disorder (e.g.,enuresis). Thirty-eight percent (48 boys) did not meetcriteria for any disorder at year I. As Figure 1 reflects,

1230 j. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 35:9, SEPTEMBER 1996

PREDICTORS OF CONDUCT PROBLEMS

Fig.2 Comorbidity of disruptive diagnoses (n = 31).

Parenting and Externalizing Behavior Problems

Associations amongParenting Measures. Year I parent­ing constructs (Involvement, Parent-Child Conflict,and Monitoring) were composed of measures selectedon an a priori basis. Table 2 presents correlationsamong year I parenting measures and CBCL scoresobtained at years I and II. The two components ofInvolvement were significantly correlated, as were thefour components of Parent-Child Conflict and thetwo components of Monitoring, suggesting that thesemeasures can be clustered together into related butdistinct constructs.

Associations with Concurrent (Year I) Behavior Prob­lems. As expected, parents who provided lower levelsof emotional support or communication had sons withmore conduct problems. Similarly, parents who usedmore punishment, more fighting, and more rejectionhad children with more conduct problems. Finally,children who reported that their parents fail to monitortheir whereabouts were more likely to have conductproblems. Demographic variables were unrelated toExternalizing scores at either assessment, and thereforethey are not included in further analyses.

Associations with Year II Behavior Problems. Parentingmeasures were similarly related to year II behaviorproblems. Lower levels of parental Involvement pre­dicted more Externalizing problems a year later. Higherlevelsof parental Rejection and Fighting were associatedwith higher year II Externalizing scores. In contrast

Internalizing scales were highly correlated (r = .72 atyear I). At year I, 14% (n = 17) had clinically elevatedscores on both Externalizing and Internalizing scales;9% (n = 11) had elevated Externalizing, but notInternalizing, scores; and 9% (n = 11) had elevatedInternalizing, but not Externalizing, scores. Year IIoverlap across domains was comparable.

As expected, rates of year I child disruptive behaviordiagnoses were related to year I Externalizing scores andrates of child anxiety/affective diagnoses were relatedto Internalizing scores. Sixty-one percent of childrenscoring above the CBCL Externalizing clinical cutoffreceived a diagnosis of at least one disruptive disorder,compared with 14% of those below the cutoff. Fifty­seven percent of children scoring above the CBCLInternalizing clinical cutoff received a diagnosis of atleast one anxiety/affective disorder, compared with 33%of those below the cutoff.

AnxietyN-39

ConductDisorder

N-4

AffectiveN-2

2011

6

Other (Enuresis. Tics. etc.>N-40

8

DisruptiveDisorder

N-31

OppositionalDisorder

N-15

Attention-DeficitDisorder --~

N-27

Fig. 1 Comorbidity of DSM-III-Rdiagnoses (no disorder: n = 48).

the sample showed considerable comorbidiry, with 33%of the children meeting criteria for two or morediagnoses.

Within the disruptive disorders (Fig. 2), 87% ofboys (27 boys) met criteria for ADHD, 48% (15 boys)for ODD, and 13% (4 boys) for CD. Twelve of the16 children with diagnoses of either ODD or CD alsomet criteria for ADHD.

Behavior Problems. At first assessment, 32% of thesample had elevated (above the 98th percentile) TotalCBCL scores. Twenty-two percent scored above theCBCL Externalizing scale cutoff (22% for Internaliz­ing). By year II, these figures had increased to 27%and 25%, respectively. As expected, Externalizing and

j. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 35:9, SEPTEMBER 1996 1231

WASSERMAN ET AL.

TABLE 2Correlations among Demographics and Parenting, and Their Associations with Child Behavior

.06 .09-.25**

2 3

Demographics1. Child age2. Ethniciry''"3. Mother's education

Involvement4. Emotional Support5. Communication

Parent-Child Conflict6. Rejection7. Punishment

(parent report)8. Punishment

(child report)"9. Fighting

Monitoring10. Monitoring

(parent report)11. Monitoring

(child report)12. Yr I Externalizing13. Yr I Internalizing14. Yr II Externalizing"15. Yr II Internalizing"

4 5 6 7 8 9 10 11 12 13 14 15

.14 .09 .17t -.01 -.10 -.01 .14 -.08 .07 .06 -.10 .01- .25** - .25** -.27** -.05 -.05. -.06 .10 .05 .01 .12 .09 .26**-.37*** .14 -.03 -.05 -.10 -.10 .14 -.07 -.09 -.08 -.In -.21*

.36*** .02 -.03 .01 -.10 .31** .22* -.23* -.13 - .32*** - .32***.14 .10 -.06 .03 .22* .08 -.19* -.20* - .37*** - .53***

.33*** .09 .29*** - .23* -.06 .21* .13 .18t -.12.24** .41*** -.10 -.10 .36*** .27** .08 -.13

.13 -.16t -.06 .30*** .10 .20* -.01

-.08 -.13 .25** .11 .28** .03

.26** -.06 -.06 -.29** -.16t

-.23* -.12 -.19t -.02

.71*** .59*** .37***.31*** .35***

.60***

an = 123.b Scored as African-American = 1, Latino = 2.en = 122.-d n = 112.*p < .05; **P < .01; *** P < .001; t p < .10.

to the lack of association between parent report ofMonitoring and year I Externalizing problems, bothparent and child report of higher levels of Monitoringpredicted increased Externalizing problems a year later.

Predicting Changes in Externalizing Behavior from

Parenting Practices

Table 3 shows results of analyses predicting year IIExternalizing scores from year I parenting measures.Controlling for the effectsofInvolvement and Monitor­ing, Parent-Child Conflict explained 10% of the vari­ance in Externalizing scores. Controlling for the effectsof the other two parenting constructs, Involvementaccounted for an incremental 13% of the variance; aftercontrolling for Parent-Child Conflict and Involvement,Monitoring was not significantly related to Externaliz­ing scores. Entry order of the three constructs did notalter these findings considerably. Thus, Parent-ChildConflict and Involvement each made a significant andindependent contribution to year II Externalizing, with

the final model accounting for 31% of the variancein Externalizing scores (F[8,95] = 5.37, P = .0000).

To examine parenting influences on changes inExternalizing scores, we next controlled for year IExternalizing scores, introducing it into the regressionequation before year I parenting. As Tables 2 and 3show, while stabiliry is high for Externalizing (r =

.58), parenting measures still account for independentvariance in outcome. Altogether, the three parentingfactors explain 17% of the variance in changes inExternalizing scores over a l-year period. Maternalreports of less Communication, less Punishment, moreFighting, and less Monitoring at year I were all associ­ated with increases in externalizing behavior betweenyear I and year II. When other aspects of year Iparenting were controlled, parental Involvement andParent-Child Conflict each explained significantly in­dependent portions of the variance in Externalizing(6% and 4%, respectively). When other aspects ofyear I parenting were controlled, the effect of year I

1232 ]. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 35:9, SEPTEMBER 1996

PREDICTORS OF CONDUCT PROBLEMS

TABLE 3Results of Regression Analyses, Explaining Year II Child Externalizing Behavior

Year II Externalizing

Year I Measures B SE B ~ P B SE B ~ P

Externalizing .525 .084 .539 .000Involvement

Emotional Support -.356 .228 -.150 .122 -.141 .196 -.059 .474Communication -1.323 .415 .299 .002 -.895 .357 -.202 .014

ConflictRejection 2.825 1.838 .146 .128 1.330 1.571 .069 .400Punishment (parent report) -.886 .917 -.095 .337 -2.121 .799 -.228 .009Punishment (child report) 1.062 .651 .145 .106 .138 .569 .019 .809Fighting .236 .095 .237 .015 .205 .081 .206 .013

MonitoringParent report -.423 .436 -.093 .334 -.840 .374 -.186 .027Child report -.207 .288 -.065 .475 .083 .247 .026 .737

Note: The first 4 columns present analyses without controlling for year I Externalizing. The second 4 columns presentanalyses first controlling for year I Externalizing. B = regression coefficient; SE B = standard error of the regressioncoefficient; ~ = standardized regression coefficient.

Monitoring on year II Externalizing was only margin­ally significant (2% of the variance). Altogether theentire model, regardless of the entry order of theconstructs, explained 51% of the variance in year IIExternalizing scores (F[9,94] = 11.05, P = .0001).Lower levels of Communication, and parent-reportedPunishment and Monitoring at year I, were all pre­dictive of increased year II externalizing behavior (Ta­ble 3).

DISCUSSION

This article examines three aspects of parenting(involvement, parent-child conflict, and monitoring)as they relate to changes in child conduct problemsover time. As hypothesized, each parenting constructwas independently related to child conduct problems.While continuiry in other dimensions of the child'ssocial environment, such as exposure to deviant peers,likely plays a role in the escalation of child behaviorproblems, the present results suggest that characteristicsof parenting play a substantial role in the worseningof conduct problems over time.

Explaining Conduct Problems

Parental Involvement. Social Control Theory (Hir­schi, 1969) posits that a child's attachments to conven­tional parents and to society's institutions protectsagainst the development of antisocial behavior (e.g.,

Smith et al., 1991). The current measure of involve­ment captures aspects of this process that relate toyoung children's family environments. Beyond theexpected concurrent relationship, lack of involvementrelated to increases in conduct problems over a l-yearperiod. The degree of communication between parentand child is the aspect of involvement that significantlyrelates to lower levels of future behavior problems.

The present measure of involvement was partlybased on observations from the HOME scale.Althoughwidely acknowledged as a predictor of 1(1 its associa­tion with childhood behavior problems is less welldocumented. Recent studies with economically disad­vantaged populations have similarly linked HOMEscale scores to child behavior problems (Dubow andLuster, 1990; Duncan et al., 1994). Using the HOMEscale for assessing parenting predictors of externalizingbehavior is particularly important because it is largelybased on blind observations ofparent-child interactionsin an ecologicallyvalid setting. Studies of the parentingof children with behavior problems are often limitedby reliance on parent report, which may reflect parents'perceptions more than actual practices, and may befurther biased by parents' efforts to provide sociallydesirable responses.

Parent-Child Conflict. Parent-child conflict was astrong predictor of conduct problems, consistent withnumerous studies showing that physical punishmentand verbal aggression increase the probability of devi­ance and aggression (Patterson et al., 1987; Nagaraja,

j. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 35:9, SEPTEMBER 1996 1233

WASSERMAN ET AL.

1984; Reiss et al., 1995). In the present sample, bothparent and child report of punishment were signifi­cantly positively correlated with current conduct prob­lems, although concurrent assessment obscures thedirectionality of the finding. The finding that parentsuse more punishment with children who exhibit moreconduct problems may indicate that parents react toincreasing levels of child behavior problems by increas­ing the severity of disciplinary and coercive strategies(Patterson et al., 1992).

Similarly, except for parent report of punishment,all aspects of parent-child conflict were positively re­lated to child outcome. However, if we examine therelation between parent-child conflict and later con­duct problems when all other aspects of earlier parent­ing and externalizing behavior are controlled,interesting findings emerge. While parent report offighting remained significantly positively related toconduct problems, and child report of discipline be­came nonsignificant, parent report of punishment wassignificantly negatively related to conduct problems.Examining the contents of the different scales mayhelp explain this apparent discrepancy. A high levelof fighting means that parents report more pushing/grabbing, insulting, or kicking in the course of adisagreement (this is positively related to subsequentconduct problems). Parent report of a higher level ofpunishment indicates that they yell/scold, slap/spank!hit, or send the child out of the house when he doessomething he is not allowed to do (and this is negativelyrelated to subsequent conduct problems). While weshould guard against post hoc interpretation of differ­ences among the component measures shown in Table3, findings may indicate that once the conflictualaspect ofparent-child interaction is taken into account,greater discipline, harsh or otherwise, reduces conductproblems. These results provide support for distinguish­ing punishment, defined here as consequent-based,from fighting, which may also reflect parental irritabilityand other unmeasured aspects of household stress.

Monitoring. Monitoring and supervision are amongthe most powerful predictors of juvenile delinquentbehavior (Loeber and Stouthamer-Loeber, 1986; Sny­der and Patterson, 1987). By monitoring their children,parents may influence their children's selection offriends and activities (e.g., Patterson et al., 1992).Although, in the present sample, monitoring accountedfor only a relatively small portion of variance, parent

report of monitoring was significantly negatively associ­ated with conduct problems over time.

Whenever we consider aspects of the parent-childrelationship at one point in time, we may lose sightof the fact that each partner reacts both to the currentstate of the relationship and to its history. In the caseof the parenting of children with conduct problems,current levels of conflict, involvement, and monitoringdepend also on the parent's prior interactions with thechild and the parent's perceptions about the child.The relative contribution of mother, child, and thehistory between them is highlighted in an observationalstudy of mothers of boys with CD and normal boys(Anderson et al., 1986). Mothers were observed inter­acting with their own sons and with boys with similarand different diagnoses from their own. Mothers weremore negative and controlling with boys who hadCD, compared with those who did not have CD,underscoring the degree to which children drive behav­ioral interactions. Mothers were more negative withtheir own CD boy than with an unfamiliar CD boy,suggesting the important influence of the history ofthe parent-child relationship on current interactions.

Limitations

We endeavored to obtain multi-informant reportsofparenting practices. While each construct was signifi­cant as a separate step in the equation predicting futureantisocial behavior, parent report was more stronglyrelated to parent-reported conduct problems than weremeasures derived from the child or from blind observa­tions. While similar results have been obtained else­where (Loeber et al., 1991a), common method variancemay have inflated the strength of associations. Further­more, this study considers processes in the families ofboys at risk for early-onset conduct problems who alsohave an antisocial sibling; therefore, the degree towhich these findings are generalizable to other groupsof children is unknown.

Clinical Implications

These results offer several suggestions for clinicalinterventions. Most notably, as supported by a largebody of literature (for a review see Loeber and Stou­tharner-Loeber, 1986), definable aspects of parentingcan be targeted for change. Behavioral parent trainingprograms have been shown to be effective in modifyingparent and child behavior. Moreover, the independent

1234 J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 35:9, SEPTEMBER 1996

contribution of the three parenting components studiedhere underscores the need for interventions that addresseach component. Treatment goals with disruptive chil­dren should include decreasing conflict, increasing posi­tive involvement, and increasing supervision. Finally,because signs of child conduct problems are very oftenapparent in early childhood, preventive interventionsthat target young children and their families, and thosethat screen for high-risk children, seem justified. Inaddition, clinicians seeing already disruptive childrenshould be aware of the potentially elevated risk statusof other siblings.

Summary

In summary, we found that three different dimen­sions of parenting practices-parent-child conflict,monitoring, and parental involvement-each contrib­ute independently to the emergence of young children'sconduct problems. This supports a cumulative modelof family risk, consistent with recent reviews (e.g.,Farrington, 1990; Loeber and Stouthamer-Loeber,1986; Lytton, 1990), such that each negative familyfactor further compounds the likelihood ofthe develop­ment of child behavior problems. The present evidencefor the predictive role of parenting suggests that alteringchildrearing practices may help alter the developmentalcourse for children with early-onset conduct problems.

REFERENCES

Achenbach TM (1991), Manualfor the ChildBehavior Checklist/4-16 and1991 Scoring Profile. Burlington: University of Vermont Deparcrnentof Psychiatry

American Psychiatric Association (1994), Diagnostic and Statistical ManualofMentalDisorders, 4th edition (DSM-IV). Washington, DC: AmericanPsychiatric Association

Anderson KE, Lytton H, Romney DM (1986), Mothers' interactionswith normal and conduct-disordered boys: who affects whom? DevPsychol 22:604-609

Andrew JM (1981), Delinquency: correlating variables. J Clin Child Psy­choll0:136-140

Bradley RH, Caldwell BM, Rock SL et al. (1989), Home environmentand cognitive development in the first 3 years of life: a collaborativestudy involving six sites and three ethnic groups in North America.Dev PsychoI25:217-235

Caldwell BM, Bradley RH (1984), Home Observation for Measurement ofthe Environment. Little Rock: University of Arkansas

Cicchetti D, Richters J (1993), Developmental considerations in the investi­gation of conduct disorder. Dev PsychopathoI5:331-344

Conduct Problems Prevention Research Group (1992), A developmentaland clinical model for the prevention of conduct disorder: the FASTTrack Program. Dev PsychopathoI4:509-527

Dubow EF, Ippolito MF (1994), Effects of poverty and quality of the homeenvironment on changes in the academic and behavioral adjustment ofelementary school-age children. J Clin Child PsychoI23:401-412

PREDICTORS OF CONDUCT PROBLEMS

Dubow EF, Luster T (1990), Adjustment of children born to teenagemothers: the contribution of risk and protective factors. J Marr Fam52:393-404

Duncan GJ, Brooks-Gunn J, Klebanov PK (1994), Economic deprivationand childhood development. Child Dev 65:296-318

Eron LD, Huesmann LR, Zelli A (1991), The role of parental variablesin the learning of aggression. In: The Development and Treatment ofChildhoodAggression, Pepler DJ, Rubin KH, eds. Hillsdale, NJ: Erlbaum,pp 169-188

Farrington DP (1983), Offending from 10 to 25 years of age. In: ProspectiveStudies of Crime and Delinquency, Van Dusen KT, Mednick SA, eds.Boston: Klumer-Nijhoff, pp 17-37

Farrington DP (1990), Implications of criminal career research for theprevention of offending. J Adolesc 13:93-113

Forgatch MS (1991), The clinical science vortex: developing a theory forantisocial behavior. In: The Development and Treatment of ChildhoodAggression, Pepler D, ed. Hillsdale, NJ: Erlbaum, pp 291-315

Garbarino J, Kostelny K, Dubrow N (1992), Children in DangerousEnvironments: Coping with the Consequences of Community Violence. SanFrancisco: Jossey-Bass

Henggeler S (1989), Delinquency in Adolescence. Newbury Park, CA: SageHirschi T (1969), Causes of Delinquency. Berkeley: University of Califor­

nia PressJensen P, Roper M, Fisher P et al. (1995), Test-retest reliability of the

Diagnostic Interview Schedule for Children (DISC 2.1): parent, child,and combined algorithms. Arch Gen Psychiatry 52:61-71

Jones MB, Offord DR, Abrams N (1980), Brothers, sisters, and antisocialbehavior. BrJ Psychiatry 136:139-145

Loeber R, Dishion TJ (1983), Early predictors of male delinquency: areview. Psychol Bull 94:68-99

Loeber R, Green S, Lahey B, Stouthamer-Loeber M (1991a), Differencesand similarities between children, mothers, and teachers as informantson disruptive child behavior. J Abnorm Child PsychoI19:75-95

Loeber R, Stouthamer-Loeber M (1986), Family factors as correlates andpredictors of juvenile conduce problems and delinquency. In: Crimeand Justice: An Annual Review ofResearch, Tonry M, Morris N, eds.Chicago: University of Chicago Press, pp 29-149

Loeber R, Stouthamer-Loeber M, Van Kammen W, Farrington DP (1991 b),Initiation, escalation and desistance in juvenile offending and theircorrelates. J Crimin Law Criminol82:36-82

Luster T, Dubow E (1992), Home environment and maternal intelligenceas predictors of verbal intelligence: a comparison of preschool andschool-aged children. Merrill-Palmer Q 38:151-175

Lyons MJ, True WR, Eisen SA et al. (1995), Differential heritability ofadult and juvenile antisocial traits. Arch Gen Psychiatry 52:906-915

Lytton H (1990), Child and parent effects in boys' conduct disorder: areinterpretation. Dev Psychol 26:683-697

McCord J (1979), Some child-rearing antecedents of criminal behavior inadult men. J Pers Soc PsychoI8:1477-1486

McGuffin P, Gottesman II (1985), Genetic influences on normal andabnormal development. In: Child and Adolescent Psychiatry: ModemApproaches, Rutter M, Hersov L, eds. Boston: Blackwell ScientificPublications, pp 17-33

Moffitt TE (1993), Adolescence-limited and life-course persistent antisocialbehavior: a developmental taxonomy. Psychol Rev 100:674-701

Nagaraja J (1984), Losing Ground: American Social Policy, 1950-1980.New York: Basic Books

National Institute of Mental Health (1995), Diagnostic Interview Schedulefor Children. Rockville, MD: National Institute of Mental Health

Patterson GR (1984), Siblings: fellow travelers in coercive family processes.In: Advances in the Study of Aggression, Blanchard RJ, Blanchard DC,eds. Orlando, FL: Academic Press, pp 173-215

Patterson GR, Bank L, Reid JB (1987), Delinquency prevention throughtraining parents in family management. Behav Manage 10:75-82

Patterson GR, Reid JB, Dishion T (1992), Antisocial Boys: A SocialInteractional Approach. Eugene, OR: Castalia

Patterson GR, Srouthamer-Loeber M (1984), The correlation of familymanagement practices and delinquency. Child Dev 55:1299-1307

]. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 35:9, SEPTEMBER 1996 1235

WASSERMAN ET AL.

Reiss D, Hetherington EM, Plomin Ret al. (1995), Genetic questions forenvironmental studies. Arch Gen Psychiatry 52:925-936

Richter LM, Grieve KW (1991), Home environment and cognitive develop­ment of black infants in impoverished South African families. InfMentHealth 112:88-101

Rowe DC, Herstand SE (1986), Familial influences on television viewingand aggression: a sibling study. Aggressive Behav 12:111-120

Rowe DC, Rodgers JL, Meseck-Bushey S (1992), Sibling delinquencyand the family environment: shared and unshared influences. ChildDev 63:59-67

Sampson RJ, Laub JH (1993), Crime in theMaking: Pathways and TurningPoints through Life. Cambridge, MA: Harvard University Press

Shaffer D, Fisher P, Dulcan M et al. (1996), The NIMH DiagnosticInterview Schedule for Children Version 2.3 (DISC-2,3): descriprion,acceptability, prevalence rates, and performance in the MECA study.I Am Acad ChildAdolesc Psychiatry 35:865-877

Smith C, Weiher WA, Van Kammen BW (1991), Family attachment anddelinquency. In: Urban Delinquency and Substance Abuse: TechnicalReport, Huizinga D, Loeber R, Thornberry TP, eds. Washington, DC:Office of Juvenile Justice and Delinquency Prevention, pp 1-28

Snyder J, Patterson GR (1987), Family interaction and delinquent behavior.In: Handbook offuoenile Delinquency, Quay HC, ed. New York: Wiley,pp 216-243

Stevens G, Cho JH (1985), Socioeconomic indexes and the new 1980census occupational classification scheme. SocSci Res 14:142-168

Straus MA (1979), Measuring family conflict and violence. I Marr Fam41:75-88

Straus MA, Gelles RJ, Steinmetz SK (1980), BehindClosed Doors: Violencein the American Family. Garden City, NY: Anchor

Tolan PH, Henry D (1994), Patterns of psychopathology among urbanpoor children: the role of aggression. Paper presented at the AnnualMeeting of the American Academy of Child and Adolescent Psychiatry,New York

Twito TJ, Stewart MA (1982), A half-sibling study of aggressive conductdisorder: prevalence of disorders in parents, brothers, and sisters. Neuro­psychobiology 8:144-150

Walter HJ, Vaughan RD, Armstrong Bet al. (1995), Sexual, assaultive, andsuicidal behaviors among urban minority junior high school students. IAm Acad ChildAdolesc Psychiatry 34:73-80

1236 ]. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 35:9, SEPTEMBER 1996


Top Related