Infant and Toddler Pathways Leading to Early Externalizing Disorders

INTRODUCTION

Due to the stability and lack of treatment efficacy of conduct problems in older children, there has been an increased focus recently on the study of externalizing problems in early childhood (Reid, 1993). One consistent finding that has emerged is the increased risk status of children who display conduct problems prior to adolescence. "Early starters" show a more persistent and chronic course of antisocial behavior from middle childhood through young adulthood (Moffitt, 1990; Patterson et al., 1992). Beginning with the pioneering efforts of Glueck (1950) and Robins (1966), it has been hypothesized, but not empirically validated, that constitutional vulnerabilities and childrearing issues play a salient role in the genesis of early externalizing problems. More recently, Moffitt (1993) has suggested that early starters are characterized by neuropsychological deficits in the child that compromise verbal and executive functioning. Psychosocial risk factors are hypothesized to interact with child characteristics to place children at further risk for a persistent course of conduct problems. In support of this model, Moffitt (1990) demonstrated that children who displayed a comorbid pattern of attention deficit disorder and delinquent behavior (ADD/JD) at age 13 showed more chronic and severe conduct problems beginning at age 5 in comparison with children who had a later onset of conduct problems, or who had symptoms of ADD without conduct problems. Children in the comorbid ADD/delinquent group also showed higher rates of psychosocial adversity than other disordered and nondisordered groups, beginning at preschool age. This research is consistent with studies of adolescents which have shown the comorbid ADD/JD pattern is a better predictor of later criminal offending than delinquency alone (Loeber et al., 1991).

Like Moffitt, Patterson's (1992) research suggests that early starters are at greater risk for more serious criminal offending, but places a greater emphasis on the role of disrupted parenting. Patterson acknowledges that specific types of children are most likely to elicit inept parenting strategies, due to the child's irritability and/or hyperactivity, but believes that patterns of coercive parenting potentiate the onset of more serious conduct problems. In a coercive cycle, the parent and child each behave in a way that is aversive to the other in an attempt to control the other's behavior. As the child's aversive behaviors increase in intensity and frequency, the parent eventually acquiesces, unwittingly reinforcing the child's behavior. As the child becomes increasingly irritating, the parent further escalates power assertion techniques and presumably, the level of hostility directed towards the child.

While both Moffitt and Patterson have discussed potential pathways by which early starters begin their trajectory toward serious conduct problems, few researchers have investigated the relative contributions of child and parenting effects prior to preschool age, particularly with at_risk samples. The present study seeks to expand our understanding by examining infant and toddler pathways leading to different types of externalizing disorders assessed at ages 5.5 and 6, including comorbid diagnoses of Attention Deficit Hyperactivity Disorder (ADHD) and Oppositional Defiant Disorder (ODD) or Conduct Disorder (CD). Based on the work of Moffitt (1990) and Patterson and colleagues (1992), our goal was to examine child and psychosocial risk factors associated with different types of early_onset, DSM_based, disruptive disorders. We hypothesized that children with a comorbid diagnosis of ADHD_ODD/CD at school entry would demonstrate a more severe and pervasive number of risk factors across domains than children with only ADHD or without a DSM_based externalizing diagnosis. However, due to the severity and degree of conduct_disordered behavior necessary to meet diagnostic criterion for CD at age 5.5_6, it was hypothesized that children in the CD_only group would show a rate of child and psychosocial risk factors similar to the comorbid ADHD_ODD/CD group. Finally, because ODD involves noncompliance but not aggressive behavior, we expected risk factors that would compromise the quality of parent_child relationships to be impaired for children in the ODD group, such as parenting, parental conflict, and maternal adjustment, in comparison to nonproblem children.

Participants

The source for subject recruitment was low_income families who use the Allegheny County's Women, Infants, and Children (WIC) Program in the Pittsburgh metropolitan area (Shaw et al., 1998). Three_hundred and ten participants were recruited from WIC sites throughout the Pittsburgh metropolitan area over the course of 2 years. Since the intent of the original investigation was to examine the developmental precursors of antisocial behavior, the sample was restricted to boys because of their greater relative risk for engaging in serious antisocial activities. Participants were recruited when target children were between 6 and 17 months old. At the time of the first assessment, at which time infants were 1.5 years old, mothers ranged in age from 17 to 43 years, with a mean age of 28. Fifty_three per cent of participants were Caucasian, 36% were African American, 5% were biracial, and 6% were other (e.g., Hispanic). At the age 1.5 visit, 65% were either married or living together, 26% were single, 7% were divorced, and 2% were other. Mean per capita family income was $241 per month ($2,892 per year), and the mean Hollingshead socioeconomic status score was 24.8, indicative of a working class sample. At the age 2 assessment, data were available on 302 participants, an attrition rate of 2.6%. At 3.5 years, data were collected on 282 participants, an attrition rate of 9.1%. Due to limited funds, only 235 participants took part in the age 5.5 assessment, but four had missing data for the K_SADS interview. However, no significant differences were found on any sociodemographic variable from the age 1.5 assessment when families retained at age 5.5 were compared to those who were not available. Teacher report data were available on 186 of the 279 children interviewed for the age 6 assessment.

Procedure

Mothers and target children were seen when children were ages 1.5, 2, 3.5, 5.5, and 6 years old in the laboratory (ages 1.5, 2, 3.5, and 6) and/or at home (ages 2 and 5.5). At ages 1.5, 2, and 3.5,

observations of parent_child interaction occurred, during which time mothers completed several questionnaires about the child's behavior, her own adjustment, and family functioning (e.g., parental conflict, family criminality). Psychiatric impairment of the child was ascertained using maternal report (K_SADS) at age 5.5 and teacher report (Achenbach Teacher Report Form) at age 6. In selecting predictor variables from early childhood, measures were designed to tap developmentally salient child and parenting behavior. Child factors included negative emotionality at ages 1.5 and 2, attention at age 2, and child disruptive behavior at ages 1.5, 2, and 3.5. Dimensions of parenting included maternal rejection at ages 1.5 and 2 and quality of the home environment at age 2. Other risk factors included parental psychological adjustment (e.g., maternal depression), family functioning (e.g., parental conflict), and neighborhood dangerousness.

Measures

Child Characteristics

Infant Characteristics Questionnaire (ICQ). At the age 1.5 and 2 year assessments, mothers rated their infant on the ICQ, which taps different aspects of infant negative emotionality and activity (Bates et al., 1979). Because of the theoretical link between negative emotionality and externalizing problems, the Difficulty factor was used in the present study. The ICQ meets customary psychometric standards for maternal reports on infant temperament (Bates et al., 1979).

Toddler Behavior Checklist (TBC). At age 1.5, mothers completed the TBC, a 103_item behavior problem checklist similar to the CBCL but designed for children 9 months to 4 years of age (Larzelere et al., 1989). Two scales were of particular interest to the present study, Oppositional (22 items) and Physical Aggression (14 items), for which Cronbach alphas are .91 and .83, respectively.

Schedule for Affective Disorders and Schizophrenia for School_Age Children_Epidemiologic Version (K_SADS_E, Orvashel and Puig_Antich, 1987). The K_SADS_E was administered to mothers about their children's adjustment at age 5.5, from which current symptomatology necessary for establishing the presence of DSM_IV diagnoses was evaluated, including diagnoses of ADHD, ODD, and CD. To establish reliability, clinical interviewers participated in an intensive training program at the Western Psychiatric Institute and Clinic regarding administration of the interview. Additionally, every case in which a subject approached or met diagnostic criteria was discussed by the team, other interviewers, and the first author before reaching a final decision on diagnosis. Children meeting diagnostic criteria for one or more of the three externalizing disorders were placed in the following groups: ADHD, ODD, CD, and ADHD_ODD/CD. Table 1 provides frequencies on the distribution of children in each of the four externalizing groups and the nonproblem group.

Child Behavior Checklist (CBCL) for Ages 2_3 and Teacher's Report Form (TRF) (Achenbach, 1992; Achenbach and Edelbrock, 1986). These versions of the Achenbach behavior checklists are the most widely used reports of child behavior problems. The CBCL was administered to mothers at the age 2 and 3.5 assessments, from which three narrow_band factors involving externalizing symptoms were selected for use: Aggressive, Destructive, and Attention. The attention factor was created from 4 items of the CBCL at age 2 (e.g., "Can't concentrate, can't pay attention," Cronbach alpha = .56), and used to examine Moffitt's hypothesis about the role of attentional deficits in executive functioning. The TRF was administered to teachers when children were age 6. Because of our interest in deriving clinical cutoff scores that were comparable to the K_SADS disorders, clinically_derived factors were created based on item content that corresponded to the following DSM_IV diagnoses: ADHD (19 items), ODD (13), and CD (8 items). Cronbach alphas for these factors were .93, .92, and .91, respectively. Teacher_based diagnostic groups were formed by establishing cutoff scores based on within_sample statistics. Children rated at or above the 88^th percentile by teachers on these factors were placed in groupings similar to those derived from the K_SADS criteria. Rates of teacher_derived externalizing disorders are presented in Table 1.

Agreement was low between maternal and teacher reports of externalizing disorders. With the exception of nonproblem children, for whom agreement was 76%, maternal and teacher concordance in the identification of specific externalizing disorders was modest: 25% for the ADHD group, 28% for the comorbid group, and 50% for the ODD and CD groups.

Insert Table 1 here

Psychosocial Factors

Parenting Daily Hassles (PDH). This is a 20_item measure of events parents typically encounter with children. Mothers completed the PDH at age 1.5, 2, and 3.5 year assessments. Parents rate the frequency of occurrence on a 4_point scale and how hassled they felt by the event on a 5_point scale (Crnic and Greenberg, 1990). The sum of the frequency and intensity factors were used in the present analyses. Cronbach alphas for frequency and intensity are .81 and .90, respectively.

Conflict Tactics Scale (CTS_Form N) (Strauss, 1991). The CTS is a widely employed measure that evaluates the use of different types of conflict resolution strategies used by adult partners. It was administered to mothers at age 3.5. In the present study, both the Verbal and Physical Aggression factors were used, for which Cronbach alpha coefficients are high, .77_.88 across studies (Strauss, 1991).

Beck Depression Inventory (BDI). Mothers completed the BDI, a widely used measure of depressive states, at the age 1.5, 2, and 3.5 year assessments (Beck et al., 1988). Split_half reliability of the scale has been found to be high (.86 to .93).

General Life Satisfaction Questionnaire (GLS). Social stress and support were evaluated using the GLS when infants were 1.5 years old (Crnic et al., 1983). Respondents are asked to indicate their level of satisfaction with potential sources of social support, including neighbors, friends or family, organized groups, and intimate relationships. The Satisfaction factor was used in the present study.

Personality Research Form (PRF). The Aggression factor of the PRF was administered at the age 1.5 laboratory visit (Jackson, 1989). The 16_item Aggression factor was selected because it was postulated to influence parenting style and subsequent child behavior. The internal consistency reliability is high (.87), as is test_retest reliability (.85, Jackson, 1989).

Home Observation for Measurement of the Environment Inventory _ Infant Version (HOME) (Caldwell and Bradley, 1984). The HOME is a widely used measure of support and stimulation in the child's home environment, which includes both observational and interview components. We chose to use the Total HOME score to supplement laboratory ratings of specific parenting strategies. In addition to evaluating parental nurturance, the Total score includes domains pertaining to the organization and variety of the home environment, and provision of play materials.

Rejecting Parenting. Maternal rejecting parenting was assessed at age 2 using the Early Parenting Coding System (EPCS), which was designed to measure a range of parenting behaviors commonly seen in young children (Shaw et al., 1998). Observations of rejecting parenting were coded from a 5_minute clean_up task that was part of the age 1.5_ and 2_year laboratory assessments. A rejection factor was derived from a combination of two molecular (verbal/physical approval and critical statements) and three global (hostility, warmth, punitiveness) ratings. For molecular codes, kappa coefficients ranged from .79 to .87. For global ratings, all kappa coefficients ranged from .83 to .94.

Neighborhood Questionnaire. This 17_item questionnaire assesses the perceived dangerousness of the neighborhood, and was administered to parents at age 2. Parents rate the prevalence of antisocial behavior (e.g., illicit drug use) and dilapidation of the neighborhood (e.g., abandoned housing), from which one factor of dangerousness is formed (Pittsburgh Youth Study, 1991, University of Pittsburgh).

Between_group differences among the five externalizing disorder groups were examined in relation to child characteristics and psychosocial risk factors using maternal and teacher reports. Because sample sizes were small across disorder groups and informants, it was decided to control for Type I error post hoc by comparing the ratio of significant to nonsignificant results. Using this approach, risk of Type II error could be minimized while maintaining the ability to assess the probability that significant results were due to chance.

For each risk factor of interest, a Oneway Analysis of Variance (ANOVA) was computed in which a child or psychosocial risk factor was the dependent variable and externalizing disorder group was the independent variable. The results for mother_reported disorder groups are presented in Table 2.

In reference to K_SADS diagnosis at age 5.5, several differences emerged among groups in relation to both child characteristics and psychosocial risk factors. In 17 of 23 comparisons (73.9%), children with the comorbid diagnosis of ADHD_ODD/CD showed significantly more deviant scores than children in the nonproblem group. Comorbid children were characterized as more temperamentally difficult, having more attentional problems, and being more oppositional, aggressive, and destructive from age 1.5 to 3.5. Mothers of comorbid children also demonstrated higher rates of depressive and aggressive personality symptoms, less social support, higher levels of rejecting parenting, and lived in more dangerous neighborhoods. Children from CD families also showed a similar pattern of negative emotionality during infancy and conduct problems from ages 2 to 3.5, but fewer psychosocial risk factors. While CD families showed significantly lower scores on the HOME and lived in the most dangerous neighborhoods, group differences in other aspects of the family environment were not evident. Children in the ODD group were less consistently found to show negative emotionality or child conduct problems in early childhood, with the exception of destructive behavior at ages 2 and 3.5, but had mothers who showed elevated symptoms of depression and aggressive personality, and the highest rejecting parenting and most deviant HOME scores. In contrast, ADHD children showed similar scores to nonproblem children in 22 of 23 instances, a percentage that would be expected to occur by chance (95.6%).

Insert Table 2 here

The same analytic strategy was applied to teacher_identified externalizing groups, reported in Table 3 below. In general, few group differences were found, particularly regarding child factors. Convergence with maternal report was more apparent for family and neighborhood factors, which were more often assessed using observational methods. Comorbid ADHD_ODD/CD children were more likely than nonproblem children to be exposed to physical parental conflict, experience higher rates of rejecting parenting, have less caring, well_organized, and more dangerous home environments. CD children were more likely than nonproblem children to be exposed to interparental verbal aggression and live in more dangerous neighborhoods. ODD children were rated as having greater destructive behavior problems at age 3.5, lower quality home environments, higher rejecting parenting, and more dangerous neighborhoods than nonproblem children. ADHD children continued to show few differences with nonproblem children (1 in 23, 4.4%), the lone exception being higher rejecting parenting at 24 months.

Insert Table 3 here

Based on maternal reports of children's psychiatric diagnosis, children in the comorbid ADHD_ODD/CD group demonstrated the most chronic and pervasive set of child, family, and neighborhood risk factors, especially in comparison to normal and ADHD children. Also according to maternal report, CD children showed a great frequency of child, family, and neighborhood risk factors in comparison to nonproblem children. Children in the ODD group were characterized primarily by family risk factors (e.g., maternal depression, rejecting parenting). ADHD children showed few differences in comparison to nonproblem children. When the same relationships were assessed using teacher_report of externalizing disorders, the comorbid ADHD_ODD/CD group, and to lesser extent, the CD and ODD groups, continued to show significantly more risk factors, but these were limited to the caregiving environment and neighborhood dangerousness.

To some extent, differences in risk factors that discriminated pathways leading to externalizing disorders reflect variation in mothers' and teachers' ratings of externalizing problems. Some of these differences may be attributed to the different measures used to ascertain clinical impairment, and the 6_month time period between evaluations. Alternatively, the results are consistent with the notion that the stronger relations found for maternal report of externalizing disorders are due to reporter bias, as mothers provided information on both risk factors and child psychiatric status. To meet criterion for DSM_IV diagnosis, mothers needed to verify that children demonstrated externalizing problems in multiple contexts outside of the home (i.e., school), and relationships were found using observed measures of rejecting parenting and quality of the home environment. Nevertheless, it is likely that, particularly for child factors (i.e., negative emotionality, attention), relations between early risk factors and later externalizing disorders were inflated. These findings are consistent with prior research on the relation between infant negative emotionality and later externalizing problems, which demonstrate modest or nonsignificant associations when informants other than parents are used to evaluate later externalizing problems (Sanson et al., 1991).

Another possible reason for the discrepancy in the identification process involves discontinuity in child behavior across contexts. While some consistency in child behavior would be expected to be demonstrated at home and in school, it is perhaps unrealistic to expect agreement in clinical cutoff scores to be high. In fact, when correlations between maternal (CBCL) and teacher (TRF) reports of clinically_derived factors of ADHD, ODD, and CD were computed at age 6, relations were significant: .18 (p < .01), .28 (p < .001), and .35 (p < .001), respectively. The findings suggest modest continuity across informant and setting, and are comparable in magnitude to those reported by others (Achenbach & Edelbrock, 1986). The findings also suggest that for many children, externalizing disorders in early childhood may be "contained" within parent_child relationships. That is, specific child characteristics may "push buttons" in parents, whose parental responses exacerbate the child's initial symptoms. Our findings suggest that only a subset of these children demonstrate clinically_elevated externalizing problems at school. It is likely that some of these "disordered" children respond differently to the structure of the classroom and/or their teachers. The results warrant caution in using the term "disorder" to describe young children's disruptive behavior. Interestingly, there is also evidence to suggest that agreement between teachers' and parents' ratings improve after the first year of school. Fisher and Fagot (1996) found convergence between parent and teacher reports to improve markedly (i.e., from correlations of _.05 to .53) between Kindergarten and second grade on symptoms of antisocial behavior, suggesting that parents' and teachers' perceptions of the child eventually influence each other, and that the child's behavior becomes more stable over time across context.

In accord with Moffitt's model, results from maternal report suggest that children in the comorbid ADHD_ODD/CD group experience severe psychosocial risk beginning in infancy relative to nonproblem children. Child characteristics, in the form of negative emotionality, attentional deficits, and early disruptive behavior, and psychosocial risk, in the form of perceived childrearing hassles, maternal depression and aggression, and rejecting parenting, contribute to the comorbid pattern at school entry. Similar child characteristics also appear to lead to the development of early_onset CD, but psychosocial risk factors are less prevalent. While for ODD children, maternal adjustment and quality of caregiving appear to be the primary influences.

Results based on maternal reports are also consistent with Patterson's model and attachment theorists (Sroufe,1983), which emphasize the salience of parenting and the family caregiving environment. That teacher ratings corroborated the significance of caregiving factors adds further credibility to models that highlight the significance of early parenting. Importantly, teacher findings cannot be attributed to shared reporting or method bias.

The results also need to be interpreted within a developmental context. The timing of the diagnostic evaluations is comparable to studying the development of schizophrenia in young adults; there are a substantial number of false negatives yet to emerge. However, data on the stability of early disruptive behavior suggest that there are relatively few children who show a "late" onset of serious externalizing problems after the preschool period (Shaw et al., 2000). Using the present sample, among boys identified at or above the 90^th percentile on the CBCL Externalizing factor at age 2, 62% remained in the clinical range and 100% remained above the median at age 6. However, only 16% of those below the median on Externalizing at age 2 moved into the clinical range at age six. Alternatively, because of the young age at which the assessments were conducted, many of the children identified as ODD will likely attain CD status in the next few years. Thus, conclusions about trajectories leading to specific externalizing disorders need to be interpreted with caution, an admonition that needs to be reinforced because of the modest cell sizes among externalizing disorder groups. It is also worth noting that measurement of toddler's neuropsychological risk was less than optimal, and could have contributed to the inconsistent support for Moffitt's model.

Finally, it is imperative to note that the validity of these findings may be limited to boys from low_income, urban settings. Further research is needed to understand risk factors associated with the development of early conduct problems in girls, particularly from comparably high_risk environments.

The results offer important information for clinicians to consider when evaluating the status of young children with serious conduct problems. Because findings across informants indicated that the quality of caregiving during infancy was associated with clinically_significant impairment at school entry, interventions during infancy and the toddler period that address both the structure of the home environment (e.g., daily bedtimes, provision of children's play areas), specific parenting strategies, and factors that compromise the quality of caregiving (e.g., maternal adjustment, social support, daily hassles) appear warranted. Such interventions have been carried out with moderate success working with preschool_ and school_age children (Webster_Stratton et al., 1994; Patterson et al., 1992). Moreover, research on the treatment efficacy of conduct problems suggests that interventions implemented prior to school_age have a higher probability of success (Reid, 1993).

Findings regarding early child characteristics also need to be taken into account. Although limited to maternal ratings of children's externalizing disorders, the relations between early negative emotionality/conduct problems and later externalizing disorders are most likely valid for child behavior within the family, and in a few instances predictive of adjustment at school. The role of child characteristics is further reinforced from inspection of the videotapes from which rejecting parenting was observed. In clean_up tasks, "rejecting" mothers were typically trying to manage a difficult infant. Thus, it is important that clinicians are sensitive to the child's actual behavior and the mother's perception of the meaning of that behavior in order to evaluate the rationale behind parenting strategies and attitudes (i.e., the child is showing "noncompliant" versus "spiteful" behavior).

In summary, the study provides data to suggest that children with specific types of early externalizing disorders show variation in the types of risk factors that precede their development. In support of Patterson's model, data from multiple sources corroborates the significance of the early caregiving environment. Only the continued study of children followed from early childhood will allow us to learn whether they are the same individuals who go on to engage in more serious antisocial activities. We hope to have data on this issue in the next few years.

References

Achenbach TM (1992), Manual for the Child Behavior Checklist 2/3 and 1992 profile. Burlington: University of Vermont Department of Psychiatry

Achenbach TM, Edelbrock C (1986), Manual for the Teacher's Report Form and Teacher Version of the Child Behavior Profile. Burlington: University of Vermont Department of Psychiatry

Bates JE, Freeland CA, and Lounsbury ML (1979), Measurement of infant difficulties. Child Dev 50: 794_803

Beck AT, Steer RA, Garbin MG (1988), Psychometric properties of the Beck Depression Inventory. Twenty_five years of evaluation. Clin Psych Review 8:77_100

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

Crnic KA, Greenberg MT (1990), Minor parenting stresses with young children. Child Dev 61: 1628_1637

Crnic KA, Greenberg MT, Ragozin AS, Robinson NM, Basham RB (1983), Effects of stress and social support on mothers and premature and full_term infants. Child Dev57: 209_217

Fisher PA, Fagot BI (1996), Development of consensus about child oppositional behavior: Increased convergence with entry into school. J Applied Dev Psychol17: 519_534

Glueck S (1950), Unraveling juvenile delinquency. New York: Commonwealth Fund

Jackson DH (1989), Personality research form manual (3rd Edition). New York: Research Psychologists Press

Larzelere RE, Martin JA, Amberson, TG (1989), The Toddler Behavior Checklist: A parent_completed assessment of social_emotional characteristics of young preschoolers. Fam Relations 38: 418_425

Loeber R, Stouthamer_Loeber M, Van Kammen V, Farrington DP (1991), Initiation, escalation, and desistence in juvenile offending and their correlates. J Crim Law and Criminology 82: 36_82

Moffitt T (1990), Juvenile delinquency and attention deficit disorder: Boys' developmental trajectories from age 3 to 15. Child Dev 61: 893_910

Moffitt T (1993), Adolescence_limited and life_course persistent antisocial behavior: A developmental taxonomy. Psychol Review 100: 674_701

Patterson GR, Reid JB, Dishion TJ (1992), Antisocial boys. Eugene, OR: Castalia

Orvaschel H, Puig_Antich J (1987), Schedule for Affective Disorders and Schizophrenia for School_Age Children: Epidemiologic 4^th Version. Ft Lauderdale, FL: Nova University, Center for Psychological Study

Robins LN (1966), Deviant children grow up. Baltimore: Williams & Wilkins

Reid J (1993), Prevention of conduct disorder before and after school entry: Relating interventions to developmental findings. Development and Psychopathology 5: 243_262

Sanson A, Oberklaid F, Pedlow R, Prior M (1991), Risk indicators: Assessment of infancy predictors of pre_school behavioural maladjustment. J Child Psychol Psychiatry 32: 609_626

Shaw DS, Gilliom M, Giovannelli J (2000), Aggressive behavior disorders. In Handbook of Infant Mental Health, Zeanah CH ed. (2^nd Edition), New York: Guilford Press, pp 397_411

Shaw DS, Winslow EB, Owens EB, Vondra JI, Cohn JF, Bell RQ (1998), The development of early externalizing problems among children from low_income families: A transformational perspective. J Abnorm Child Psychol 26: 95_107

Sroufe LA (1983), Infant_caregiver attachment and patterns of adaptation in preschool: The roots of maladaption and competence. In Minnesota symposium in child psychology (Vol. 16), Perlmutter M ed. (Vol. 16,), Hillsdale, NJ: Erlbaum, pp 41_81

Strauss MA (1991), Family violence in American families: Incidence rates, causes, and trends. In:Abused and battered: Social and legal responses of family violence, Knudsen DD, Miller JL eds. New York: Aldine de Gruyter, pp 17_34

Wesbster_Stratton C, Martin H (1994), Troubled families _ problem children: Working with parents: A collaborative process. Chichester, England: John Wiley and Sons