Running head: IN SEARCH OF GROWTH

Running head: CHANGES IN MATERNAL RATINGS

Changes in Maternal Ratings of Children’s Overt and Covert Antisocial Behavior

Gerald R. Patterson^a, Daniel S. Shaw^b, James J. Snyder^c, and Karen Yoerger^a

^aOregon Social Learning Center, Eugene, OR

^bUniversity of Pittsburgh, Pittsburgh, PA

^cWichita State University, Wichita, KS

Address correspondence to:

Gerald R. Patterson, Ph.D.

Oregon Social Learning Center

160 East Fourth Ave.

Eugene, OR 97401

541-485-2711

541-485-7087 FAX

carleenr@OSLC.org

Keywords: antisocial, growth, overt, covert, early onset

Changes in Maternal Ratings of Children’s Overt and Covert Antisocial Behavior

Abstract

Maternal ratings of overt and covert forms of aggression were collected for two samples of children ranging in age from 2 through 12 years. It was hypothesized that longitudinal analyses would show the slope scores for these two forms of aggression to be quite different from each other. The data were consistent with this hypothesis. An effort was made to find alternative explanations for the negative slope for overt antisocial behavior. Alternatively, it was hypothesized that the more extreme cases would not show this negative slope. It was also hypothesized that careful examination of intraindividual curves would identify a significant number of individuals growing from normal to clinical levels of overt antisocial behavior. The findings led to the rejection of both alternative hypotheses. It was hypothesized that overt and covert scores would correlate significantly for first grade boys. It was also assumed that both covert and overt scores would show moderate stability over the 5 year interval. The findings were consistent with both of these hypotheses.

Acknowledgments

An earlier version of this paper was presented at the annual conference of the American Society of Criminology in San Diego, November 1997. We gratefully acknowledge the support provided by Grant RO1 MH 38318 (Prevention and Behavioral Medicine Research Branch, Division of Epidemiology and Services Research, National Institute for Mental Health [NIMH], U.S. Public Health Service [PBS]) in collecting data and Grants MH 37940 (Center for Studies of Violent Behavior and Traumatic Stress, NIMH, U.S. PHS) and P50 MH 46690 (Prevention and Behavioral Medicine Research Branch, Division of Epidemiology and Services Research, NIMH, U.S. PHS) and MH 37940 in analyzing data and preparing the report. Grants supporting the Pittsburgh research were MH 46925 and MH 50907 (NIMH, U.S. PHS).

Changes in Maternal Ratings of Children’s Overt and Covert Antisocial Behavior

Most investigators agree that early arrest is a prime predictor for status as a chronic offender [Blumstein et al., 1986; Moffitt, 1993; Patterson et al., 1989.]. This perspective, in turn, emphasizes the importance of building models that identify the childhood antecedents that significantly predict early-onset arrest. In the coercion model, the developmental antecedent to early arrest is thought to be a sequence that begins during early childhood with overt forms of antisocial behavior [Patterson & Yoerger, 2002]. The second stage is characterized by the emergence of high rates of covert forms of antisocial behavior during later childhood and early adolescence [Patterson & Yoerger, 1997]. Presumably, boys who perform high rates of overt antisocial behaviors when entering school are at particular risk for performing at least low levels of covert forms as well [Patterson & Yoerger, 2002].

Several developmental models for children’s aggression emphasize the key role played by overt and covert forms of antisocial behaviors in predicting early arrest for juvenile crime [Loeber & Schmaling, 1985; Loeber et al., 1999; Patterson, 1982; Patterson & Yoerger, 1997; 1999; 2002]. According to the coercion model, it is the sequence of high rates of overt antisocial behavior during childhood followed by high rates of covert antisocial behavior during early adolescence that constitutes the key predictors for early-onset arrest [Patterson & Yoerger, 1997; 1999; 2002].

It seems reasonable, then, in building a theory of aggression to begin by describing the early growth in these two forms of antisocial behavior. Overt antisocial behaviors were defined by maternal ratings of behaviors (e.g., stubbornness, temper tantrums, and fighting) as they occurred during two longitudinal studies during the interval between ages 2 and 12 years. One of the longitudinal studies also collected data that described changes in covert behaviors, such as stealing, vandalism, and lying, for the interval from ages 6.5 through 12 years.

A prior examination of longitudinal ratings by mothers for at-risk adolescent boys showed a negative slope for overt forms but a positive slope for covert antisocial behaviors [Patterson & Yoerger, 1997]. In that study, rates for covert behaviors were very low at age 10 and remained low until about age 13 years at which time there was accelerated growth. This finding suggests that in the current study, for boys between ages 6 and 12, the mean level for covert behaviors would remain close to zero and display a neutral slope. However, findings from the research literature suggest the slope for overt forms of antisocial behavior would be negative for younger as well as for adolescent samples. For example, maternal diary reports in the classic study by Goodenough [1931] showed frequency of anger outbursts decreased from about .13 per hour at 18 months to .07 per hour by 8 years. Griffith [1952] used parental ratings of children, ages 6 through 14 years, and found decreases in angry outbursts. A similar effect was noted for data from both teachers [Achenbach, 1991a] and parents [Achenbach et al., 1991b; Tremblay et al., 1999]. Based on parent report, the effect was replicated in the large-scale study by Bongers et al. [2003]. In that study, aggressive behaviors decreased at a much faster rate for boys than for girls.

Alternatively, the effect may be some function of parent (and teacher) report data. However, the effect was also obtained in observation data collected in the home. For example, the study by Fawl [1963] showed a correlation of -.76 between age and frequency of conflict bouts. The negative slope seems to hold for both normal and clinical samples as shown in observation data collected in homes [Patterson, 1982]. Decreases in frequency are also cross cultural, as shown for observed age-related decreases in child assault behavior in six cultures studied by Whiting and Whiting [1975] and in the Isle of Wight study by Rutter et al. [1970]

J. Snyder has put forward the hypothesis that adults successfully apply discipline to the more obvious forms of overt antisocial behavior [Snyder et al., 2003]. As he puts it, the aversive contingencies “drive overt forms of antisocial behavior underground.” In the present context, the Snyder hypothesis may account for the consistent findings for a negative slope for overt forms.

It is well known that for older samples of children, the overt and covert forms of antisocial behavior are highly correlated. [Loeber & Schmaling, 1985]. Observation data collected on the playground for kindergarten and first grade boys showed that the overt and covert forms are also intercorrelated [Snyder et al., 2003] for younger children. It was hypothesized in the current sample of younger children that maternal ratings for overt and covert forms will also be significantly correlated. It was also assumed that both forms would show moderate stabilities in maternal ratings over the 5-year interval.

The findings suggest that the sequence of growth in overt behavior followed by growth in covert forms of antisocial behavior may serve as a necessary antecedent for early-onset delinquent behavior. Based on the Oregon Youth Study (OYS), the data showed that boys who scored high on overt antisocial behavior at age 10 years and who then showed the intense growth in covert forms during adolescence were most at risk for early arrest before the age of 14 years [Patterson & Yoerger, 1999, 2002]. Their risk for arrest as adults was .57 [Dishion & Patterson, in press]. There was a second group characterized by low scores on overt antisocial behavior but a marked acceleration in covert behaviors during midadolelscence. This group tended to be arrested after age 14 years (late onset) but as adults were most likely to be desistors [Patterson & Yoerger, 1997]. For this group, the likelihood of adult arrest was .27. The third group’s path was characterized by high scores at age 10 years on overt antisocial behavior and by a failure to move into peer deviancy training as evidenced by low scores on covert antisocial behavior. The likelihood of adult arrest for this group was .30. The Dunedin sample studied by Moffitt et al. [2002] also identified a group like this. The finding suggests that the study of the sequence of growth may contribute to our understanding of which adolescents will become early-onset and chronic offenders.

A series of analyses were carried out that examine alternatives to the idea that the negative slope for overt forms is generalizable. For example, it is possible that the decrease in overt forms of antisocial behavior may characterize some age groups but not others. In the current article, maternal ratings of overt antisocial behavior from two longitudinal studies span the interval from 2 through 12 years. These data will provide a test for the contribution of age to negative slope.

There are other alternative hypotheses that are also of theoretical interest. For example, the coercion model implied that the longer a child was engaged in a coercion process, the greater the increase in frequency of more severe forms of overt antisocial behaviors [Patterson, 1982]. Thus, it would be expected that longitudinal data for problem children would show steady increases in severe antisocial behaviors, such as hitting, temper tantrums, and physical attacks. It could be that only the more childish forms of overt antisocial behavior decrease as a function of age. For example, it could be that juvenile forms of overt antisocial acts (such as teasing, screaming, and showing off) may decline, but the more serious problem behaviors (such as hitting and fighting) may have a neutral slope or actually increase. In the present article, this hypothesis was tested by examining maternal ratings from age 2 to age 12 years for a subset of overt antisocial behaviors that included only the more severe forms, such as fighting, physically attacks people, temper tantrums, and cruelty to animals.

Another possibility is that only a tiny subsample of boys actually grows in antisocial behavior. As toddlers, they might show normal levels of overt antisocial behavior, and then during middle childhood steadily grow until eventually they become cases that would be clinically identifiable. Given that the N for such a group might be quite small, it might not significantly contribute to the means for the larger sample. To test this possibility, intraindividual growth curves for overt forms from the Oregon (LIFT) sample were examined to determine the extent to which some boys changed from normal to clinical levels.

Two longitudinal samples were used to test six hypotheses about growth in overt and covert forms of antisocial behavior during the period from ages 2 to 12 years. The first hypothesis was that maternal ratings for overt antisocial behavior would be described by a significant negative slope. The second hypothesis tested was that growth in the most extreme forms of overt antisocial behavior would be characterized by a neutral or a positive slope. Third, it was also hypothesized that there would be a significant number of individuals growing from a normal range in antisocial behavior during earlier developmental periods to attain clinical status in middle childhood. Fourth, it was proposed to examine maternal ratings of covert antisocial behavior to determine whether there was significant positive growth in that form during middle childhood. Fifth, it was hypothesized that at school entry, high levels of overt antisocial behavior would be a significant predictor for early forms of covert antisocial behavior. Sixth, it was assumed that maternal ratings for both forms would be moderately stable over the 5-year interval.

Methods

Samples

The younger sample was drawn from the Pittsburgh Mother and Child Project (PMCP). The sample consisted of 310 families with children recruited from waiting rooms of Women, Infants, and Children’s (WIC) Nutritional Supplement Program [Shaw et al., 1999]. When recruited at WIC clinics, the children were between 6 and 17 months old. When the infants were 18 months old and admitted to the current study, the mothers ranged in age from 17 to 43 years. Ethnic groupings of participants were 54% Caucasian, 40% African‑American, and 6% other. At the 18-month visit, 64% of the parents were either married or living together. The mean per capita family income was $2,892 per year ($11,616 for a family of four), and the mean Hollingshead [1975] socioeconomic status (SES) score was 24.5, indicative of a working-class sample. Of the 310 families seen at the 18-month assessment, data were available on 279 at age 6 years (an attrition rate of 10%). In the present study, maternal reports of child behavior were collected when the children were 2.0, 3.5, 5.0, and 6.0 years of age. After listwise deletion, 208 cases were available for analysis.

The Oregon LIFT sample was drawn from families with children in the first grade of 12 schools that had been identified earlier as the highest crime neighborhoods of a moderate-size metropolitan area [Reid, 1993; Reid & Eddy, 1997; Reid et al., 1999]. Of the families contacted, 12% declined to participate in the study. Of those families participating, 89% were Caucasian and 20% received public assistance. The majority of the families earned $15,000 to $50,000 per year. About 54% of the children lived with both biological parents, 20% of the families were headed by a single parent, and roughly another 20% of the families included a stepparent. In the majority of the families, the parents ranged in age between 25 and 50 years.

The resulting sample included 144 first grade boys who were intensively assessed at approximately 6.5, 8.0, 9.0, 10.0, 11.0, and 12.0 years of age. After listwise deletion, data for 101 first graders were available for analysis.

Measures

The boys’ overt and covert antisocial behaviors were assessed using items derived from the Child Behavior Checklist (CBC-L) completed by mothers at each developmental point. Each item was scored as 0 (not true), 1 (sometimes or somewhat true), or 2 (very true or often true) in reference to the child behavior described by the item. Unlike observational data, these maternal ratings do not solely reflect the frequency of the behavior, but in some general sense do reflect the prevalence of the behavior (some composite of perceived behavioral frequency, social impact or intrusiveness, and warranted concern relative to the mothers’ own “normative” base for same‑aged children).

Overt antisocial behavior. Scores were based on mothers’ ratings on two versions of CBC-L at ages 2 to 3.5 years and at ages 5 to 18 years [Achenbach, 1991a; Achenbach et al., 1991b]. Based on previous factor analyses [Loeber & Schmaling, 1985; Patterson et al., 1992], a set of 21 items was selected that assessed overt antisocial behavior. For each item, Patterson and Yoerger [1997] examined linear and quadratic trends describing change over time. A principle axis factor analysis was carried out separately for linear and quadratic terms. Items that loaded less than .30 on either the linear or quadratic factor scores were dropped.

Nine items sampling overt antisocial behaviors were selected that were shared in common by the two versions of the CBC‑L. The 9 items were: (a) doesn’t get along well with other children; (b) gets into many fights; (c) physically attacks people; (d) screams a lot; (e) showing off or clowning (at 2 to 3.5 years, shows off); (f) stubborn, sullen, or irritable; (g) sudden changes in mood or feelings; (h) sulks a lot, and (i) temper tantrums or hot temper. Cronbach’s alphas for the 9-item scales ranged from .69 to .86 across all time points. The mean of the 9 items formed a composite overt antisocial behavior score.

Severe overt antisocial behavior. Subscales of the most severe examples of antisocial behavior were formed at each time point by maternal ratings on three of the items: gets into many fights; physically attacks people; and temper tantrums or hot temper. Cronbach’s alphas were somewhat lower (.42 to .70) for this subscale of low base‑rate items. The mean of these three items formed the severe antisocial behavior score.

Covert antisocial behavior. Loeber and Schmaling [1985] examined self-report data for a sample of 195 preadolescent and adolescent boys to form a measure of covert antisocial behavior. They selected 11 behaviors to form their scale that were also identifiable as items in the CBC-L [Achenbach et al., 1991b] filled out by the mothers. The items included: cruelty to animals; socializes with kids who get in trouble; lying or cheating; runs away from home; sets fires; steals at home; steals outside the home; swearing or obscene language; truancy or skips school; uses alcohol or drugs, and vandalism. Cronbach’s alphas for these scales ranged from .43 to .79. The covert items were not included in the assessment of the younger ages of the PMCP, so the sample was not included in this particular analysis.

Mother and teacher scales for prediction study. In order to maximize predictive power, expanded sets of items were taken from the mother and teacher versions of the CBC-L [Achenbach & Edelbrock, 1979] of the Oregon (LIFT) sample at Grades 1 and 6. These items defined the child overt and covert antisocial behavior scales for the prediction study.

For mothers, the overt antisocial scale added 10 items to the 9 previously described. The 10 additional items included: (a) argues a lot; (b) bragging, boasting; (c) cruelty, bullying, or meanness to others; (d) demands a lot of attention; (e) doesn’t seem to feel guilty after misbehaving; (f) easily jealous; (g) impulsive or acts without thinking; (h) not liked by other children; (i) teases a lot; and (j) threatens people. Cronbach’s alphas for the 19-item scales at Grades 1 and 6 were .90 and .92, respectively.

The teacher overt antisocial scale consisted of 23 items, including: (a) argues a lot; (b) defiant, talks back to staff; (c) bragging, boasting; (d) cruelty, bullying, or meanness to others; (e) demands a lot of attention; (f) disobedient at school; (g) doesn’t get along with other pupils; (h) doesn’t seem to feel guilty after misbehaving; (i) easily jealous; (j) gets in many fights; (k) impulsive or acts without thinking; (l) not liked by other pupils; (m) physically attacks people; (n) disrupts class discipline; (o) screams a lot; (p) showing off or clowning; (q) stubborn, sullen, or irritable; (r) sudden changes in mood or feelings; (s) sulks a lot; (t) talks too much; (u) teases a lot; (v) temper tantrums or hot temper; and (w) threatens people. Cronbach’s alpha was .95 for Grade 1 and .93 for Grade 6.

The covert scale for mothers was the same as previously described. The covert scale for teachers consisted of 10 items: (a) destroys his own things; (b) destroys property belonging to others; (c) hangs around with others who get in trouble; (d) lying or cheating; (e) behaves irresponsibly; (f) steals; (g) swearing or obscene language; (h) tardy to school or class; (i) truancy or unexplained absence; and (j) uses alcohol or drugs. Cronbach’s alphas for Grades 1 and 6 were .81 and .80, respectively. Composite scores for overt and covert antisocial behavior were computed as the means of the standardized mother and teacher scales.

Analyses

Three types of analyses were performed. First, at the item level, similar patterns of change over time among 9 items for overt and 11 items for covert scales were confirmed within each sample by reliability analyses of linear trend scores. The writers were concerned about the possibility that some items in the scales might be less sensitive as measures of change than others. To examine this possibility, linear trend scores were calculated for each of the items in the overt and covert scales using orthogonal coefficients appropriate to the spacing of the time points in each sample. Reliability analyses of those scores demonstrated similarities among the items in patterns of change over time. The majority of the linear trend score means for the 9 items in the overt scales were negative. Cronbach’s alphas for these scales of overt linear trends were .74 (PMCP) and .57 (LIFT). In contrast, the majority of linear trend scores for the covert items were positive; and Cronbach’s alpha for the LIFT scale of covert linear trends was .47.

Second, scores for maternal ratings of overt antisocial behavior were plotted separately for each boy from the LIFT sample. The mean for the sample at Grade l was .34 with a standard deviation of .34. Growth curves were examined separately to identify boys who experienced significant growth during the interval between Grades 1 and Grade 6. Clinical growth was defined by those boys scored at or below the mean value when first tested at Grade l and then scored at or above l SD above the mean for two or more subsequent points in time.

Third, for each sample, repeated measures MANOVA was used to test the significance of linear effects for overt, severe overt, and covert composite scores. To test the difference in linear effect between the extreme group and the remainder of the sample, a second set of repeated measures MANOVAS was conducted on the severe overt composite scores.

It might be noted that an effort was made to fit a growth model to these data in order to obtain a more precise estimate of the early contribution of overt to growth in covert forms. We found that there was simply not enough between-subject variation in growth in covert forms to provide a basis for such an analysis. In fact the within-subject variation across time in covert forms exceeded the between-subject variance. The findings are not detailed here.

Results

Figure 1 describes changes in maternal ratings of overt antisocial behavior as a function of the child’s age. In keeping with findings from earlier studies, data for each longitudinal sample describe a clearly defined negative slope. In both developmental ranges, the data showed that mothers’ ratings of overt antisocial behaviors decreased as a function of age. On the average, mothers of 2‑year‑old children reported a score of .55 for overt antisocial items, and mothers of 12 year olds reported a mean of .26. The findings suggest a steady decrease in the maternal composite reports of overt antisocial behavior, approximately .029 item level decrease per year. Notice that at age 6 years, the Oregon and Pittsburgh samples provide very different estimates for overt forms of antisocial behavior. It is assumed that the difference reflects a greater prevalence of families with disrupted parenting practices in the Pittsburgh sample. In that measures of parenting practices employed in the two studies were not comparable, this hypothesis could not be tested at the present time.

_____________________________

Insert Figure l and Table I about here

_____________________________

A repeated-measures MANOVA was applied separately to data from each sample to examine the significance of the linear effects. Findings summarized in Table I show that changes in mean scores for overt antisocial behavior were significant for both samples. The data also showed that negative linear coefficients were significant for each of the samples. Estimates of quadratic and cubic effects were also made; none were significant and are not included here.

_____________________

Insert Figure 2 about here

_____________________

Figure 2 shows the item means for maternal ratings of the severe overt antisocial behavior. The data were used to test the hypothesis that slope for an overt scale based just on the more severe items would be neutral or positive rather than negative. It can be seen that the findings do not support the hypothesis. Findings for repeated-measures analyses are summarized in Table II. Again, each sample of maternal ratings for items describing more severe overt antisocial behavior showed a significant negative linear effect. The findings are consistent with the hypothesis that a negative slope for overt forms is a general case rather than being limited to items that only describe more trivial forms of overt antisocial behavior.

____________________

Insert Table II about here

____________________

We also examined the findings for a subset of extreme children identified as 1 SD above the mean on the overt scale at Grade 1. Examining their changes in scores for the subset of extreme items showed the now familiar general negative slope. The findings are not detailed here. Taken together, the findings are consistent with the expectation that maternal ratings of overt antisocial behavior may, as a general case, be expected to conform to a negative slope.

It had been hypothesized that during the middle childhood interval, growth in overt antisocial behaviors might characterize only a small group of children identified as performing within normal levels at Grade 1 but moving toward, and maintaining, significant levels of growth in overt antisocial behavior over the ensuing 5 years. Examination of data from the LIFT sample [Patterson & DeGarmo, 1997] showed that only two boys identified as being at the normal level in Grade l moved into the clinical range on maternal ratings of overt antisocial behavior during the ensuing years. According to these findings, only a very small number (.015) of boys can be expected to show clinically significant growth in overt antisocial behavior during the quiescent interval. This is in keeping with findings from the Montreal study showing that after 6 years of age, teacher ratings showed that there were no individuals who initiated and maintained moderate to high levels of physical aggression [Nagin & Tremblay, 1999].

Maternal ratings for the covert antisocial behavior scale were available only for the Oregon LIFT sample. We wished to determine whether maternal ratings for covert forms would show a neutral or a positive slope for the interval from Grades 1 through 6. The findings are summarized in Figure 3. The data showed a nonsignificant positive slope. Although positive (.011), the linear coefficient was minimal and nonsignificant. (t =1.245, p = .216).

______________________

Insert Figure 3 about here

______________________

It was hypothesized that in disrupted families, growth in overt antisocial behavior begins in the preschool years and is accompanied by very low rates of covert antisocial behavior [Patterson, 1982]. It was hypothesized that composites based on maternal and teacher ratings at Grade 1 for both the overt and the covert forms would be significantly correlated. Consistent with the hypothesis, the correlation between the two composite scores was .72 (p = .000). The data also showed modest levels of stability from Grades 1 through 6 for both the overt (.60, p = .000) and covert (.48; p = .000) scales.

Discussion

Analyses of maternal ratings showed strong support for the hypothesis that the age-related decline in overt antisocial behavior may represent a general case. The findings were consistent across two samples that covered ages 2 through 12 years. The literature review showed similar findings across teacher and parent-report data and across observation-based studies as well. The fact that findings for the more severe forms of overt antisocial behavior also followed a negative slope provides further support for the general case hypothesis. The failure to find evidence for new cases being added during middle childhood further underscores the importance of the preschool interval in the development of conduct disorders. The narrow focus in the present study on maternal ratings emphasizes the need to systematically sample data across agents and methods to determine the precise shape for growth in both overt and covert as it varies by age and by gender. The current article is but a small step in this direction.

The fact that there is a negative slope across measures of overt antisocial behavior demands an explanation. As noted earlier, a social interactional perspective would suggest that the slope could reflect age-related changes in reactions from adults and peers to overt forms [Snyder et al., 2003]. As the child matures physically, unpredictable explosions of coercive behaviors become increasingly unacceptable, particularly to adults. For example, observation data showed that the 10-year-old aggressive problem children performed about .70 coercive responses per minute [Patterson, 1982]. This rate of coercion is roughly equivalent to the rates for a normal preschool child. As the child ages, he learns that there are fewer and fewer settings were overt forms will be reinforced. The child becomes increasingly selective and generally decreases performance in the presence of adults. According to Snyder’s underground hypothesis, the child may shift from overt to more covert manifestations. This is consistent with the present finding that early levels of overt antisocial behavior were reliably correlated with higher levels of covert antisocial behavior.

The available evidence suggests that there may be an important developmental sequence that begins during the preschool years with the reinforcement by family members for overt forms of antisocial behavior. Our studies showed that high rates of overt behavior are often accompanied by covert behavior such as stealing and fire setting. Typically the latter occur at much lower rates than do the overt forms. As shown in the present report, the rate differential is maintained throughout the childhood interval. Prior to age 14, the antisocial child becomes increasingly at risk for extensive involvement with members of the deviant peer group. This is accompanied by dramatic growth in new forms of covert antisocial behavior [Patterson et al., 2000]. In effect, the contingencies for the first stage in the sequence are provided by the family during preschool years, and the contingencies supporting the second stage are provided during early adolescence by deviant peers. Notice, too, that during adolescence, the slopes for overt and covert forms became negative and positive, respectively.

The survey study by Bongers et al. [2003; p. 190] has a very different view about the relation between early forms of overt antisocial behavior and adolescent and adult adjustment. They see aggressive behaviors of children and adolescents as being relatively “transitory in nature and resolved by the beginning of adulthood.” The perspective is evidently based on the fact that the aggressivity scores at age 4 were twice as high as they were at age 18 years.

However, some contemporary developmental models of delinquency take a very different position. For example, Loeber et al. [1997] take the position that the covert and overt forms of antisocial behavior in early childhood represent distinctly different paths to delinquency. In effect, childhood forms of aggression may be changing over time; but they are predictive of adolescent and adult crime. The models described in the present report represent a similar position. These correlational models suggest that reducing childhood levels of antisocial behavior should lead to less risk for adolescent delinquency. Recent randomized trial prevention designs have shown that procedures that reduce antisocial behavior during childhood demonstrate reductions in delinquency when followed up years later [Vitaro, 200l; DeGarmo & Forgatch, 2000]

From the mother’s perspective, there is simply little growth in covert forms during the middle childhood interval. The linear coefficient for these ratings was only slightly positive and nonsignificant. Although teacher ratings of covert behaviors show higher rates of activity than do mothers’ ratings [Patterson & Yoerger, 2002], their data concur in offering little evidence for growth during this interval. It is a matter of some interest, therefore, to find data showing that what little covert activity there is can be predicted from knowing the general level of overt activity at Grade 1. The hypotheses that need to be tested are relatively straightforward. It would be hypothesized that the relative rate of reinforcement for deviancy is relatively stable across the middle childhood interval. Second, it would be hypothesized that the time allotted to deviant peer involvement is limited but relatively stable during this same interval. Based on the literature reviewed earlier, the amount of time spent with deviant peers would be expected to increase dramatically during early adolescence. It is instructive to note that during adolescence, the availability of deviant peers in the social experience of most adolescents shows massive increases in contacts during the same intervals for peak delinquency [Elliott & Menard, 1996].

Progress in understanding development from early childhood through adolescence requires that we begin by accurately describing the growth in the various forms of antisocial behavior. The next stage will be to explain how these changes come about. Studies of this kind will eventually move the field away from a simplistic trait to a more developmental perspective.

References

Achenbach TM. 1991a. Manual for the Teacher's Report Form and 1991 Profile. Burlington, VT: University of Vermont.

Achenbach TM, Edelbrock CS. 1979. Child behavior checklist. Bethesda, MD: National Institute of Mental Health.

Achenbach TM, Howell CT, Quay HC, Conners CK. 1991b. National survey of problems and competencies among four- to sixteen-year-olds: Parents' reports for normative and clinical samples. Monographs of the Society for Research in Child Development 57:3, Serial No. 231.

Blumstein A, Cohen J, Roth JA, Visher CA, editors. 1986. Criminal careers and career criminals, vols. I & II. Washington, DC: National Academy Press.

Bongers IL, Koot HM, van der Ende J, Verhulst FC. (2003). The normative development of child and adolescent problem behavior. Journal of Abnormal Psychology 112:179-192.

DeGarmo, D. S., & Forgatch, M. S. (2000, November). Preventing the "early start" within transitional divorce families: An experimental test of precursors influencing antisocial behavior and delinquency. Paper presented at the American Society of Criminology Annual Meeting: Crime and Criminology in the Year 2000, San Francisco.

Dishion TJ, Patterson GR. in press. The development and ecology of antisocial behavior. In D Cicchetti, D Cohen editors. Developmental psychopathology. Vol. 3: Risk, disorder, and adaptation (Revised ed.). New York: Wiley & Sons.

Elliott DS, Menard S. 1996. Delinquent friends and delinquent behavior: Temporal and developmental patterns. In Hawkins D, editor. Delinquency and crime: Current theories Cambridge: Cambridge University Press. P. 28-67.

Fawl CI. 1963. Disturbances experienced by children in their natural habitats. In Barker RG, editor. The stream of behavior New York: Appleton-Century-Crofts. p. 99-126.

Goodnough FL. 1931. Anger in young children. Minneapolis: University of Minnesota.

Griffith W. 1952. Behavior difficulties of children as perceived and judged by parents, teachers, and children themselves. Minneapolis: University of Minnesota Press.

Hollingshead AB. 1975. Four factor index of social status (Unpublished manuscript). Department of Sociology, Yale University, New Haven, CT.

Loeber R, Keenan K, Zhang Q. 1997. Boys' experimentation and persistence in developmental pathways toward serious delinquency. Journal of Child and Family Studies 6:321-357.

Loeber R, Schmaling KB. 1985. Empirical evidence for overt and covert patterns of antisocial conduct problems: A meta-analysis. Journal of Abnormal Child Psychology 13:337-352.

Loeber R, Wei E, Stouthamer-Loeber M, Huizinga, D, Thornberry, T. 1999. Behavioral antecedents to serious and violent juvenile offending: Joint analyses from the Denver Youth Survey, Pittsburgh Youth Study, and the Rochester Youth Development Study. Studies in Crime and Crime Prevention 8:245-263.

Moffitt TE. 1993. Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review 100:674-701.

Moffitt TE, Caspi A, Harrington H, Milne BJ. 2002. Males on the life-course persistent and adolescence-limited antisocial pathways: Follow-up at age 26. Development and Psychopathology 14:179-207.

Nagin D, Tremblay RE. 1999. Trajectories of boys' physical aggression, opposition, and hyperactivity on the path to physically violent and nonviolent juvenile delinquency. Child Development 70:1181-1196.

Patterson GR. 1982. A social learning approach: Vol. 3: Coercive family process. Eugene, OR: Castalia.

Patterson GR. 1999. A proposal relating a theory of delinquency to societal rates of juvenile crime: Putting Humpty Dumpty together again. In Cox MJ, Brooks-Gunn J, editors. Conflict and cohesion in families: Causes and consequences. Mahwah, NJ: Lawrence Erlbaum. p. 11-35.

Patterson GR, DeBaryshe BD, Ramsey E. 1989. A developmental perspective on antisocial behavior. American Psychologist 44:329-335.

Patterson GR, DeGarmo DS. 1997, November. In search of growth in antisocial behavior: Prelude to early-onset delinquency. Paper presented at the American Society of Criminology, San Diego, CA.

Patterson GR, Dishion TJ, Yoerger K. 2000. Adolescent growth in new forms of problem behavior: Macro- and micro-peer dynamics. Prevention Science 1:3-13.

Patterson GR, Reid JB, Dishion TJ. 1992. A social interactional approach: vol. 4: Antisocial boys. Eugene, OR: Castalia.

Patterson G R, Stouthamer-Loeber M. 1984. The correlation of family management practices and delinquency. Child Development 55:1299-1307.

Patterson GR, Yoerger K. 1997. A developmental model for late-onset delinquency. In D. W. Osgood, editor. Motivation and delinquency: Nebraska Symposium on Motivation, vol. 44. Lincoln: University of Nebraska Press. p. 119-177.

Patterson GR, Yoerger K. 1999. Intraindividual growth in covert antisocial behavior: A necessary precursor to chronic and adult arrests? Criminal Behaviour and Mental Health 9:86-100.

Patterson GR, Yoerger K. 2002. A developmental model for early- and late-onset delinquency. In Reid JB, Patterson GR, Snyder J, editors. Antisocial behavior in children and adolescents: A developmental analysis and model for intervention. Washington, DC: American Psychological Association. p. 147-172.

Reid JB. 1993. Prevention of conduct disorders before and after school entry: Relating interventions to development findings. Journal of Development and Psychopathology 5:243-262.

Reid JB, Eddy JM. 1997. The prevention of antisocial behavior: Some considerations in the search for effective interventions. In Stroff DM, Breiling J, Maser JD, editors. Handbook of antisocial behavior. New York: John Wiley & Sons. p. 343-356.

Reid JB, Eddy JM, Fetrow RA, Stoolmiller M. 1999. Description and immediate impacts of a preventive intervention for conduct problems. American Journal of Community Psychology 27:483-517.

Rutter M, Tizard J, Whitmore K. 1970. Education, health, and behavior. London: Longmans.

Shaw DS, Winslow EB, Flanagan C. 1999. A prospective study of the effects of marital status and family relations on young children's adjustment among African American and European American families. Child Development 70:742-755.

Snyder JJ, Reid JB, Patterson GR. 2003. A social learning model of child and adolescent antisocial behavior. In BB Lahey, TE Moffitt A Caspi editors, The causes of conduct disorder and juvenile delinquency New York: Guilford Press. p. 27-48.

Tremblay RE, Japel C, Perusse D, Boivin M, Zoccolillo M, Montplaisir J, McDuff P. 1999. The search for the age of "onset" of physical aggression: Rousseau and Bandura revisited. Criminal Behaviour and Mental Health 9:8-23.

Vitaro F, Brendgen M, Tremblay RE. 2001. Preventive intervention: Assessing its effects on the trajectories of delinquency and testing for mediational processes. Applied Developmental Science 5:201-213.

Whiting BB, Whiting JM. 1975. Children of six cultures. Cambridge, MA: Harvard University Press.