Evidence-Based Preventive Intervention for Preadolescent Aggressive Children: One-year Outcomes Following Randomization to Group versus Individual Delivery

Abstract

Objective

Some research suggest that group interventions with antisocial youth may on occasion have iatrogenic effects. This is the first study to test the effects of group versus individual delivery of evidence-based intervention for aggressive children.

Method

Three hundred sixty fourth-grade children were randomly assigned by school to Group Coping Power (GCP) or Individual Coping Power (ICP). Longitudinal assessments of teacher, and parent reports of behavior (BASC; PASA) were collected from baseline through a one-year follow-up.

Results

Growth curve analyses revealed children in both conditions reduced teacher and parent reported externalizing behavior problems and internalizing behavior problems by the end of the one year follow up. However, the degree of improvement in teacher-reported outcomes was significantly greater for children receiving an individual version of the program. In addition, children’s baseline level of inhibitory control moderated intervention effects, showing children with low initial levels of inhibitory control to respond poorly in teacher-rated outcomes to group interventions compared to those delivered individually.

Conclusions

This study suggests overall benefits to children for either group or individual delivery of the Coping Power program under high fidelity conditions, however, for children with low levels of initial self-regulation, individualized interventions will likely yield the most significant reduction in externalizing behavior in the school setting in preadolescence.

Translational research over the past three decades has resulted in the development, testing and refinement of effective prevention and treatment of antisocial behavior. Despite the success of these structured research-based interventions, some of which are delivered in a group formats, it is critical to respond to emerging scientific findings that might suggest intervention refinement. One such finding is that interventions that aggregate high-risk children into groups are potentially iatrogenic (Dodge, Dishion, & Lansford, 2006). As a result of deviant peer effects, group interventions may escalate or maintain, rather than reduce, youth behavior problems. Developmental research has indicated that children with problem behavior are likely to affiliate with each other, and that involvement with deviant peers leads to increased risk for adolescent problem behaviors. Intervention research has raised some similar concerns, although meta-analyses have not found across-the-board evidence for iatrogenic effects (Weiss et al, 2005). In a review of the literature on peer contagion in general and iatrogenic research in particular, the evidence suggested was sufficiently worrisome to lead some researchers to suggest caution when high-risk youth in clinical, educational, or correctional settings are aggregated for treatment (Dodge, Dishion & Lansford, 2006; Dishion & Tipord, 2011).

Although there are concerns about deviant peer effects in interventions for aggressive and conduct problem children, there are also may be important potential benefits to using a highly structured group format for intervention delivery (Poulin et al, 2001). The benefits include: a) potential cost-benefits of group intervention, b) more natural practice of learned skills through role-playing, c) receipt of peer reinforcement for attaining behavioral goals, and d) creation of a group norm that using non-aggressive strategies is useful.

One of the several research-based, structured group-based intervention programs for children with antisocial behavior is the Coping Power program which targets preadolescent children just prior to the additional risks they will experience during the adolescent years. Research trials have shown that randomization to the Coping Power intervention results in improvements, relative to control groups, through one-year follow-ups in substance use, delinquent behavior, and teacher-rated behavioral problems, that these intervention effects are mediated by changes in children’s social cognitions and parents’ use of consistent discipline, and that real-world school counselors could be trained to provide the intervention that affected children’s externalizing problems according to teacher, parent and self-ratings at the end of intervention (Lochman et al., 2012a) compared to a control group. The intervention has helped to prevent the declines seen in control children in language arts grades in the two years following intervention (Lochman et al., 2012b). Long-term follow-up effects, three to four years after the intervention, have been found on youths’ externalizing behavior in school settings in two separate studies (Lochman et al., 2013, 2014) and on substance use in a clinic setting (Zonnevylle-Bender et al., 2007). Because of the existing evidence-base for group-administered Coping Power, it provides a unique opportunity to compare the effects of group vs. individual formats with implications for structured research-based interventions for preadolescent children. Even though there are no overall negative effects of the Coping Power program, and in fact there are significant prevention effects, deviant peer group effects may be minimizing the strength of the intervention’s potential effects.

In the review of potential individual moderators of group intervention effects on high risk children, it has been suggested that children’s poor self-regulation might augment peer contagion effects (Dishion & Tipsord, 2011). Inhibitory control is a key measure of children’s temperamental trait to be cautious and controlling of one’s personal behavior. Poor inhibition is uniquely prognostic of children’s growth in problem behaviors (Pardini et al., 2004; Young et al., 2009) as well as vulnerability to peer influence (Dishion & Tipsord, 2011).

The Current Study

The current study uses the Coping Power child component as the common evidence-based intervention to compare group (GCP) versus individual (ICP) delivery, and has implications for the effects of group formats for research-based interventions for preadolescent aggressive children. It was hypothesized that ICP would produce greater reductions in externalizing behavior outcomes at a one-year follow-up in comparison to GCP. It was also hypothesized that youth with low inhibitory control would be most vulnerable to deviancy effects in the group intervention (GCP) and therefore would show smaller effect sizes than youth higher in inhibitory control at baseline, in comparison to similar children in the ICP condition.

Method

Participants

Fourth grade teachers from 20 schools completed ratings on all students in their classrooms. At the first screening gate, children whose scores fell at or above the cut-off score set at the 25th percentile were considered eligible for participation. Six students were recruited from each school for each annual cohort. At the second screening gate children whose parents rated them within the average range or above on the BASC Aggression scale remained eligible for enrollment. The three annual cohorts resulted in a total sample size of 360 parent-child pairs¹ (see Table 1 for sample information). Random assignment to condition was made at the school level. Schools were paired based on size, ethnic distribution, and eligibility for free and reduced meals. One school from each pair was randomly assigned to each condition, resulting in 10 schools in the Individual Coping Power (ICP) condition, and 10 schools in the Group Coping Power (GCP) condition. At GCP schools, each intervention unit of six children participated in a Coping Power group together². Overall, students participated in an average of 28.75 sessions (range = 0 to 34); GCP students participated in an average of 28.54 GCP sessions (range = 0 to 34), while the mean number of sessions attended for ICP students was 28.96 (range = 3 to 34)³.

Table 1.

Sample Characteristics at Baseline

Age
	Mean	10.17
	Range	9.17–11.79
Sex
	Boys	65%
	Girls	35%
Race and Ethnicity
	African American	78.1%
	Caucasian	20.3%
	Hispanic	1.4%
	Other	0.3%
Family Income
	none	5.0%
	< $15,000	24.9%
	$15,000 to $29,999	31.8%
	$30,000 to $49,999	20.5%
	>$50,000	17.6%

Procedure

Preintervention (Time 1) measures were completed with children and parents at the time of enrollment, during the spring semester of students’ fourth grade year. The Coping Power intervention was delivered during the end of fourth grade and throughout fifth grade. Mid-intervention assessments (Time 2) occurred in the summer after fourth grade, post-intervention assessments (Time 3) occurred in the summer after fifth grade, and one-year follow up assessments (Time 4) took place during the summer after students completed sixth grade. Children and parents were interviewed separately, typically in their homes, by research staff who were blind to the children’s condition assignment. At school, baseline data (Time 1) were gathered from teachers during spring of fourth grade, post-intervention assessments were collected in the late spring of fifth grade (Time 3), and one-year follow-up data (Time 4) were collected in the spring of sixth grade. The study was approved by the University’s IRB.

Coping Power Intervention

Since this study focuses on the effects of the delivery format of the Coping Power child intervention, only 32 sessions of the Coping Power child component (Lochman, Wells, & Lenhart, 2008) was offered, not the parent component. Children in Group Coping Power (GCP) participated in small groups of 5–6 children, with two co-leaders, and group sessions were 50–60 minutes. Children in GCP were exposed to all of the same intervention content as children in ICP, as well as some additional group-level activities (e.g., role-plays with peers, generating a group name or cheer, opportunity to earn group-level points and rewards). In addition to the group sessions, children in GCP also received the brief monthly individual sessions typically included in the Coping Power intervention. GCP leaders did not receive explicit training about deviancy training in groups, but they did receive routine training in setting and enforcing group behavior rules. Children in Individual Coping Power (ICP) met with a Coping Power leader individually for each of the 32 planned 30-minute sessions. Children in ICP completed role-plays and other interactive activities with their Coping Power leader, rather than with their peers.

Intervention Fidelity and Quality

Each leader served a similar number of GCP and ICP participants. To ensure high fidelity of implementation, two doctoral level psychologists who had substantial experience implementing Coping Power met with the interventionists weekly to monitor and provide feedback on program implementation. The interventionists also received detailed supervisory feedback on video-recorded GCP and ICP sessions on a monthly basis to ensure that program implementation remained consistent. GCP leaders and ICP leaders rated that they completely or partially completed 91.07% and 86.43% of objectives, respectively.

Measures

To identify at-risk aggressive students, fourth grade teachers completed the Teacher Report of Reactive and Proactive Aggression (Dodge, Lochman, Harnish, Bates, & Pettit, 1997) for all of the children in their class.

The study used the Externalizing and Internalizing scores from teachers’ and parents’ ratings on the Behavior Assessment System for Children (BASC; Reynolds & Kamphaus, 1992), which has demonstrated strong reliability (Cronbach’s alpha of .80–.89) and construct validity. The Peer Affiliation and Social Acceptance Measure (PASA; Dishion, Kim, Stormshak, & Nelson, 2014) was also completed by parents and teachers to assess for children’s deviant peer relations. Internal consistency is adequate for mother (α=.73) and teacher (α=.80) report and evidence supports the PASA’s convergent, concurrent, and predictive validity.

The Early Adolescent Temperament Questionnaire – Parent Report (EATQ-P; Rothbart et al., 2003) assessed the child characteristic of inhibitory control. Internal consistency for this subscale was good in prior Coping Power samples (α =.78; Pardini et al., 2004).

Analytic Strategy

The growth curve model included three levels: (a) times of measurement, (b) nested within children, and (c) nested within the intervention units (six children per cohort per school). HLM 6.02 software was used to perform the data analyses with full maximum likelihood (FML) estimation method (Raudenbush & Bryk, 2002). The unconditional curvilinear growth models were tested by adding a time’s quadratic term to the level-1 model, and the Deviance Tests indicated that the six outcomes changed over time with a significant curvature pattern, thus requiring a three-level curvilinear growth model. For teacher outcomes, time was coded as 0 as baseline, 1 as post-intervention, and 2 as 1-year-follow up, while for parent outcomes (where data collection took several months for each wave), we used the actual time interval from baseline as the time variable, with baseline set to zero. Each of the growth parameters in level-1 model has a substantive meaning. The intercept π_0ij (initial status at baseline), time slope π_1ij (the linear change rate over time), and quadratic term π_2ij (curvilinear change across time, captures the curvature or acceleration in each growth trajectory) were estimated in level-1 model equation (1)⁴ where Y_tij is the outcome score at time t for child i in intervention unit j. At level-2, the person level, we examined a child characteristic child’s inhibitory control (X_ij), that moderated rate of change, which was group centered. The intercept and time slope were treated as random effects at level-2. The quadratic term was treated as a fixed effect for teacher outcomes with three times of measurements, and as a random effect for parent outcomes at level-2 because of four times of measurements. The level-2 model equations were (2), (3), and (4) for parent outcomes and equations (2), (3) and (5) for teacher outcomes⁴. The intervention condition (W_j, ICP=1 and GCP=0) was included at Level-3. The intercept and time slope were random effect at Level-3, and all interaction effects were fixed effects. The Level-3 model equations were (6) to (10)⁴, where u_00j, and u_10j were the variance of population intercept and growth rate associated with intervention units. In this model, the variation in the growth parameters was partitioned as follows: (a) the variation among children within intervention unit was captured in the level-2 model, and (b) the variation among intervention units is represented in the level-3 model. To calculate the effect size for ICP and GCP on the growth rate, we constructed a 2-level growth curve model to the six child’s behavior outcomes with only intervention condition in the model. We used the Raudenbush and Liu (2001) formula for estimation standardized effect size on linear growth rate:

$$\delta =\frac{\mid {\beta }_{11}\mid }{\sqrt{{\tau }_{11}}}$$

where β₁₁ is intervention condition difference on the linear change rate, $\sqrt{{\tau }_{11}}$ is the population variation in annual linear change rate and δ is the standardized mean difference between ICP and GCP on the annual change rate.

Results

Table 2 provides the means and standard deviations for the six outcome variables from the BASC and PASA measures by intervention condition at each of the four time points. There were no significant differences between the attrition rates for the two conditions at any of the four time points for any of the six outcome measures. Attrition bias was tested by examining whether children’s characteristics (gender; age; racial status; initial level of externalizing behavior in school; inhibitory control) and intervention condition status differentiated attriters from non-attriters in logistic analyses. Higher levels of baseline teacher-rated externalizing behaviors predicted higher attrition levels (across both conditions) for teacher ratings at Time 4, and older age predicted greater attrition in Time 4 parent data. Only two of the 25 predictors (the five baseline characteristics and intervention condition affecting across T2–T4 for parent data, and T3–T4 for teacher data) of possible attrition bias were statistically significant, and only one of the 25 tests of interactions between baseline predictors and condition status (baseline teacher-rated externalizing predicting Time 4 teacher ratings) serving as predictors of attrition was significant. Thus, there was no systematic pattern of attrition bias across time.

Table 2.

Mean and Standard Deviations of Behavioral Outcomes Across Time

Measure	Time 1			Time 2			Time 3			Time 4
	M	SD	N	M	SD	N	M	SD	N	M	SD	N
	Coping Power: Group Format
Externalizing (BASC)
-Teacher-rated	38.79	17.53	177	---	---	---	30.70	17.84	168	31.88	22.87	148
-Parent-rated	27.40	13.68	180	25.28	13.52	178	23.78	11.97	165	23.71	12.55	157
Internalizing (BASC)
-Teacher-rated	14.24	9.58	177	---	---	---	9.94	7.94	168	10.07	10.13	148
-Parent-rated	24.53	11.16	180	22.17	9.85	179	21.88	10.52	165	21.17	9.98	157
Deviant Peer Involvement
-Teacher-rated	1.86	0.56	177	---	---	---	1.88	0.64	168	2.00	0.76	150
-Parent-rated	1.56	0.46	180	1.52	0.47	179	1.48	0.48	168	1.51	0.47	156
	Coping Power: Individual Format
Externalizing (BASC)
-Teacher-rated	43.28	17.26	178	---	---	---	35.21	20.01	161	28.07	18.57	137
-Parent-rated	30.99	14.71	180	27.46	13.65	175	27.49	14.08	161	28.51	15.44	152
Internalizing (BASC)
-Teacher-rated	16.50	8.95	178	---	---	---	10.81	7.53	161	7.88	7.46	137
-Parent-rated	28.51	13.84	180	24.72	11.48	175	24.61	11.51	161	24.34	11.51	152
Deviant Peer Involvement (PASA)
-Teacher-rated	2.01	0.57	177	---	---	---	1.88	0.66	162	1.92	0.65	138
-Parent-rated	1.62	0.48	180	1.50	0.44	178	1.48	0.47	163	1.55	0.46	152

Notes: BASC: Behavior Assessment System for Children. PASA: Peer Affiliation and Social Acceptance. Teacher ratings were not collected at Time 2

When the between-source associations among the six outcome variables were examined at Time 4, all of the 15 correlations were below .34. Overall, the outcome variables were distinct enough from each other to warrant running separate growth models for each outcome.

Table 3 summarizes the results of the six HLM analyses. The quadratic growth curve was significant for the six outcomes, and thus linear slopes in the analyses need to be interpreted within the context of quadratic terms.

Table 3.

Summary of 3-level Growth Curve Analyses on Growth Rate

Variable	Fixed Effect					Random Effect
						Level-3				Level-2				Level-1
	Coef.	SE	t-value	DF	p-value	Var	χ2	DF	p-value	Var	χ2	DF	p-value
Externalizing -Teacher
Model for Intercept, P0
Intercept, G000	38.09	1.69	22.53	58	0.000	40.82	117.78	58	0.000	93.49	499.29	285	0.000	185.60
IGCP, G001	5.91	2.36	2.51	58	0.015
Inhibitory, G010	−1.17	1.31	−0.89	358	0.372
Model for Time Slope, P1
Intercept, G100	−8.72	2.21	−3.95	58	0.000	22.98	126.87	58	0.000	9.80	371.96	285	0.001
IGCP, G101	−3.95	1.71	−2.31	58	0.025
Inhibitory, G110	−1.54	1.15	−1.34	358	0.181
Inhibitory*IGCP,G111	3.63	1.60	2.27	358	0.024
Model TimeSQ Slope, P2
Intercept, G200	2.69	0.95	2.84	958	0.005
Externalizing -Parent
Model for Intercept, P0
Intercept, G000	27.29	1.12	24.32	58	0.000	9.32	82.02	58	0.021	156.1	1909.4	279	0.000	33.31
IGCP, G001	2.85	1.55	1.84	58	0.070
Inhibitory, G010	−3.36	1.04	−3.24	358	0.002
Model for Time Slope, P1
Intercept, G100	−5.13	0.83	−6.18	58	0.000	0.03	51.20	58	>.500	55.49	464.63	279	0.000
IGCP, G101	0.09	0.50	0.19	58	0.850
Inhibitory, G110	0.55	0.52	1.07	358	0.287
Inhibitory*IGCP,G111	0.13	0.76	0.18	358	0.860
Model TimeSQ Slope, P2
Intercept, G200	1.73	0.32	5.46	359	0.000					6.32	422.31	339	0.002
Internalizing –Teacher
Model for Intercept, P0
Intercept, G000	14.08	1.01	13.92	58	0.000	20.75	192.78	58	0.000	18.78	464.03	285	0.000	43.23
IGCP, G001	2.40	1.42	1.69	58	0.096
Inhibitory, G010	−0.12	0.61	−0.20	358	0.842
Model for Time Slope, P1
Intercept, G100	−5.34	1.13	−4.72	58	0.000	10.24	196.34	58	0.000	0.92	352.42	285	0.004
IGCP, G101	−2.05	0.99	−2.07	58	0.043
Inhibitory, G110	−0.15	0.51	−0.30	358	0.767
Inhibitory*IGCP,G111	0.15	0.65	0.23	358	0.818
Model TimeSQ Slope, P2
Intercept, G200	1.60	0.46	3.52	958	0.001
Internalizing –Parent
Model for Intercept, P0
Intercept, G000	24.13	0.84	28.61	58	0.000	0.06	49.16	58	>.500	109.9	1822.5	279	0.000	32.47
IGCP, G001	3.24	1.16	2.79	58	0.008
Inhibitory, G010	−1.24	0.90	−1.38	358	0.169
Model for Time Slope, P1
Intercept, G100	−3.54	0.78	−4.56	58	0.000	0.39	69.07	58	0.152	36.30	387.65	279	0.000
IGCP, G101	−0.53	0.44	−1.20	58	0.235
Inhibitory, G110	0.21	0.42	0.52	358	0.606
Inhibitory*IGCP,G111	0.45	0.59	0.76	358	0.448
Model TimeSQ Slope, P2
Intercept, G200	1.05	0.29	3.61	359	0.001					2.90	339.00	339	>.500
Peer Deviance-Teacher
Model for Intercept, P0
Intercept, G000	1.86	0.05	36.85	58	0.000	0.03	93.94	58	0.002	0.05	406.94	285	0.000	0.27
IGCP, G001	0.16	0.07	2.23	58	0.030
Inhibitory, G010	0.06	0.04	1.29	358	0.199
Model for Time Slope, P1
Intercept, G100	−0.06	0.08	−0.71	58	0.479	0.01	85.26	58	0.011	0.01	370.75	285	0.001
IGCP, G101	−0.13	0.05	−2.50	58	0.016
Inhibitory, G110	−0.10	0.04	−2.36	358	0.019
Inhibitory*IGCP,G111	0.10	0.06	1.71	358	0.088
Model TimeSQ Slope, P2
Intercept, G200	0.07	0.04	1.85	961	0.064
Peer Deviance-Parent
Model for Intercept, P0
Intercept, G000	1.56	0.03	45.79	58	0.000	0.005	71.17	58	0.115	0.13	994.18	279	0.000	0.08
IGCP, G001	0.03	0.07	0.60	58	0.551
Inhibitory, G010	−0.09	0.03	−2.68	358	0.008
Model for Time Slope, P1
Intercept, G100	−0.17	0.04	−4.20	58	0.000	0.0001	53.61	58	>.500	0.13	470.30	279	0.000
IGCP, G101	−0.02	0.02	−0.78	58	0.436
Inhibitory, G110	0.02	0.02	1.18	358	0.240
Inhibitory*IGCP,G111	0.01	0.03	0.22	358	0.826
Model TimeSQ Slope, P2
Intercept, G200	0.07	0.02	4.13	359	0.000					0.02	463.05	339	0.000

Note:

1) Inhibitory Control (from the Early Adolescent Temperament Questionnaire) is a Level-2 variable

2) IGCP is the Level-3 intervention condition, and is dummy coded as 1=ICP and 0=GCP.

Effects of Time, Condition, and Moderation

The HLM analyses indicate that there was significant reduction across time in five outcomes (all except teachers’ ratings of youths’ involvement with deviant peers). These significant change were replicated when slopes were examined within unconditional models.

Condition effects indicate that ICP produced greater reductions in growth rates of teacher-rated externalizing behavior (effect size: δ = .35) and teacher-rated deviant peer involvement (δ = .40) across time than did GCP, but the two conditions did not significantly differ in parent-rated externalizing behavior (δ = .06), or in parent-rated deviant peer involvement (.δ = .03). To provide context for the condition effects on teacher-rated externalizing behaviors and deviant peer involvement, within condition analyses were conducted to explore whether there was a significant reduction in growth rates within both conditions. Results of the within condition growth analyses for teacher-rated externalizing problems indicated that there was a significant declining rate for both the GCP, t(29) = −4.597, p<.001, and ICP, t(29)= −2.873, p<.01, conditions. A different pattern was evident on the within condition growth analyses for teacher-rated deviant involvement, as there was a significant decline in growth rate for ICP, t(29)= −2.081, p<.05) but not for GCP, t(29)= −0.310, ns⁶.

The exploratory analyses examining condition effects on internalizing problems found that ICP did lead to significant declines in growth rates of teacher-rated internalizing problems across time in comparison to GCP (δ = .30), although condition differences were not evident in parent-rated internalizing problems (δ = .01). Results of the within condition growth analyses for teacher-rated internalizing problems indicated that there was a significant declining rate for both the GCP, t(29) = −4.087, p<.001, and ICP, t(29)=−4.818, p<.001, conditions.

The moderation hypothesis that children with lower levels of inhibitory control would be more vulnerable to deviancy-promoting effects operating in the group format was partially supported, although only for teacher-rated externalizing behaviors. As depicted in Figure 1, children with low inhibitory control had greater reductions in growth rates for teacher-rated externalizing behavior problems when they received intervention in an individual format than when they received intervention in a group format; in contrast, children with high levels of inhibitory control benefited from the two formats in similar ways. This interaction qualifies the main effect of condition on teacher-rated externalizing behavior.

Figure 1.

Interaction between Condition and Inhibitory Control on Teacher-rated Externalizing Behavior

Discussion

As hypothesized, growth curve analyses indicate that aggressive children receiving the individual format did indeed have lower rates of change on teacher-rated externalizing problems and teacher-rated deviant peer involvement across time, although there were no significant intervention format effects on parent-rated measures of conduct problems. Children’s baseline level of inhibitory control significantly moderated the format effect on teacher-rated outcomes. As expected, children with lower levels of initial inhibitory control were less responsive to the group intervention as reported by teacher ratings of problem behavior 1 year later.

Interpretation of the limiting effect of the group format for Coping Power is made more complicated by the fact that the two formats did not influence outcomes from parent reports. From the vantage point of parents, it does not seem to matter whether interventions are delivered in a group or individual formats; reductions in problem behavior are reported. It is also inaccurate to say that the effects of the group intervention are iatrogenic. The effect sizes for externalizing behavior reduction in the present study through the one-year follow-up (D=.34) were similar to those found in a randomized control trial of the Coping Power program (Lochman and Wells, 2003) (D=.40). Although the current study lacked a control group, and the decline in children’s problems in the current study can’t be directly linked to the intervention, this reduction in behavior problems (similar to the prior randomized control study) is occurring even though the Coping Power parent component was not delivered in the current study.

Children in the group and individual condition had similar decreases in teacher-rated externalizing, internalizing and deviant peer involvement at the end of the intervention, but additional teacher-reported improvements in the year after intervention were only evident for children who had received Coping Power in an individual fashion. The strong additional improvements for children seen individually may have been due to the fact that they had avoided some negative peer effects in the year after intervention. In addition to differences in peer effects, children seen individually may have learned intervention concepts and skills more deeply, and thus retained them better in the follow-up year. The individual format may have permitted the interventionists to provide more repetition of central ideas, and more individualized role-playing and practice of anger-management and problem-solving activities.

Theoretically and clinically important is the moderation of teacher-reported intervention format effects by children’s self-regulation. Specifically, children in the group condition with lower levels baseline levels of inhibitory control showed less teacher-reported behavior change compared to those in the individual condition over the course of one year. This finding can be explained by two possible mechanisms. First, children with low levels of inhibitory control may be more reactive to provocations in the group (e.g., anger, aggression, etc.), or as found in previous research, more vulnerable to peer inducements to engage in problem behavior despite the group leader’s guidance.

Limitations

One of the limitations of the current study is that we did not follow the youth more than 1 year following the end of the intervention. Another set of limitations are that current findings generalize best to other structured research-based interventions which intervene with children in the preadolescent age range. This age range is a transition period developmentally important for prevention purposes, but is younger than the adolescent period when clearer evidence of iatrogenic group effects are obtained (e.g. Poulin et al., 2001). Since the goal of the study was to examine how group effects operate in evidence-based interventions, the findings also do not address how poorly structured groups might operate in this age period. The Coping Power intervention was closely supervised weekly by doctoral level supervisors with extensive knowledge of the best practices in the intervention. There is some evidence to suggest that group interventions lose their effectiveness mostly when they are implemented in the real world under conditions of less supervision and guidance (Cho, Hallfors et al, 2005).

Next Steps and Implications

First, when group interventions are being delivered, child characteristics should be carefully considered and assessed as potential moderators of intervention outcomes. Although children with relatively higher inhibitory control (at least high in the context of children screened for aggressive behavior problems) were found to profit adequately from a group format for intervention, children with low levels of inhibitory control appeared to profit more from individual delivery of the Coping Power intervention. Other predictors of children’s response to group interventions should also be explored, including neurobiological and neuropsychological indicators of emotional activation and regulation. Children’s weak inhibitory control may be addressed in part by the typical behavioral management training for parents in interventions like Coping Power (in this study of format effects, we did not include the Coping Power parent component) to reduce their risk to some deviant peer influences. In addition, the child components of interventions like Coping Power could be adapted to address children’s neuropsychological limitations related to their weak inhibitory control (e.g. provide shorter intervals for behavioral monitoring; greater attention to long-term consequence for social decisions) (Matthys et al., 2012).

Second, we know remarkably little, from an empirical standpoint, about how group therapists can optimally manage and run group sessions to best counter the deviancy-promoting effects that may be occurring moment-to-moment during sessions. Group therapists who allow less unsupervised time within group sessions and who carefully use behavioral management strategies may have best outcomes (e.g., Dishion & Dodge, 2006). Rigorous research examining how different types of therapist behaviors affect children within the group sessions is crucial.

Public Health Significance.

This study finds that group formats for aggressive children can limit intervention effects in school settings for some children.