EVALUATING THE EFFECTIVENESS OF CHILD AND ADOLESCENT GROUP PSYCHOTHERAPY: A META-ANALYTIC REVIEW

by

Print Friendly, PDF & Email

Corey D. Randall and Dr. Matthew J. Haag, Department of Psychology and Clinical Psychology

Introduction

Effective treatment of children and adolescents in psychotherapy is crucial to alleviating the distress and suffering resulting from psychopathology (Kazdin, 1993). Dysfunctional behavior, if not appropriately addressed, can result in irreversible developmental outcomes. The importance of evaluating treatment to address childhood difficulties becomes even more vital in a managed health care system that stresses accountability and demands evidence of progress. It is for this reason that an examination of the current treatment approaches and techniques must be undertaken.

Methods and Design

Variables that have been described as significant to understanding child, adolescent, and adult group treatment in other reviews of the literature were identified (e.g .. methodology, client, therapist, and process variables) (Abramowitz, 1976; Dagley et al., 1994; Dies & Riester, 1986; Gazda & Larsen, 1968; Kraft, 1968; Sugar, 1993).

Articles were obtained for this meta-analysis by a computer search of PsycLit to locate all articles published between 1974-1994 on child and adolescent group psychotherapy. The reference sections of previous reviews, recent relevant publications, and articles identified in the computer search were examined for studies which could be included in the analysis (e.g., Casey & Berman, 1985; Dagley, et. a!. 1994; Dies & Riester, 1986; Sugar, 1993; Tramonta, 1980; Weisz eta!., 1987). Initially, well over 100 studies were identified for possible inclusion in this study. Articles were then included or excluded from the meta-analysis based on criteria similar to those which Dagley et al., (1994) and Stinchfield, Owen, and Winters (1994) used in their narrative reviews of group therapy for children and adolescents and substance abuse respectively. These criteria resulted in 43 articles being identified which were included in the present analysis.

Effect sizes were calculated from the outcome statistics in each study using the DSTAT computer software package Uohnson, 1989) according to the within-study meta-analysis formula given below. a Where dis the estimated effect size, M1 and M2 are the means of the groups being compared, and Sp is the pooled within groups standard deviation (Cohen, 1977). This formula was utilized to calculate effect sizes directly comparing group therapy with controls or placebos as well as examining pre-to-post-test improvement. Thus, an effect size of 1.00 would indicate that the Ml group achieved an effect one standard deviation above that obtained by the M2 group. It could then be said that the average person in the Ml group achieved an outcome that was better than 84% of the people in the M2 group. Likewise, an effect size of -1.00 would indicate that the average person in the Ml group fared worse than 84% of the subjects in the M2 group. Treatment (Ml). control or placebo (M2), and pre- (Ml) post( M2) groups were analyzed using this formula.

Results

Effect sizes were calculated comparing group treatment to wait list control as well as placebo control groups. However, an independent samples t-test indicates that the wait list mean effect size (M=.51) did not differ significantly from the placebo controls (M=.49), t(41)=.14, p,=.89. As a result, we combined effect sizes from these two groups for our analysis of the overall effect size as well as our examination of variables. In addition, when studies reported results for both a wait list and a placebo control group, the effect sizes were averaged to maintain independence. This resulted in 37 of the 43 studies being utilized for the analysis of treatment compared to wait list and placebo control groups.

The overall effect size of the 37 studies which compared group treatment to a wait list or placebo control group was.50 (significantly different from zero, p,=.OOO). In other words, the average treated child or adolescent was one half a standard deviation better than untreated children or adolescents. Thus the child in treatment can be placed at the 67th percentile of those not in treatment. In addition, an effect size was calculated which reflects overall pre-post differences. This computed effect size was .62 and also was significantly different from zero (P.=.OOO).

The second question posed by this study regarding whether levels of each variable contribute differential effectiveness to the overall effect size was examined. The variables examined and the overall effect size provide significant findings in client, treatment, and the methodological domains.

Analysis indicates that experimenter allegiance had a significant effect on the effect size. In other words, when it was clear that the experimenter had a preference for one of the types of therapy, this therapy did significantly better than therapies in studies with no stated preference, E(1,11)-4.63, p,=.05.

A treatment variable that obtained significance was treatment standardization. Briefly, treatment standardization examines whether a study uses a manual or monitors adherence to a theoretical model. Results indicate that studies high in treatment standardization were significantly more improved than those that were medium in treatment standardization !’:(2,34) = 4.11, p,=.03.

A client variable that also obtained significance was that of socioeconomic status (SES). An examination of the studies in this meta-analysis found that the following categories emerged: 1) low, 2) middle, and 3) lower to middle. Analysis found that studies that classified children and adolescents as “middle” SES produced significantly greater results than those children that were classified as “low” SES, !’:(2, 13)=5.64, p,=.02. Unfortunately, these were the only SES classifications provided by the studies under investigation.

Two group variables also approached significance and their results seem relevant to our present discussion. Groups that were oriented to counseling or therapy were more effective than psychoeducational groups (i.e., structured or didactic experience), !’:{1,34)=3.90, p,=.06. In addition, groups that were larger in size (greater than 1 0) produced larger effect sizes than moderately sized groups (between 5-9 children or adolescents), E(2, 32)=2.50, p,=.10.

No other client, therapist, treatment, group, or methodological variables approached significance or were significant in their impact on the overall effect size. Nonsignificant client variables included: gender, age, race, diagnosis, chronicity of symptoms, and entrance in therapy. Therapist variables which failed to reach significance included gender, experience, and number of therapists leading the group. Frequency and length of treatment were variables in the treatment domain that also did not reach significance. Whether the group was heterogenous or homogenous and whether it was open or closed with regard to membership were two nonsignificant group variables. Methodological variables that failed to reach significance included validity of the studies and source, content, and reactivity of the measures.

Discussion and Conclusion

This study investigated whether the efficacy of group psychotherapy with children and adolescents could be established via meta-analysis. While narrative reviews have suggested that research investigating group therapy with children and adolescents suffered from methodological flaws and failed to demonstrate efficacy, meta-analytic reviews of group therapy with children and adolescents have continued to demonstrate effectiveness. However, these meta-analytic studies were not investigating group therapy specifically, but saw group therapy as one variable among many being examined. This study has shown that group therapy with children and adolescents is effective. Children and adolescents that received group therapy improved significantly more than wait list or placebo controls. The overall effect size obtained by this study is comparable to those found by other meta-analyses which utilized heterogenous subjects (Casey & Berman, 1985; Prout & DeMartino, 1986; Russell eta!., 1991; Shirk & Russell, 1992; Weisz, eta!, 1987).

It is difficult to ascertain what contributes to the overall effectiveness of this treatment modality. Few client, therapist, and group variables contributed to differential effectiveness. The only client variable described as significant in the current study was socioeconomic status. Sugar (1993) noted that SES was one variable with no research on outcome. While this study provides tentative data on middle and low SES children and adolescents, it represents only a beginning in identifying which children respond best to group therapy.

Further research should investigate the specific variables which make group psychotherapy effective. It appears from this report that those studies which utilize strict adherence to therapeutic models or utilize manuals produce greater effects. Perhaps other variables exist which provide further support to the group therapist as he/she prepares to utilize this treatment modality. Only through increased attention to methodological rigor will these variables be identified.

Additional details and data of this meta-analysis are found in a larger document presented at the 1995 National Convention of the American Psychological Association and may be obtained upon request.

References

  • Abramowitz, C. V. (1976). “The effectiveness of group psychotherapy with children.” Archives of General Psychiatry 33:320-326.
  • Casey, R. J,, & Berman,]. S. (1985). “The outcome of psychotherapy with children.” Psychological Bulletin 98(2):388-400.
  • Cohen,]. (1977). Statistical Power Analysis for the Behavioral Sciences (rev. ed.). New York: Academic Press.
  • Dagley,]. C., Gazda, G. M., Eppinger, S. J .. & Stewart, E. A. (1994). Group psychotherapy research with children, preadolescents, and adolescents.
  • In A. Fuhriman & G. M. Burlingame (Eds.). Handbook of Group Psychotherapy. 340-369. New York: Wiley.
  • Dies, R. R., & Riester. A. E. (1986). “Research on child group psychotherapy. Present status and future directions.” In A. E. Riester & 1. A. Kraft (Eds.), Child Group Psychotherapy: Future tense (pp. 173-220). Madison: International Universities Press, Inc.
  • Gazda, G. M .. & Larsen, M.]. (1968). “A Comprehensive Appraisal of Group and Multiple Counseling Research.” Journal of Research and Development in Education. 1 (2):5 7-132.
  • Johnson, B. T. (1989). DSTAT: Software for the Meta-analytic Review of Research Literatures. Hillsdale, NJ: Lawrence Eribaum.
  • Kazdin, A. E. (1993). “Psychotherapy for children and adolescents: Current progress and future research directions.” American Psychologist. 48(6):644-657.
  • Kraft, I. A. (I %8). “An Overview of Group Psychotherapy with Adolescents.” International Journal of Group Psychotherapy. 39(2):255-263.
  • Prout, H. T., & DeMartino, R. A. (1986). “A Meta-Analysis of School-Based Studies of Psychotherapy.” Journal of School Psychology. 24:285-292.
  • Russell, R. L .. Greenwald, S., & Shirk, S. R. (1991). “Language Change in Child Psychotherapy: A Meta-analytic Review.” Journal of Consulting and Clinical Psychology. 59(6):916-919.
  • Shirk, S. R., & Russell, R. L. (1992). “A Reevaluation of Estimates of Child Therapy Effectiveness.” Journal of American Academy of Child and Adolescent Psychiatry. 31 (4):703-709,
  • Stinchfield, R., Owen, P. L., and Winters, K. C. (1994). “Group Therapy for Substance Abuse: A Review of the Empirical Research.” In A. Fuhriman & G. M. Burlingame (Eds.), Handbook of Group Psychotherapy. 458-488. New York: Wiley.
  • Sugar, M. (1993). “Research in Child and Adolescent Group Psychotherapy.” Journal of Child and Adolescent Psychotherapy. 3(4):207 -226.
  • Tramonta, M.G. (1980). “Critical Review of Research on Psychotherapy Outcome with Adolescents: 1967-1977 ,” Psyf!:!Q)Qgjff!LI!ul)etin. 88(2):429-450.
  • Weisz,J. R., Weiss, B .. Alicke, M.D., & Klotz, M. L. (1987). “Effectiveness of Psychotherapy with Children and Adolescents: A Meta-analysis for Clinicians.” Journal of Consulting and Clinical Psychology. 55(4):542-549.