LOOKING BACK 20 YEARS
Implications of Outcome Research on Residential Treatment Services
What does the outcome research tell us about the effectiveness of residential treatment? The short answer is, "it depends." It depends on how one reads the data, when one reads the data, and how confident one is about the methodology of the various investigators. To take the last first, methodological problems involving internal and external validity and reliability abound in evaluations of residential treatment. Typically, these involve absence of control/contrast groups; absence of random assignment; vaguely defined service units; narrow or inappropriate selection of outcome criteria; biased sample selection; and observer/rater bias [Whittaker 1984]. On the remaining question of how and when one reads the data, the following generalizations appear to be borne out across many studies. First, if one limits analysis to objective indicators of adjustment, like school behavior, court contacts and the like, residential treatment fares rather poorly. On more subjective indicators, such as therapists' judgment of progress and various forms of consumer evaluation, residential treatment appears to be rather effective. Second, if one samples behavior during residential treatment, or immediately at discharge from the treatment milieu, residential treatment looks reasonably effective; but if one samples behavior at increasing intervals in such distal environments as school, community, and family, the decay of treatment effects appears to be pronounced.
Bearing in mind the caveats on the potential weaknesses in research design and the difficulties in interpretation of results, the corpus of outcome research in residential treatment yields several general findings. First, the postdischarge environment appears to be a powerful factor in determining successful long-term adjustment, irrespective of gains made while in the program. Studies by Allerhand et al. , Taylor and Alpert , Cavior et al.  and others all support this general finding. The Allerhand et al. study, which involved extensive follow-up of 50 graduates of a sophisticated residential treatment agency (Bellefaire) in Ohio, summarizes the major findings as follows:
Perhaps the most striking finding of the study is that none of the measurements of within-Bellefaire performance at discharge, either in casework or in cottage and school roles, were useful in themselves in predicting postdischarge adaptability and adaptation. Only when the situation to which the child returned was taken into account were performances at Bellefaire related to postdischarge adequacies. In a stressful community situation, strengths nurtured within the institution tended to break down, whereas in a supportive situation, these strengths tended to be reinforced. [p. 140]
The importance of the postplacement environment was underscored in the later research by Taylor and Alpert, which indicated that contact with biological parents while the child was in placement was positively correlated with postdischarge adjustment. These same researchers conclude that neither the child's presenting symptoms, nor any specific treatment variables were strongly associated with postdischarge adjustment: "[it] is not possible to predict a child's postdischarge adaptation on the basis of a given set of preadmission characteristics" [p. 35]. Similar findings with respect to specific treatment variables and outcome were identified in studies conducted by Davids and his associates. They conclude, "treatment variables, especially conventional psychotherapy, seem to bear little relationship to subsequent adjustment" [Davids et al. 1978].
Part of this lack of demonstrated effect may stem from our still primitive attempts to conceptualize and measure the key treatment variables in a complex intervention like residential treatment. Nelson et al.  found that children who left residential treatment with supportive community ties to family, friends, neighbors, schools, and the like were more likely to maintain their treatment gains than those who did not. Those with support maintained over 70% of their gains, while those without support maintained only 50%. Though the sample size was small (22), this study is notable in that it measured behavior at four points in time beginning with a pretreatment community baseline. Similar results were obtained by Lewis  in a follow-up of a ReEd program, though again, small sample size limits generalization.
The similarity of these findings should not be surprising to anyone involved in either service delivery or program evaluation in services to children. The maintenance and generalization of treatment effects, as well as, ultimately, replication of program models remain paramount issues for those involved in mental health services for children.
JAMES WHITTAKER and ANTHONY MALUCCIO
Whittaker, J.K. and Maluccio, A.N. Changing paradigms in residential services for disturbed/disturbinbg children: Retrospect and prospect. In Hawkins, R.P. and Breiling, J. (1989). Therapeutic foster care: Critical issues. Washington: Child Welfare League of America.
Allerhand, M. E.; Weber, R.; and Haug, M. 1966.Adaptation and adaptability: The Bellefaire follow-up study. New York: Child Welfare League of America.
Cavior, E. C; Schmidt, A.; and Karacki, L. 1972. An evaluation of the Kennedy Youth Center Differential Treatment Program. Washington, DC: U.S. Bueaus of Prisons.
Davids, A.; Ryan, R.; and Salvatore, P. 1978. Effectiveness of residential treatment. American Journal of Orthopsychiatry 38: 469-475.
Lewis, W. W. 1982. Ecological factors in successful residential treatment. BehavioralDisorders 7: 149-156.
Nelson, R. H.; Singer, M. J.; and Johnsen, L. O. 1978. The application of a residential treatment evaluation model. Child Care Quarterly 7: 164-175.
Taylor, D. A., and Alpert, S. W. 1973. Continuity and support following residential treatment. New York: Child Welfare League of America.
Taylor, R. B. 1981. The kid business. Boston: Houghton Miftin.
Tharp, R. G., and Wetzel, R. J. 1969. Behavior modification in the natural environment. New York: Academic Press.
Trieschman, A. E; Whittaker, J. K.; and Brendtro, L. K. 1969. The other 23 hours: Child care work in a therapeutic milieu. New York: Aldine.
U.S. Children's Bureau. 1983. Child Welfare Research Note #1. Washington, DC: Department of Health and Human Services.
Whittaker, J. K. Formal and informal helping in child welfare services: Implications for management and practice. Child Welfare 65: 17-25.