Raymond M. Berger, MSSW, is a doctoral candidate and Irving Piliavin, DSW, is Professor, School of Social Work, University of Wisconsin, Madison. The authors greatly appre ciate the help given them by the late Margaret Blenkner, her research staff, and Mary Hemmy, Executive Director of the Benjamin Rose In stitute. All the analyses reported here made use of the data file originally developed by Dr. Blenkner and her staff. In a review of eleven controlled experimentais and controls other than studies, Fischer concluded that social that the experimentais were less likely casework services—defined broadly as to survive and more likely to be in the services provided by professional stitutionalized than were the controls, caseworkers—yielded no greater imAlthough institutionalization was re provement among a variety of client lated to poor survival for both groups, populations than did no intervention it did not account for the different at all.1 One of the studies reviewed by survival rates between experimentais Fischer was the Protective Services and controls. Statistical tests were not Project for Older Persons, a longitudireported but experimentais among nal study directed by Blenkner under both institutionalized and noninstitu the auspices of the Benjamin Rose Intionalized participants were said to be stitute in Cleveland, Ohio.2 Accordless likely to survive than the controls, ing to Fischer, the results of this study The reason for the poorer survival rate suggested that the special, enriched among experimentais remained some services of the institute in fact worked thing of a mystery, against the clients, since the mortality In reviewing the analytic procedures rate of the clients who received the utilized in the Rose Institute study, special services was higher than that the authors found two possible ex of the control clients. Three years planations for the findings reported after the beginning of the program, by Blenkner and her colleagues. The 47.4 percent of the special service first concerns the influence of im clients were still alive compared to portant covariates of survival. AI 64.9 percent of the controls.8 The though the original investigators implication that social services can found that pretest data on age, physi have debilitating effects is obviously a cal functioning, and mental status of matter of concern. participants were linked to survival This research note presents a more (see Table 1), these variables played elaborate statistical analysis of the only a limited role in the analysis of original data than that carried out by experimental and control survival the researchers of the Rose Institute rates.4 This was unfortunate, since study. Such analysis indicates that the these variables indicated that the ex apparently negative effects of the inperimentals were more debilitated stitute program are attributable to facthan the controls, tors other than the special services proThe second possible explanation for gram itself. Blenkner's findings involves the sub Participants in the Rose Institute stantial number of cases with missing study were 164 elderly persons who pretest data. The overall comparison were in need of protective services of control and experimental death and who had come to the attention rates included individuals for whom of various community agencies. Sevinformation on physical functioning enty-six persons were randomly asand mental status was lacking. Ad signed to an experimental group and mittedly, the percentage of such in received intensive casework and other dividuals was about the same for ex services for a period of one year, perimentals and controls, but it is Eighty-eight persons were assigned to possible that the health status of non a control group and received only assessed experimentais was worse than those incidental services normally that of nonassessed controls. Unfor provided by community agencies. The tunately, there is no way of directly effects of service were evaluated by confirming this possibility since the a variety of variables that measured original Rose Institute study provides mental and physical competence, envery little information on the proce vironmental protection, affect, and dures by which diagnostic information effect on family and close friends. on participants was obtained at the Analyses reported by the researchinception of the experiment. However, ers revealed few differences between there are data—the variables just dis