These home characteristics are favorable social climate, extensive linkages with the community, and good quality of these linkages. Program graduates' scores on questionnaires to measure perceptions about significant others, expectation and aspirations, and recidivism rates were compared to determine a relationship between program characteristics and results. The comparative evaluation research design was adapted from Campbell and Stanley's nonequivalent control group design and separate-sample pretest/posttest design. Programs were evaluated by analysis of variance techniques. The group home programs displayed deinstitutionalized features and significantly differed from each other in the level of development of these features. They did not significantly differ from each other in their effectiveness in rehabilitating delinquents. Three reasons are offered for this: the sample may have been too small, missing data for some subjects could have affected overall study results, and, finally, the one aspect on which all programs scored relatively equal (extent of community linkages) may be the most important of all the variables for influencing rehabilitation. Study data, footnotes, about 160 references, and study instruments are provided. (Author abstract modified)