The inconsistency of results to be found in the study of the impact of school desegregation arises in part from the failure to appreciate the complexity of the phenomenon. This report takes a broader view of school desegregation; the basic dependent variables are school level means in the following basic areas: (1) racial attitudes, (2) political attitudes; (3) self-confidence; and (4) achievement. Using a one-way analysis of variance approach, the investigation examined the ability of each of three mediating variables--the school environment, the family, and the peer group--to reduce the original differences in the inter-school means (considered separately for each race). The following theoretical approaches to understanding inter-school differences are considered: (1) contact theory; (2) deprivation theory; and (3) socialization theory. Seniors in six Southern urban high schools were surveyed, and 85 percent of them were interviewed. It was found that, although significant relationships may exist between a covariate and a dependent variable, there may be no reduction in inter-school differences when the covariate is removed. This factor may explain a good deal of the contradiction in earlier findings. Overall, contact theory was found to be unsatisfactory as an approach to understanding inter-school differences. Socialization theory is satisfactory only for whites, and only in the areas of political and racial attitudes. Deprivation theory is most successful, for both races and for all four of the tested areas. Data are presented on eight tables. An appendix lists the survey questions. A 40-item list of references is included. (Author/BJV)