Researchers analyze the cost effectiveness of educational administrator training programs for Native Americans at four universities: Harvard, Pennsylvania State, Minnesota, and New Mexico. The programs vary in scope, duration, clientele, admissions, and five other characteristics. The programs' average costs per student are computed and are related to differences in program characteristics. The researchers then compile effectiveness rankings for seven different outcome measures, including program completion rate, post-program jobs and salaries, shifts in job aspirations, and students' perceptions of program quality, changes in their decision-making responsibilities, and realization of preprogram expectations. Program rankings on each outcome are weighted and averaged, and a measure of program cost per unit of effectiveness is computed. While one unidentified school is found to have lower costs, the authors conclude that no school can be considered more cost effective because the use of different outcomes or weights would yield different figures for cost effectiveness. They discuss the problems of analyzing the cost effectiveness of social action programs, especially in relation to variation in program goals, structures, resource constraints, and outcome measures. (Author/RW)