A HOTLY CONTESTED ISSUE IN EVALUATIVE RESEARCH IS THE IMPORTANCE OF COMPARISON GROUPS. THE FUNCTION OF COMPARISON GROUPS--INDIVIDUALS NOT IN THE EVALUATED PROGRAM WHO ARE OTHERWISE COMPARABLE TO PROGRAM PARTICIPANTS--IS TO PROVIDE A MEANS FOR ESTIMATING GAINS ATTRIBUTABLE TO PROGRAM INVOLVEMENT. OFTEN THE IMPORTANCE OF COMPARISON GROUPS IS DOWNPLAYED BECAUSE THEY ARE DIFFICULT TO COME BY AND DO NOT ADD ENOUGH INFORMATION TO JUSTIFY INCLUDING THEM IN RESEARCH. ANOTHER FACTOR IS THE GENERAL IMPRESSION THAT COMPARISON GROUPS MUST RECEIVE ABSOLUTELY NO PROGRAM BENEFITS, THAT RESEARCH DESIGNS USING RANDOMLY CHOSEN COMPARISON GROUPS ARE TOO RIGID FOR EVALUATIVE RESEARCH, AND THAT THE RANDOMIZATION PROCESS ITSELF IS FRAUGHT WITH MORAL AND POLITICAL PROBLEMS. TO INCREASE ACCEPTANCE AND USE OF COMPARISON GROUPS, ESPECIALLY THOSE CHOSEN RANDOMLY, IN EVALUATIVE RESEARCH, THREE RESEARCH DESIGNS--COHORT-CONTROL DESIGN, NO-PREFERENCE DESIGN, AND INTENSIVE (MINIMAL) PROGRAM DESIGN--USED TO EVALUATE STUDENT SERVICE PROGRAMS AT SOUTHERN ILLINOIS UNIVERSITY ARE REVIEWED. EACH DESIGN ALLOWS AT LEAST THE OPTION OF RANDOMIZATION, BUT NONE REQUIRES A NO-PROGRAM-PARTICIPATION COMPARISON GROUP. TWO FACTORS THAT HELPED TO GAIN ACCEPTANCE FOR THE RANDOM SELECTION PROCESS AT SOUTHERN ILLINOIS ARE NOTED: (1) THE TARGET PROGRAMS WERE IN SHORT SUPPLY (MORE ELIGIBLE PARTICIPANTS THAN THE PROGRAMS COULD HANDLE), AND (2) THE RANDOMIZATION PROCESS WAS PRESENTED TO POTENTIAL PROGRAM PARTICIPANTS AS BEING ANALAGOUS TO A LOTTERY. THE DESIGNS ARE EXPLAINED BY DESCRIBING THEIR APPLICATIONS. TABLES ILLUSTRATING THE ASSIGNMENT OF RESEARCH SUBJECTS TO STUDY GROUPS UNDER EACH DESIGN ARE INCLUDED, TOGETHER WITH A LIST OF REFERENCES. (LKM)