The programs were selected to represent a range of program format in terms of duration and services. The first 20 to 25 men appearing for program intake each month were recruited as subjects, until approximately 210 men had been recruited at each of the 3 research sites. The vast majority of the men were mandated to the programs by the courts as opposed to others that entered the programs voluntarily. The program outcome was based primarily on phone interviews with the initial or new partners of the men in the sample. Instrumental variable analysis was used to test for a program effect in the multi-site evaluation of batterer intervention and evidence was found of a moderate program effect across three sites. A man who completed a batterer program was 44 percent to 64 percent less likely to re-assault his partner than those men who did not complete a program. If all the men who enroll in a batterer program could be motivated or coerced to complete the program, the overall re-assault rate of 32 percent (for program completers and dropouts) would decrease a minimum of 44 percent to an 18 percent re-assault rate. In the follow-up sample of 480 women, these percentages implied that only 86 women instead of 155 women would have been re-assaulted if all partners had completed the programs. Program length is not a significant predictor of program completion, and completion of any of the programs, regardless of required length, leads to a reduction in the probability of re-assault. There is a need for a more complex analysis of program effect--an analysis that goes beyond the prevailing designs and analytical approaches. 1 table, 4 notes, 40 references