Abstract
Most previous evaluations of batterer programs have been quasi-experimental designs or evaluations with no comparison group. Two recent experimental evaluations attempted more rigorous science and produced little or no evidence of program effect for conventional batterer counseling compared to a control group. These experimental evaluations must be viewed with caution because of methodological limitations of their own. As with many experimental evaluations, there were difficulties in implementing the random assignment of subjects and reliance on an intention-to-treat approach amid substantial dropout rates. Also, the response rates to the follow-up are relatively low, leaving one study to use probation records rather than victim reports for the outcome. Other studies have shown the influence of a variety of contextual factors on program outcome. There are alternative analytical methods and research designs to address some of these problems. These alternatives, along with practitioner and clinical experience, need to be weighed in the balance.

This publication has 19 references indexed in Scilit: