Lessons from a successful and failed random assignment testing batterer program innovations

With increasing pressure to conduct experimental evaluations of domestic violence interventions, it is important to weigh further the most challenging aspect of experimental designs: the implementation of random assignment. This paper reviews two attempted experimental evaluations of counseling programs for domestic violence offenders, and formulates implications for conducting and interpreting program evaluations. The two case studies offer an instructive comparison of a maximally implemented experiment and a failed one at the same setting. In the first study, the random assignment was introduced within the counseling program and with implicit leverage of court sanctions for non-compliance to the assignment. In the second, random assignment was disrupted by unforeseeable events and inter-agency breakdowns in the complex referral system. Interestingly, implementation issues in both studies raised divergent interpretations from researchers and practitioners. They appear to imply a need for more disclosure of implementation problems in experimental evaluations and for more caution about over interpreting the existing experimental evaluations in the field.