Examining Differential Officer Effects in the Minneapolis Domestic Violence Experiment

Under the MDVE, which began on March 17, 1981, all cases of misdemeanor domestic assault where both the suspect and victims were present when the police arrived were randomly assigned to receive one of three distinct police responses. The three experimental treatments were arrest, under which the suspect was taken to jail; advise, which could include mediation; and separate, under which the offender was ordered from the premises. Based on analyses, the study concludes that "the arrest intervention certainly did not make things worse and may well have made things better," thus leading the authors to "favor a presumption of arrest." A new way of approaching the issue of differential officer effects, which is applied here, involves asking whether or not those officers who turned in the majority of the cases were more or less likely than other officers to follow the experimental criteria. This question examines whether or not the quantity of the cases generated by the officers was related to the quality of those cases regarding the proper implementation of the experimental guidelines. More specifically, the study examined whether the officers who contributed most of the cases in the MDVE were responsible for a lesser share of the 21.82 percent of the cases in which the delivered treatment was different from the randomized treatment. The findings show that this was indeed the case, strongly suggesting that there was an officer effect at work in the MDVE. One interpretation of the weight of the evidence from the current analyses is that there is need to rethink the original conclusion of the MDVE regarding the effectiveness of arrest. Although there may well be other reasons why arrest would be beneficial, such as getting the participants into the criminal justice system to receive services otherwise unavailable, apparently arrest is not the panacea that many had hoped in terms of specifically deterring spouse abusers. 6 tables, 5 notes, and 16 references