The Minneapolis Domestic Violence Experiment (1981 – 1982) was conducted within the theory of statistical inference. It is an example of experimental design (meaning it was a controlled experiment), with the initial hypothesis that further behavior of domestic violence offenders depends highly on the measures taken by police. Before this experiment police had neither right nor obligation to interfere with domestic violence cases, as they were referenced as private family matter. Only in cases of serious violence arrest was applicable. However, in 1970s clinical psychologists began to insist on the necessity of police responses, and under financial support of the National Institute of Justice Lawrence W. Sherman initiated the experiment in Minneapolis, Minnesota. During about 17 months 25 field experiments were conducted, involving randomly chosen families. The control group held up supervision over the offender during half a year after the registered act of offense. Police used either isolation from the family for 8 hours, or mediation counseling or arrest. The latter turned out to be the most effective in terms of further behavior of the offenders.
However, the experiment has demonstrated certain shortcomings both in methodology and evaluation of results. First of all, the mandatory arrests within the experiment under consideration were based on deterrence theory. This theory has an assumption that being intimidated of punishment, the offenders tend to refrain from further offenses. But the matter is this theory derives from the condition that the offenders act according to rational decisions, while those who are accused of violent actions actually rarely show rational behavior. Secondly, there was rough generalization in selection of cases, when neither cultural background of the family nor the context of the action was taken to account. Besides, the period in six months is considered to be rather short to make proper conclusions, as the accidence of offences may be periodical and defend on some seasonal and other factors. Thus, episodic and cyclical patterns were not captured. The amount of offenders who took part in the design experiment was approached as one, homogeneous group, with no diversification mentioned. However, in such studies there is a strong need to take into account the social, cultural, financial, behavioral differences and be careful with tracing negative effects of the intervention. On the one hand, deterrence may really work in some cases, but it was not mentioned that the offenders were often released almost immediately; police officers have shown low level of participation and follow-through in truly randomizing the police responses was failed. Further on, it was not mentioned that arrests may provoke even stronger, more retributive violence. On the other hand, the further cases of offense in the same families could have been missed not because the offenders were afraid of punishment, but because their victims were afraid to call, either because of retribution and recidivism of their offenders or because they did not want their offenders to be arrested, while their will was not taken to account either.
In this way, the main disadvantage of the experiment and conclusions that were made after it is that in reality there can be no singular approach for police to response to domestic violence. The need for diversification has been proved by further experiments, which became the replication of the Minneapolis Domestic Violence Experiment in Colorado Springs, Colorado; Miami-Dade-County, Florida; Omaha, Nebraska; Charlotte, North Carolina and Milwaukee, Wisconsin.
Buzawa, E. S., & C. G. Buzawa (1990). Domestic Violence: The Criminal Justice Response. Thousand Oaks, CA: SAGE.
Fagan, Jeffrey (1989). “Cessation of Family Violence: Deterrence and Dissuasion.”¯ Crime and Justice: an Annual Review of Research, 11, 377”“425.