Minneapolis Domestic Violence Experiment Police Foundation ReportsPage 2 of 13 April 1984
surveyed. Despite the fact thatviolence is reported to be presentin one-third (Bard and Zacker,1974) to two-thirds (Black, 1980)of all domestic disturbancespolice respond to, policedepartment data show arrests inonly five percent of thosedisturbances in Oakland (Hart,n.d., cited in Meyer and Lorimer,1977:21), six percent of thosedisturbances in a Colorado city(Patrick, Ellis, and Hoffmeister,n.d., cited in Meyer and Lorimer,1977:21), and six percent in Los Angeles County (Emerson, 1979).The best available evidence onthe frequency of arrest is theobservations from the Black andReiss study of Boston,Washington, and Chicago policein 1966 (Black, 1980:182). Policeresponding to disputes in thosecities made arrests in 27 percentof violent felonies and 17 percentof the violent misdemeanors. Among married couples (Black,1980:158), they made arrests in26 percent of the cases, but triedto remove one of the parties in 38percent of the cases.The apparent preference of manypolice for separating the partiesrather than arresting theoffender has been attacked fromtwo directions over the past 15years. The original critique camefrom clinical psychologists whoagreed that police should rarelymake arrests (Potter, 1978:46;Fagin, 1978:123-124) in domesticassault cases and argued thatpolice should mediate thedisputes responsible for theviolence. A highly publicizeddemonstration project teachingpolice special counseling skills forfamily crisis intervention (Bard,1970) failed to show a reductionin violence, but was interpretedas a success nonetheless. By1977, a national survey of policeagencies with 100 or moreofficers found that over 70percent reported a family crisisintervention training program inoperation. Although it is notclear whether these programsreduced separation and increasedmediation, a decline in arrestswas noted for some (Wylie, et al.,1976). Indeed, many soughtexplicitly to
the number of arrests (University of Rochester,1974; Ketterman and Kravitz,1978).By the mid-1970s, policepractices were criticized from theopposite direction by feministgroups. Just as psychologistssucceeded in having many policeagencies respond to domesticviolence as “half social work andhalf police work,” feminists beganto argue that police put “toomuch emphasis on the socialwork aspect and not enough onthe criminal” (Langley and Levy,1977:218). Widely publicizedlawsuits in New York andOakland sought to compel policeto make arrests in every case of domestic assault, and statelegislatures were lobbiedsuccessfully to reduce theevidentiary requirements neededfor police to make arrests formisdemeanor domestic assaults.Some legislatures are nowconsidering statutes requiringpolice to make arrests in thesecases.The feminist critique wasbolstered by a study (PoliceFoundation, 1976) showing thatfor 85 percent of a sample of spouse killings, police hadintervened at least once in thepreceding two years. For 54percent of those homicides, policehad intervened five or moretimes. But it was impossible todetermine from the data whethermaking more or fewer arrestswould have reduced the homiciderate.
How the Experiment Was Designed
In order to find which policeapproach was most effective indeterring future domesticviolence, the Police Foundationand the Minneapolis PoliceDepartment agreed to conduct aclassic experiment. A classic experiment is aresearch design that allowsscientists to discover the effectsof one thing on another byholding constant all otherpossible causes of those effects.The design of the experimentcalled for a lottery selection,which ensured that there wouldbe no difference among the threegroups of suspects receiving thedifferent police responses (Cookand Campbell, 1979). The lotterydetermined which of the threeresponses police officers woulduse on each suspect in a domesticassault case. According to thelottery, a suspect would bearrested, or sent from the sceneof the assault for eight hours, orgiven some form of advice, whichcould include mediation at anofficer’s discretion. In thelanguage of the experiment,these responses were called thearrest, send, and advisetreatments. The design called fora six-month follow-up period tomeasure the frequency andseriousness of any futuredomestic violence in all cases inwhich the police intervened.The design applied only to simple(misdemeanor) domesticassaults, where both the suspectand the victim were presentwhen the police arrived. Thus,the experiment included onlythose cases in which police wereempowered, but not required, tomake arrests under a recentlyliberalized Minnesota state law.