الگوهای بازداشت در برخوردهای خشونت خانگی: شناسایی یک مدل تصمیم گیری پلیس

Arrest patterns of police officers in domestic violence events were explored for a stratified random sample of domestic violence incidents (from police reports) in two police districts in Boston for the calendar year 1993. The initial analytic strategy used was the chi-square automatic interaction detector, which conducts segmentation modeling useful for identifying interaction effects among a predefined set of variables. The interaction effects were then entered into several logistic regression models to generate odds ratios in the predictions of arrest. Results showed that risk to the victim is the most important decision-making criterion for officers, and that officers use variable pieces of information in deciding when to make arrests in domestic violence events based on the different levels of risk.

It has been suggested that historically, more than half of all American couples have experienced minor incidents of domestic violence (Straus and Gelles, 1990), and further, that acts of serious violence have been frequently committed between domestic partners. As reported by Straus and Gelles (1990), findings from the National Family Violence Survey revealed that over one-third of all violence committed against wives was serious and involved punching, kicking, biting, beating, and attacks with guns and knives. Finally, it has been shown that domestic violence is not limited to any particular (or predictable) social class. Most criminal justice interventions have involved members of low socioeconomic status, though several researchers have found that domestic violence transcends SES stratification ( Bassett 1980 and Coates and Leong 1980; Final Report, 1984; Hart et al., 1984).

The vast body of literature related to domestic violence and the intervention of the criminal justice system is a reflection of its importance as a major social issue and problem in American culture. Considering the breadth of research conducted by capable scientists on the police response to domestic violence, it is hoped that the present study makes a substantive contribution to the understanding of how police dispose of domestic violence events. Past inquiries into the correlates of arrest in assaultive incidents between intimate couples have investigated several direct effects of situational-, offender-, and victim-specific conditions on the arrest outcome. Such inquiries presumed that independent conditions (i.e., predictors) were normally distributed across the samples studied. The present study made an alternative presumption that subgroup differences interacted to predict an offender’s relative risk of arrest at a domestic violence incident. This was important because it facilitated an awareness of police officer decision making, and detected potential biases among a sample of officers, especially as the potential risk to the victim diminished. The findings here were limited to the extent that neither the individual officer characteristics, nor the offender demeanor were included in the analysis, which represented a weakness of the data set. Future analyses using similar analytical tools should address these limitations. In addition, while the present findings described how arrest patterns might operate, they did little to explain why. The results produced here should be a basis for a theoretical expansion in this area of police behavioral research, leading to tests of theoretically derived research questions that consider sample subgroup differences. Final Report. 1984