Bridging Disciplinary Boundaries (January 11 - 14, 2007)

PURPOSE OF THE STUDY. The police have actively intervened in domestic violence through pro-arrest policies since the 1980s, but research results of the effectiveness of arrest are inconclusive. Some insisted that more batterers have been arrested, but others refuted it. Some suggested that arrest has effectively reduced domestic violence, while others did not support it. Moreover, the majority of previous studies focused on the effect of arrest on individual batterers' recidivism. Considering that arrest has deterrent effects on the public (i.e., general effect) as well as arrested individuals (i.e., specific effect), it is necessary to examine the effectiveness of arrest in the general effect perspective. This study attempted to examine the general effect of arrest by using the National Crime Victimization Survey (NCVS) from 1987 to 2003.

METHOD. The NCVS annually gathers detailed crime data from a nationally representative sample of 50,000 households with 100,000 individuals. The target population of the study is the domestic violence victim age 18 and over. The general effect of arrest was examined by time series analysis using the ARIMA model. Two time series were calculated based on the NCVS; 204 arrest rates and 204 domestic violence incidence rates. First, univariate time series analysis was conducted to see if arrest rates have increased over time. Then, regression time series analysis was conducted to see if arrest rates have been related to incidence rates in temporal order.

RESULTS. First, overall trends of two time series were examined. Arrest rates have increased from 19% in 1987 to 35% in 2003. Also, domestic violence incidence rates at the same period fell from 5.6 to 2.9 per 1,000 women. Univariate time series analysis confirmed these results. Arrest rates were correctly identified with ARIMA (1, 1, 0) model and shown as having increased over time (β = .088, p < .001), while incidence rates, identified with ARIMA (0, 1, 1), have decreased (β =- .008, p < .001). Regression time series analysis showed that arrest rates were negatively related to incidence rates (β = -.026, p < .003), but the effect was realized eight months later. In other words, it has taken eight months for arrest rates to affect domestic violence incidence rates.

IMPLICATIONS. The study results showed that arrest rates have increased over time and substantially reduced domestic violence incidence rates. It suggests that arrest of batterers has a general effect on the public. Given that the effectiveness of arrest can be comprehensively understood by examining the general effect as well as the specific effect, it is necessary for future research to further investigate how to interpret the general effect delayed by eight months, and how to combine the results of general effect with those of specific effect to assess overall effectiveness of arrest. Also, the police needs to be encouraged to continue to actively intervene in domestic violence. Finally, social work practice needs to improve and strengthen cooperation with the criminal justice system to provide victims with broad legal options available and shown as effective.