In the 1980s, accepted practices rose out of the widely accepted theory that “arrest works best” for reducing recidivism in misdemeanor domestic violence (DV) cases. However, a collection of studies supported by the National Institute of Justice revealed that mandatory arrest policies were, at best, producing mixed results and, at worst, not working at all. From this, police departments began looking for new strategies to combat DV. For some agencies this led to the creation of specialized units. Research on these units has historically been limited and mixed regarding success, resulting in this particular study attempting to examine a number of factors affecting recidivism. These factors include the creation of a DV unit which provided a multi-faceted community response to DV cases, which included enhanced investigation and victim assistance.
To provide a more intensive and multi-faceted approach to combating domestic violence, The Charlotte Police Department formed a specialized DV unit. This police unit investigates charges based on evidence, versus a common standard Patrol practice of taking victim’s wishes regarding charging into consideration. The DV Unit, along with various community resources, also provides victim assistance in the form of basic needs, referrals to services, and the safety plans.
Primary Research Question(s)
Does assignment to a specialized domestic violence (DV) unit affect the recidivism rate of domestic violence offenders?
One thousand cases were randomly chosen from the approximately 6,000 cases that were reported with the Charlotte Police Department. From here, these cases narrowed down to 891 single offender-single victim cases. Those assigned to the DV unit had been assigned by a DV sergeant based on severity and level of recurrence. Due to significant differences between the cases assigned to the DV unit and standard Patrol, propensity scoring was used to weigh the cases, so the two groups were statistically similar.
Data for the study came from electronic incident reports spanning the 18 to 30 months following initial incidents. To determine recidivism, additional domestic violence incidents against any victim were reviewed.
Researchers identified three limitations in particular.
Frequency of recidivism was limited to only three measurement benchmarks – 0, 1, or 2+. This impeded the researchers’ ability to identify any differences in the frequency rate between the two groups beyond reoffending one or more times.
Unlike previous studies, this study did not include victim interview, resulting in the possibility that recidivism may have occurred, but never resulted in an incident report.
This study was limited to the DV events in the jurisdiction of the Charlotte Police Department only, so any reoffending events outside of that jurisdiction were not counted.
Offenders whose cases went through the DV unit had a lower rate of recidivism than those managed through standard Patrol practices. However, if offenders did reoffend, the frequency (1 or 2+ additional events) rate was similar across both groups.
Utilizing a specialized DV unit shows promising results for reducing recidivism. The scope of this studied was limited to studying the effect of the DV unit as a whole, so there was a limited ability to identify the effectiveness of specific DV unit features. Four possibilities for the reduction in recidivism were identified:
Intensive investigation led to more case closures by arrest.
Offenders suspected of more serious crimes were routed to the DV unit; therefore increased incarceration terms could have reduced
Increases in victim assistance may have led to reducing target suitability. Although, a separate study showed that the DV unit did not decrease re-victimization.
Charging based on the evidence regardless of victim’s wishes may have increased procedural justice, i.e. perceived fairness, in the eyes of the offender.
Exum, M. L., Hartman, J. L., Friday, P. C., & Lord, V. B. (2010). Policing domestic violence in the post-SARP era: The impact of a domestic violence police unit. Crime & Delinquency, 60(7), 999-1032.)