Forecasting Domestic Violence: A Machine Learning Approach to Help Inform Arraignment Decisions

Abstract

Arguably the most important decision at an arraignment is whether to release an offender until the date of his or her next scheduled court appearance. Under the Bail Reform Act of 1984, threats to public safety can be a key factor in that decision. Implicitly, a forecast of “future dangerousness” is required. In this article, we consider in particular whether usefully accurate forecasts of domestic violence can be obtained. We apply machine learning to data on over 28,000 arraignment cases from a major metropolitan area in which an offender faces domestic violence charges. One of three possible post‐arraignment outcomes is forecasted within two years: (1) a domestic violence arrest associated with a physical injury, (2) a domestic violence arrest not associated with a physical injury, and (3) no arrests for domestic violence. We incorporate asymmetric costs for different kinds of forecasting errors so that very strong statistical evidence is required before an offender is forecasted to be a good risk. When an out‐of‐sample forecast of no post‐arraignment domestic violence arrests within two years is made, it is correct about 90 percent of the time. Under current practice within the jurisdiction studied, approximately 20 percent of those released after an arraignment for domestic violence are arrested within two years for a new domestic violence offense. If magistrates used the methods we have developed and released only offenders forecasted not to be arrested for domestic violence within two years after an arraignment, as few as 10 percent might be arrested. The failure rate could be cut nearly in half. Over a typical 24‐month period in the jurisdiction studied, well over 2,000 post‐arraignment arrests for domestic violence perhaps could be averted.

Accuracy and Fairness for Juvenile Justice Risk Assessments

“…Arrests for violent crimes are rare and difficult to predict. The results from the first column of Table are roughly consistent with past machine‐learning efforts to forecast crimes that are very troublesome but relatively uncommon (Berk ; Berk et al ). In this application, considerable success forecasting the absence of an arrest for a violent crime is to be expected because the base rate for violent crime is low (i.e., 0.10).…”

A Misdirected Principle with a Catch: Explicability for AI

“…This example is in contrast with the examples used by those proposing a principle of explicability for AI. ML used for medical diagnosis (de Bruijne 2016;Dhar and Ranganathan 2015;Erickson et al 2017), judicial sentencing (Berk et al 2016;Barry-Jester et al 2015), and predictive policing (Ahmed 2018;Ensign et al 2017;Joh 2017) are just a few of many realworld examples. Using the decisions of ML algorithms in these contexts without explanation is wrong, so the argument goes, unless that ML algorithm is explicable.…”

Risk and punishment: The recent history and uncertain future of actuarial, algorithmic, and “evidence‐based” penal techniques

“…Not only are risk practices expanding, they are continuing to evolve. Some recent risk assessment instruments are beginning to incorporate machine learning (Berk, 2008;Berk, Sorenson, & Barnes, 2016), and there are discussions of incorporating dynamic data sets-where the instruments train on new incoming data in real time (Rothschild-Elyassi et al, 2019). Related to this, criminal justice institutions are increasingly working with computer scientists and software engineers trained in big data analytics to develop new ways of thinking about and assessing risk (Hannah-Moffat, 2018; see also Danaher, Hogan, & Noone, 2017).…”