RG-T3272 : Development of Predictive Model to Build Pilot Risk Tool to Identify Victims of Domestic Violence

Domestic Violence (DV) cannot be deterred by routine patrol in crime "hot spots." The most serious forms of DV  those that involve serious injury or even death  occur at a tragically high rate in an absolute sense, but account for a small share of the literally hundreds of thousands of DV events that occur in a given jurisdiction each year. Identifying the victims who are truly at greatest risk from among many victims who appear, a priori, to be at risk, is like finding a needle in a haystack for the law enforcement. The objective consists in building a world-class prediction model that can assist stakeholders like law enforcement agencies or victim service providers in doing exactly this. Building upon the experience of the Chicago Crime Lab (CCL), the research team would engage with 2-3 jurisdictions in Latin America (cities or central governments) in order to build a state-of-the-art risk tool to identify victims of domestic violence who are at greatest risk of experiencing repeat victimization. This exercise will produce two outputs: 1) a probability of repeat victimization for each victim known to law enforcement on the basis of that victim's prior contact with law enforcement and other public agencies and 2) a list of predictors that are most highly correlated with repeat victimization. By efficiently extracting the true "signals" from the large amount of available data, this tool will be helpful to law enforcement in targeting their most promising victim- or offender-based intervention strategies.