Using NIBRS Data to Predict Likelihood of an Arrest on Single Offender Incidents

• In Event: NIBRS in Action: New Horizons for Data Analysis and Availability

Abstract

Using Wisconsin NIBRS data extracted from the state repository housed at the Wisconsin Department of Justice and accessible through the FBI’s Crime Data Explorer, this study looks at factors that predict whether known suspects in sex offense cases were arrested. Funded by the Joint Statistical Analysis Program, two different datasets were created: the first contains approximately 10,000 Wisconsin NIBRS incidents involving offenses of rape/sodomy/sexual assault with an object/sodomy and the NIBRS data elements that were included in those incidents, including characteristics of the crime, characteristics of the suspect/victim, and characteristics of the agency. The second dataset contains approximately 1,400 statutory rape incidents. Logistic regression was performed to determine the overall model fit and odds ratios of the crime, person, and agency characteristics that significantly predict whether the single offender was arrested. The study was designed to be replicable at the local, county, state, regional, and national levels to determine whether results for one geographic area are similar to others and benefits the criminal justice community by improving our understanding of the quality of NIBRS data and law enforcement arrest decision-making. Benefits, limitations, and opportunities for strengthening the predictive capabilities of NIBRS data will also be discussed.

Author

• Ashley K. Billig, Wisconsin Department of Justice, Bureau of Justice Information and Analysis