From Arrest to Post Conviction: Matching Data Across Criminal Justice Agencies

• In Event: Data Issues in Criminal Justice and Criminology Research

Abstract

Data from the criminal justice system frequently offers a snapshot in time. Department of corrections data may have information on case sentencing, but may lack relevant court case information prior to disposition. Likewise, data from courts may omit important pre-filing information that is available in other agency data, such as police or public defender data. Only by combining information across agencies can we obtain more complete information on cases. We introduce a matching methodology built on machine learning techniques to accomplish this task and apply it using agency data from two states. The method takes two or more agency data sets as input and returns the same data with unique identifiers for both defendants and cases that are consistent across all data sets. This technique not only improves our understanding of how cases evolve as they are processed through the criminal justice system, but also allows us to develop new metrics that evaluate criminal justice practices. In this project, we estimate the percentage of non-violent felonies sentenced to prison across counties in a midwestern state using matched data. Implications for research and the field will be discussed.

Authors

• Mason DeLang, Measures for Justice

• Robert Hutchison, Measures for Justice

• Sema Taheri, Measures for Justice