Record Linkage of Criminal-Legal System Data: From Arrest to Post-Conviction Supervision

• In Event: Novel Uses and Approaches to Administrative Data

Abstract

Record linkage refers to matching data from multiple files record-by-record and enables the creation of linked data files richer than individual files in isolation. Record linkage can substantially enhance how we utilize the wealth of existing data to advance knowledge of the cumulative effects of criminal-legal system processing. However, record linkage can be challenging and is often error prone. In a recent study of racial disparities in criminal-legal system decisions and outcomes in Indiana, we conducted system-wide record linkage for seven years of police, jail, court, probation, prison, and parole data drawn from a single county. In this presentation, we discuss record linkage and post-linkage data analysis in the context of our study. Our study required linking records from several local and statewide data sources with varied recording practices. Linking identifiers were often non-unique or noisy, leading to incorrect or missed links. We discuss how linkage errors can influence the quality of the resulting linked data and should be considered to ensure valid data analysis downstream.

Authors

• Priyanjali Bukke, George Mason University

• Martin Slawski, George Mason University

• Evan Marie Lowder, George Mason University

• Hayley Carlisle, George Mason University

• Anna Knes, George Mason University

• Morgan Goins, George Mason University