Practical Remedies for Inaccurate Criminal Justice Race and Ethnicity Data

Reliable and accurate data is the bedrock for evidence-based criminal justice policy. When a sentencing commission or other agency evaluates fairness and uniformity, they need to draw on accurate race and ethnicity data. This data is often missing or inaccurate. That problem is particularly acute for Hispanic ethnicity data. This article summarizes both prospective and retrospective remedies for missing and inaccurate race and ethnicity data. Research on gathering race and ethnicity data from other domains offers important best practices for criminal justice agencies. Collecting better data prospectively requires considerable coordination, planning, and administrative resources. Retrospective remedies are a crucial tool to enable data analysis and policy making in the interim. This article presents an overview of surname analysis, geocoding, and machine learning, three strategies to impute missing or inaccurate race and ethnicity data. Each of these strategies has trade-offs in their accuracy, complexity, and resources needed, but any of them is far better than the status quo of poor-quality data.