Ethics and Empathy in Using Imputation to Disaggregate Data for Racial Equity: Recommendations and Standards Guide

• Published in: Policy File 2021
• Main Authors: Brown, Steven, d, LesLeigh D, Ashley, Shena
• Format: Report
• Language: English
• Subjects: Big Data; Ethics; Urban Institute

To identify, and subsequently address, racial disparities, changemakers need access to high-quality data disaggregated by race and ethnicity. In many important policy questions, data disaggregated by race and ethnicity are unavailable, and efforts to collect new, self-reported data to fill these gaps is costly, time-consuming, or impermissible. Imputation and other methods for appending or integrating different data sources are critical tools for filling these gaps. However, these methods do not typically require the input of the people whose data are being combined or augmented, creating ethical risks and a potential lack of empathy for people whose data are used in the process. This report explains what imputation is, why it is an important and needed tool for disaggregated data and race-conscious policy making, and how to approach it with ethics and empathy. It explores key questions that stakeholders should weigh when creating and using imputed race and ethnicity data, including whether imputation is the right approach for disaggregating data in a particular use case, who should be involved in the process (for example, researchers, community partners and representatives, and end users), and how to ethically apply imputation methods for racial equity analysis. We recommend that the field adopt standards on relevance, interpretability, coherence, accuracy, privacy, and institutional environment when using imputation to disaggregate data by race and ethnicity. Critically, we also recommend that stakeholders incorporate community-engaged methods into any imputation project, drawing on insights from impacted communities to inform the use, design, and application of imputation to create disaggregated data for their community. If researchers and data analysts ask themselves the questions posed here, and apply these standards to their own work, we believe that imputation can be an effective, ethical, and empathetic tool to address critical gaps in race and ethnicity data.