| Sumario: | This article explores the use of Bayesian Improved First Name and Surname Geocoding (BIFSG) to estimate race and ethnicity in data that lack this information. The authors propose a new method to estimate the uncertainty associated with using BIFSG estimates in analyzing disparities. They apply this method to estimate differences in earned income tax credit (EITC) dollars claimed among various racial and ethnic groups in the United States. The article also discusses the challenges and sources of uncertainty in BIFSG estimates, including sampling error, measures of race and ethnicity, clustering, and covariance between errors in predicted race and ethnicity. The text suggests a bootstrap procedure to incorporate this uncertainty when estimating differences in outcomes across racial and ethnic groups, which produces more conservative standard errors and confidence intervals. However, it does not address potential statistical bias introduced by BIFSG.
|
|---|