Multiple Imputation of Race for Project TALENT (2014)

This document describes a procedure for multiple imputation of race for all 377,015 students in the primary PT student datafile. The procedure incorporates not only information in this file, but also family information derived by AIR. It employs school-level racial composition information derived from this file, and also uses school-reported information in the General School Characteristics (GSC) datafile. Specifically, it employs the responses to questions GSC95–GSC99. As discussed in §3, the procedure correctly predicts race for more than 95% of the 1952 respondents in the 2012 Pilot Study.