The working paper offers several improvements for the imputation of variable values. Using data from the 1999-2000 National Postsecondary Student Aid Study (NPSAS), the paper describes the procedures and results for resolving discrepancies in variable values obtained from several data sources; imputing missing values on variables with low levels of missing data; imputing values for derived variables of great policy importance; and variance estimates for imputed variable values. Identified improvements will be applied to the NPSAS:2004 study.
Online Availability
