This paper gives an overview of the methods used to handle missing data in the 2004 National Postsecondary Student Aid Survey (NPSAS:04). More generally, this paper deals with the concept of mass imputation – or the process of simultaneously filling in large blocks of missingness in a data file. As surveys become larger and data sets get bigger, dealing with large amounts of missing data is quickly becoming an issue to data producers. Holes in the data, whether caused by unit or item nonresponse, are beginning to become particularly problematic. Another motivating factor is the requirement by some agencies to provide complete data sets. The statistical literature is replete with research on single variable imputation with very little on mass imputation whose underlying theory and guidelines have not been widely discussed, established, or understood.
