Variance Estimation for Nearest Neighbor Imputation for U.S. Census Long Form Data
Jae Kwang Kim, Wayne A. Fuller, William R. Bell
KEY WORDS: Income estimation, Fractional imputation, Hot deck imputation, Nonresponse, Poverty estimation, Replication variance estimation
ABSTRACT
Variance estimation for estimators of state and county income and poverty characteristics derived from the Census 2000 long form are discussed. The variance estimator must account for (1) uncertainty due to imputation, and (2) raking to census population controls. An imputation procedure that imputes more than one value for each missing item using donors that are neighbors is described and the procedure using two nearest neighbors is considered in detail. The Kim and Fuller (2004) method for variance estimation under fractional hot deck imputation is adapted to this problem. Numerical results from the 2000 long form data are presented.
CITATION: Kim, Jae Kwang, Fuller, Wayne A., and Bell, William R.. (2008). Variance Estimation for Nearest Neighbor Imputation for U.S. Census Long Form Data. Statistical Research Division Research Report Series (Statistics #2008-13). U.S. Census Bureau. Available online at <http://www.census.gov/srd/papers/pdf/rrs2008-13.pdf>.