Weighting and imputation comparison in small area estimation

Abstract

In this paper, different methods of nonresponse adjustment for the totals of small area domains are examined. To improve quality of estimations linear model with random parameters at domain level is used. The empirical results are based on Monte Carlo simulations with repeated samples drawn from a finite population constructed from the Lithuanian survey on short-term statistics on service.

Keywords

• small area
• response probability
• imputation
• model