Small Area Estimation for Skewed Data in Presence of Zeros

Abstract

The standard approach for small area estimation (SAE) based on linear mixed models often yields inefficient estimates for skewed data. Chandra and Chambers (2011a) described SAE for skewed data that are linear following a log-log transformation. In this case, they extended the model-based direct estimation (MBDE) approach of SAE (Chandra and Chambers, 2009). In particular, they derived sample weights via model calibration (Wu and Sitter, 2001) based on a log-log transform model with random area effects. We discuss SAE for skewed data in presence of zeros. In this context, we extend the Chandra and Chambers (2011a) methods of SAE under a mixture model. Results generated by simulation studies show that the method works well and produces an efficient set of small area estimates.