Shrinkage Estimation of Proportion via Logit Penalty

Abstract

By releasing the unbiasedness condition, we often obtain more accurate estimators due to the bias-variance tradeoff. In this paper, we propose a class of shrinkage proportion estimators which show improved performance over the sample proportion. We provide the “optimal” amount of shrinkage. The advantage of the proposed estimators is given theoretically as well as explored empirically by simulation studies and real data analyses.