643 – Small Sample Property and Statistical Inference
Bias-Corrected Estimators of Scalar Skew Normal
Guoyi Zhang
University of New Mexico
Rong Liu
University of Toledo
One problem of skew normal model is the difficulty in estimating the shape parameter, for which the maximum likelihood estimate may be infinite when sample size is moderate. The existing estimators suffer from large bias even for moderate size samples. In this paper, we proposed five estimators of the shape parameter for a scalar skew normal model, either by bias correction method or by solving a modified score equation. Simulation studies show that except bootstrap estimator, the proposed estimators have smaller bias compared to those estimators in literature for small and moderate samples.