Efficiency of Adaptive Methods Using Simulated Alpha Skew Normal Two-Stage Data

Abstract

Two-stage sampling improves variances for estimators of means and regression coefficients because of intra-class homogeneity. To choose the appropriate way of allowing for clustering using sample data, an adaptive method will be evaluated in this paper based on testing the null hypothesis that the variance component of the random effect is zero. Rejecting the null hypothesis, clustering will be
allowed for in variance estimation; otherwise clustering will be ignored. The data will be simulated from alpha-skew normal distribution with different values of the parameter.

Key words and phrases. Primary Sampling Units, Two-stage Sampling, Linear Mixed Model, Alpha Skew Normal Distribution, Balanced Data, Huber-White estimator.

2000 Mathematics Subject Classification. 40H05, 46A45