On synthetic and composite estimators for small area estimation under Lahiri - Midzuno sampling scheme

Abstract

This paper studies performance of synthetic ratio estimator and composite estimator, which is a weighted sum of direct and synthetic ratio estimators, under Lahiri – Midzuno (L-M) sampling scheme. The synthetic estimator under L-M scheme is unbiased and consistent if the assumption of synthetic estimator is satisfied. Further, this paper compares performance of the synthetic and composite estimators empirically under L-M and SRSWOR schemes for estimating crop acreage for small domains. The study shows that both the estimators perform better under L-M scheme as having comparatively smaller absolute relative biases and relative standard errors.