Performa Restricted Maximum Likelihood and Maximum Likelihood Estimators on Small Area Estimation

Abstract

In this paper, it was studied performance of maximum likelihood and restricted maximum likelihood method to estimate random factor on small area estimation. To evaluate of their performances, it was used mean square error as criteria. Based on data simulation, it was found that mean square error of restricted maximum likelihood estimator smaller than ML. Therefore, it could be concluded that restricted maximum likelihood method is better than maximum likelihood. Hereafter, it was applied those methods to human development index in South Sulawesi. The result of study showed that restricted maximum likelihood method is also better than Ordinary Least Square or maximum likelihood. The results of the study on human development index data in South Sulawesi Province; Gross Regional Domestic Product is one of the components that significantly affect the human development index.