<![CDATA[The Small Area Estimations (SAE) method is used as a reliable approach in overcoming sample insufficiency problem in sample surveys. BPS produces small area statistics using popular SAE methods such as Empirical Best Linear Unbiased Prediction in Fay-Herriot (EBLUP-FH) model. The EBLUP-FH method is a parametric approach that requires the assumption of normality and is free from outliers on the random effect. However, it is difficult to satisfy because real data often behave in extreme ways. The SAE M-quantile Chambers-Dunstan (CD) method relaxes parametric assumptions and is robust to outliers. This study examines M-quantile CD method in increasing robustness of small area estimation through its application to real data in estimating average household expenditure per capita at sub-district level in DI Yogyakarta 2018. This study uses Susenas and Podes data in 2018. The result shows that M-quantile CD succeeds in improving the precision of EBLUP-FH. By implementing M-quantile CD, it is expected that the estimation of extreme data is more accurate for local area policymaking.]]>
Copyrights © 2021
