ESTIMATION OF INFANT MORTALITY RATES IN INDONESIA BY USING EMPIRICAL BEST LINEAR UNBIASED PREDICTION

Abstract

Infant Mortality Rate (IMR) is one of the many indicators that can measure the health status of a population in an area. IMR is also part of the third Sustainable Development Goals (SDGs), namely to ensure healthy lives and promote well-being for all of all ages. IMR was produced with direct estimation from the Indonesian Demographics Health Survey (IDHS). However, the result of the 2017 IDHS publication indicated that several direct estimations of IMR in 34 provinces in Indonesia had high relative standard error (RSE) values. Accurate data (from the RSE value) is needed for policy making. Therefore, this paper focused on small area estimation (SAE) by using the empirical best linear unbiased prediction (EBLUP) method and estimated IMR to the provincial level. SAE works by using the strength of several variables from the village potential data (Potensi Desa) which correlates strongly with IMR. The results of the analysis with the RSE used as a measure of model accuracy showed that by using the SAE EBLUP method in the IDHS data, an average RSE value of 15.23% was obtained, which is smaller than the direct estimate of the average RSE value of 29.51%. This research paper concludes that SAE using the EBLUP method is good for estimating the Provincial level IMR value in Indonesia in 2017.