Fitri Mudia Sari, Luh Putu Widya Adnyani, Anang Kurnia, Hendry Frananda
The Maximum Likelihood method is one of the most commonly used estimators in econometric spatial modeling because it provides the Best Linear Unbiased Estimation (BLUE). Still, it is weak that when the data has many spatial units, this method becomes biased in parameter estimation. Bayes method is an alternative solution to this problem. The Bayesian method is suitable for use in cases of spatial heteroscedasticity, namely, the error variance of the spatial model is not homogeneous. The two methods were applied to diarrhea data in Surabaya with several predictor variables, including population density (X1), number of health facilities (X2), number of households with access to proper sanitation facilities (X3), and number of drinking water facilities that meet the requirements (X4). The high cases of diarrhea in a nearby sub-district are thought to be influenced by the geographical location of an area. The adjacent sub-districts have several characteristics that are almost the same. Based on this, the possibility of diarrhea cases in Surabaya has a spatial effect. The results showed that the SAR model with the MLE approach was better than the Bayes approach. © 2022 American Institute of Physics Inc.. All rights reserved.
Statistics Department, Universitas Negeri Padang, Padang, Indonesia; Informatics Department, Indraprasta University, Jakarta, Indonesia; Department of Statistics, IPB University, Bogor, Indonesia; Geography Department, Universitas Negeri Padang, Padang, Indonesia
