Relative Risk for Poverty in Kelantan –A Bayesian Approach

Abstract

Abstract Poverty eradication among poor household head becomes a significant concern. Previous research employed the traditional statistical method to model the poverty data. However, these traditional statistical methods do not consider the spatial elements of poverty data. This study compares the performance of Poisson log-linear Leroux Conditional Autoregressive (CAR) model with difference neighbourhood matrices. A Poisson Log-Linear Leroux Conditional Autoregressive model with different neighbourhood matrices was fitted to the poverty data for 66 districts in Kelantan for 2010. The results show that the performance of the model with the contiguity matrix was nearly similar to the Delaunay triangulation neighbourhood matrix in estimate poverty risk. The variables that are significantly associated with the poverty in Kelantan are the number of non-education, number of female household head and the average age of the household head.