Ordering Dually in Triangle (ORDIT) and Hotspot Detection in Generalized Linear Model for Poverty and Infant Health in East Java

Abstract

The objcctive of this research is to build a generalized linear model using ordinal response variable with some covariates. One of the covariates in the model is an indtcator variable hotspot as the results the hotspot detection, while the response variable is ordinal data as the result of ORDIT ranking method. The data in this research is about infant health and poverty in some districts of East Java; i.e. Blitar, Kediri, and Jember taken from 3 different infant health district levels. GLM is implemented for 200 villages from each district, so there are 600 villages (sub-districts) as unit observation for modeling. The modeling analysis gives results that number of farmer, families, hotspot, and district are statistically significant as the predictor to the poverty level of villages.