Kajian Overdispersi pada Regresi Poisson Berdasarkan Model Generalized Poisson

Abstract

Generalized Poisson (GP) model is one model that can be used to overcame overdispersion in Poisson data. Detection of overdispersion in simulation data used score test, SSR-LRT test, and Wald test. Simulation data was generated based on the GP distribution. Goodness of fit of GP models in this study were analyzed by using the Root Mean Square Error (RMSE) and Akaike Information Criteria (AIC). In addition to the simulation data , this study also used applied case study data which malnutrition data in West Lombok. The results showed that the applied case study used indicated overdispersion at 5% significance level and the percentage of families living in slum area have impact on the spread of malnutrition in West Lombok. The simulation data showed that power of test of score test is more better than the other test, so that score test is consider as the most appropriate test in detecting overdispersion.