Overdispersion Data Analysis Using Gaussian Inverse Poisson Regression Model

Abstract

One of the violations of assumptions in regression models in the calculation data is the case of overdispersion, where the value of variance is greater than the mean of the response variable. To overcome cases of overdispersion, a regression model is formed by mixing the Poisson distribution with several other distributions. Poisson Inverse Gaussian (PIG) is one of the regression models formed by mixed models designed for overdispersion data. The purpose of this study is to analyze the variables that affect the infant mortality rate in East Java Province in 2019. The model parameter estimator used is the Maximum Likelihood Estimator (MLE). From the minimum AIC value achieved, it can be seen that the Gaussian Inverse Poisson regression model is better than the Poisson regression model.