Modeling and hypothesis testing for the factors affecting infant’s diarrhea using Generalized Poisson Regression

Abstract

Infants are weak individuals. Number of infants with Diarrhea is count data. Count data can be modeled using Poisson Regression. Poisson Regression has assumption that must be met. In the real case, overdispersion or underdispersion often occurs in data. This condition causes Poisson Regression cannot be used to model the data. Another alternative used to model the data with violation of assumption in Poisson Regression is Generalized Poisson Regression. This article will estimate the parameters of Generalized Poisson Regression using Generalized Poisson Regression. After getting the estimate parameters, parameters hypothesis testing simultaneously is done using Maximum Likelihood Ratio Test. There are three independent variables. They are percentage of infants who get exclusive breastfeeding, percentage of infants who get complete basic immunization, and percentage of households who have healthy living behavior. Significant parameter used to build the model. So, model for the factors affecting Diarrhea in infants in Pasuruan Regency is a model consisting complete basic immunization and healthy living behavior in the model.