• A WALD TEST FOR OVER DISPERSION IN ZERO-INFLATED POISSON REGRESSION MODEL

Abstract

We have considered the regression models to fit the count data especially in the field of Biometrical, Environmental, Social Sciences, and Economical areas. In the field of medical applications the count data with extra zeros are also common. The zero-inflated Poisson (ZIP) regression model is helpful to examine such data. Here we awareness on the use of ZIP model for analysis of count data including maximum likelihood estimation for regression coefficients using Fisher scoring method, compare between Poisson and ZIP models by various tests: likelihood ratio test, score test, chi-square test, test based on a confidence interval test and Cochran test. Model selection using Deviance method, AIC and BIC. We can find a Wald test for ZIP model in a single sample case for detecting zero-inflation in Poisson model and conduct a small simulation study in order to investigate sampling distribution of Wald test and power of Wald test. From our study we found that distribution can be used to detect the zero-inflation in counts.