Statistical Models for Discrete Data

Abstract

The theory of the generalized linear model (GLM) represents a significant advancement on that oflinearregression,specificallyinthechoiceofprobabilitydistributionswherethelinearregression is based on normal distribution, as the linear model is extended to the natural exponential family. We introduce the generalized linear regression model for analyzing binary, nominal and ordinal logistic response. The aim of this research is to focus on models used if the response variable have more than two categories . We outline two approaches ,nominal logistic regression and ordinal logistic regression. To introduce the concept of assumption and to aid understanding of nominal and ordinal logistic regression is whether that if the response variable takes on the values of ordered categories or not, a number of methods of ordinal logistic regression are discussed and illustrated with examples using real data.