A Bayesian Approach to Binary Logistic Regression Model with Application to OECD Data

Abstract

In spite of being a common method for estimating the model parameters, Maximum Likelihood (ML) method may give bias results for small sample sizes. To overcome this problem, Bayesian method is usually utilized to obtain the estimates of the model parameters as an alternative to the ML method. In this study, a real data set was analyzed by using the binary logistic regression model. Parameters of the binary logistic regression model were estimated by using ML and Bayesian methods. Modeling performance of the binary logistics regression model based on the Bayesian estimates was compared with the model based on the ML estimates. Well-known information criteria such as AIC and BIC were used in this comparison.