Investigation of an Ordered Logistic Model for Consumer Debt

Abstract

In many situations surveys are carried out with the aim of determining how an important dependent variable is related to characteristics of individuals, perhaps with the intention of predicting, on the basis of their characteristics, the value of the dependent variable for new individuals. For example, major utility companies would like an easy way of predicting which customers are likely to become bad debtors. Frequently such surveys are analysed using multiple linear regression models, even when the dependent variable is categorical and cannot be expected to comply with the distributional assumptions of a multiple regession model.