Modeling Categorical Factors with Two Levels

Abstract

Often we need to consider the effects of categorical information in a regression model. Inclusion of categorical variables allow us to model the differences between two (today) or more groups (Class 7). When the categorical variable has two levels, the group membership is easily coded in a special numerical variable, known as a dummy variable. Since these dummy variables are numerical representations of qualitative, not quantitative, information, the coefficients of dummy variables require special interpretation. The testing methods of Class 5 are valuable tools for comparisons among models using qualitative factors.