Dummy Variables

Abstract Categorical explanatory variables can be included in models via the use of dummy variables. In a regression model with an intercept, K  − 1 linearly independent dummy variables must be defined to represent a K level categorical variable. Indicator variables are commonly used to define a set of dummy variables. In a regression model with an intercept, indicators are derived for K  − 1 of the K possible levels and the remaining level is the referent or baseline category. Overall tests regarding the categorical variable will not depend on the choice of dummy variable definitions. However, the interpretation of the coefficient for each dummy variable will depend on the coding used.