We recommended that the following simple rule always be applied to avoid
collinearity in defining a dummy variable for regression analysis; if the nominal
independent variable of interest has k categories, then one must define exactly k-1 dummy variables to index these categories, provided that the regression model contains a constan term (i.e.,intercept B0). If the regression model does not contain an intercept, then k dummy variables are needed to index the k categories of interst. For example. If ther are k = 3 categories, the number of dummy variables should be k-1=2 for a model containing an intercept In applying this rule, the following should be noted: 1. If a intercept is used in the regression equation, proper deiniton of the k-1 dummy variales automatically indexes all the k categories 2. If k dummy variables are used to describe a nominal variable with k categories in a model containing an intercept, all the coefficients in the model cannot be uniquely estimated because collinearity is present 3. There are many different ways to properly define the k-1 dummy variables to index the k categories of a given nominal variable. For example, teo equivalent dummy variables in regression
A dummy, or indicator, variable is any variable in regression equation that takes on
a finite number of values so that different categories of nominal variable can be identified.