In Chapter 5 and subsequent sections of this book we have discussed the analytic utility of thinking of independent variables as members of a smaller number of sets. These sets, each of which may have one or more members, may represent a distinct role in the research, such as a set of potential confounders (common causes of independent and dependent variables) or control variables, whose central role is to rule out certain alternative reasons for a relationship between Y and the IYs of interest. Alternatively, they may represent the multiple facets or aspects of a research construct, as, for example demographic factors , or treatment features. Or a set may include the g-1 variables needed to represent the g groups in a categorical variable . Or a set may include curvilinear or interactive aspects of one or more variables.