A NOTE ON MULTIPLE DISCRIMINANT ANALYSIS

Although the discussion of the equations for DA was applied to the case of two groups, it should be emphasized that it was said earlier that the same equations apply to DA with any number of groups. With more than two groups, more than one discriminant function is calculated. The number of discriminant functions that can be calculated is equal to the number of groups minus one or to the number of dependent variables, whichever is smaller. Thus, with three groups, for example, only two discriminant functions can be calculated, regardless of the number of dependent variables. If, on the other hand, six groups but only three dependent variables are present, the number of discriminant functions that can be calculated is three (the number of the dependent variables ).

In the beginning of this chapter, it was mentioned that for the case of two groups, DA can be calculated by multiple regression analysis in which the groups are represented by a coded vector. With more than two groups, it is necessary to use more than one coded vector. Under such circumstances, multiple regression analysis cannot be used; instead, a canonical analysis with coded vectors may be used to calculate DA for any number of groups.