Personality Correlates of Innovative Buying Behavior: A Reply

Abstract

multiple group discriminant analysis the reverse is true: the criterion score is not entered into the data matrix, but category membership is [7, p. 89]. It is true that the coefficients of the two group discriminant function can be determined by means of multiple regression. That is, one could use an artificial criterion variable (composed of zeroes and ones) and compute the multiple regression weights of the predictors on the artificial criterion. The weights resulting from this would be proportional to those in a two-group discriminant function computed from the same data, because the two techniques are statistically identical in structure. It would not be appropriate, however, to handle the multiple-group case with multiple regression, since the criterion cannot be represented as a single variable.