58
Information Descriptions were removed from the summary variable. Information Descriptions were removed from the Prosocial Behavior variable because they reflect neutral statements and are not necessarily "prosocial." Statistically significant differences were not found on father Total Commands or child Prosocial Behavior (with or without Information Descriptions).
A discriminant function analysis was performed on the summary variables to
evaluate the extent to which DPICS II variables could be used to distinguish between the clinic and non-clinic samples. The cross-validation procedure, which helps to diminish the optimistic bias, was used. In this procedure, each case is classified into a group according to the classification functions computed from all the data except the case being classified.
The seven DPICS II summary variables of child Compliance, child Prosocial Behavior, child Inappropriate Behavior, father Total Commands, father Prosocial Behavior, father Direct Command Ratio, and father Inappropriate Behavior were entered together to yield the following results: Wilk's Lambda = .752 (p < .05), Standardized Canonical Discriminant Function Coefficients: child Compliance = .787, child ProsocialRevised = -.129, child Inappropriate Behavior = .268, father Total Command = .127, father Prosocial Behavior-Revised =.544, father Direct Command Ratio = -.488, father Inappropriate Behavior = -.287, n = 60. When the combination of these variables was used, 68% of the original grouped cases were correctly classified. When the crossvalidation procedure was applied, 60% of the grouped cases were correctly classified. Forty percent of cases from each group were misclassified. The classification of each case is presented in Table 26 of Appendix A.