Discriminant Function Analysis Results - Advanced Tab

Click this button to produce a spreadsheet with summary statistics for all variables currently in the model:

Wilks' Lambda. The Wilks' Lambda for the overall model that will result after removing the respective variable. Remember that Wilks' Lambda can assume values in the range of 0 (perfect discrimination) to 1 (no discrimination).

Partial Lambda. The Wilks' Lambda associated with the unique contribution of the respective variable to the discriminatory power of the model.

The F-value associated with the respective partial Wilks' Lambda.

The p-value associated with the respective F to remove. Note that in stepwise analyses, these p-values should be interpreted with caution; refer to the Overviews.

The tolerance value of a variable is computed as 1 - R-square of the respective variable with all other variables in the model (in this spreadsheet). Thus, the tolerance is a measure of the respective variable's redundancy. For example, a tolerance value of .10 means that the variable is 90% redundant with the other variables in the model.

The R-square value of the respective variable with all other variables in the model (in this spreadsheet).

Click this button to produce a spreadsheet with the summary statistics for all variables that are currently not in the model. The statistics displayed in this spreadsheet are essentially identical to those described in Variables in the model. Note that this option is only available if either the Forward stepwise or Backward stepwise method is selected on the

Model Definition

dialog box

- Quick

tab.

Click this button to produce a spreadsheet with the squared Mahalanobis distances between the group centroids. The Mahalanobis distance is similar to the standard Euclidean distance measure, except that it takes into account except that it takes into account the correlations between variables. The larger the differences in this spreadsheet, the farther are the respective groups apart from each other, and the more discriminatory power does our current model possess for discriminating between the respective two groups.

Together with the spreadsheet of Mahalanobis distances, two other spreadsheets will be produced: one with the F-values associated with the respective distances and another with the respective p-values. Again, those p-values should be interpreted with caution, unless you bring to the analysis strong a priori hypotheses concerning which pairs of groups should show particularly large (and significant) distances.

Click this button to display the Canonical Analysis dialog box. Note that this option is only available if at least three groups were specified for the analysis, and if there are at least two variables in the model. In that case, Statistica will perform a complete canonical analysis, and compute the discriminant functions.

Click this button to produce a spreadsheet that summarizes for each step:

1. which variable was entered or removed (indicated by the letters E or R in parentheses after the respective variable name in the first column of the spreadsheet);
2. the step number;
3. the F to enter or remove;
4. the respective degrees of freedom for that F-value;
5. the p-value for that F-value;
6. the number of variables in the model after the respective step;
7. the overall value of Wilk's Lambda after the respective step;
8. the F-value associated with the Lambda;
9. the degrees of freedom for that F-value;
10. the p-value for that F-value.