Desirability Profiling in GLM, GRM, and MANOVA

Select the Profiler tab of the GLM Results, GRM Results, or the ANOVA Results dialogs to access the options described here. The options for desirability profiling are very similar to those offered in the Experimental Design module, and described there in the Desirability topic. Also, the options available on this tab are essentially the same as those offered on the Response Profiles dialog in the Experimental Design module, with the following exception to accommodate the general nature of GLM:

• For categorical factors, the grid points for the desirability profiles can only be set at the observed (and selected) factor settings; for example, if a factor Gender has two levels (Male, Female), then the grid can only be set at those two levels for the factor.
• When referencing the factor levels of categorical predictors in the Specifications for Factor Grid dialog, use consecutive integer values (i.e., 1, 2, 3, ...), regardless of the actual factor codes that were used in the data to identify the levels of those factors (e.g., do not use 101, 110, -99).
• Because for categorical predictors only the observed (and selected) levels are available for the grid points, the Use general function optimization method of finding the optimum combinations of factor levels, available on the Desirability Profiler Options dialog in Experimental Design, is not available on the GLM Desirability Profiler Options dialog; when you select the Optimum option button, STATISTICA will instead always perform a search over the user-defined grid of factor values/levels.
• If the design contains repeated measures factors, the desirability profiler will ignore the within-subject (repeated measures) design; in other words, STATISTICA will treat the levels of the repeated measures factors as separate dependent variables.
• This tab is not available if there are mixture variables in the model. If the current design is a standard mixture design, you may want to use the Experimental Design module to analyze the data, and to perform the desirability profiling analysis. That module has facilities to produce triangular graphs, and designated options to find the best factor settings, given the desirability function, within the constrained (mixture) experimental region; these options are not supported in GLM.

In all other respects, the functionality of these options available on this tab is identical to that described for the Desirability Profiler Options dialog in Experimental Design.

See also GLM - Index.