GDA Models Wizard Effects in Design - Custom Design Tab

Select the Custom design tab of the GDA Models Wizard Effects in Design dialog to access options to select the specific effects for the selected variables that are to be included in the model. Note that categorical or continuous predictor variables that are not selected into the Effects in the design group box, but were selected via the Variables button on the GDA Models Wizard Effects in Design dialog, will automatically be dropped from the analysis.

Effects can be specified in the following manner:

• Select the Categorical and/or Continuous predictor variables in the Predictor Variables group box by clicking on them; select multiple variables by holding down the CTRL key while clicking on the respective variables; select a list of consecutive variables by dragging the mouse over the list.
• Click on the desired action in the Method group box.

The specified effects will automatically be listed in the Effects in the design group box. You can further edit the effects in the Effects in the design group box by using the following options:

Predictor variables.; Categorical, Continuous.

The Predictor variables group box contains the Categorical and/or Continuous predictor variables selected for the current analysis. Select the variables you want to include in the design. See also the Note of caution for models with categorical predictors, and other advanced techniques to learn about the possible limitations of including categorical predictors in discriminant analysis problems.

Method

The Method group box contains the available options for including the selected variables as effects in the design. Some options may only be available if the appropriate number and types of variables were chosen in the respective Predictor variables group box. For a discussion of various designs (e.g., surface, mixtures, etc), see Types of Analyses in the Introductory Overview.

Add

Click the Add button to add the main effects for the chosen categorical and continuous predictor to the list of Effects in the design.

Full cross

Click the Full cross button to create the respective interaction term for all selected variables (e.g., Gender*Group*Time). Note that this button is available only if two or more categorical and/or continuous predictor variables are selected.

Full factorial

Click the Full factorial button to create a full factorial design for the selected variables (e.g., Group Gender Time Group*Gender Group*Time Gender*Time Group*Gender*Time). Note that this button is only available if two or more categorical and/or continuous predictor variables are selected.

Surface

Click the Surface button to create a response surface design. If a categorical variable is selected, the design will be blocked (e.g., Block Degrees Degrees*Degrees Time Time*Time Degrees*Time; see also the Experimental Design Overview for details). That the variables involved in the surface design will be flagged with an {S} next to them in the Effects in the design group box. Note that this button is available only if two or more continuous predictor variables are selected; it is not available if more than one categorical predictor (blocking variable) is selected.

Fact. to deg.

Click the Fact. to deg. (Factorial to degree) button to create a factorial design (up to the degree specified in the Degree field) for the selected variables (e.g., Group Gender Time Group*Gender Group*Time Gender*Time). Note that this button is available only if two or more categorical and/or continuous predictor variables are selected.

Poly. to deg.

Click the Poly. to deg. (Polynomial to degree) button to create a polynomial design (up to the degree specified in the Degree field) for the selected variables (e.g., Degrees Time Degrees*Degrees Time*Time). Note that this button is available only if one or more continuous predictor variables are selected and no categorical predictor variable is selected.

Degree

Use the Degree field in conjunction with the Fact. to deg. and Poly.to deg. options; see above.

Effects in the design

The Effects in the design group box shows the effects currently in the design. These effects can further be edited via the following options:

Cross

Click the Cross button to cross two effects currently selected in the Effects in the design group box; e.g., if two effects A and B(C) are currently selected, then Cross will add the term A*B(C) to the effects in the design. If more or fewer than two effects are selected, this button is not available.

Remove

Click the Remove button to remove the selected effects from the Effects in the design group box.

Clear all

Click the Clear all button to remove (clear) all effects from the Effects in the design group box.

Note: Reordering effects. To reorder the effects in the Effects in the design group box, first select them (e.g., click on the effect or list of effects), then move the mouse cursor to the desired location in the list, and when the mouse cursor changes to the insertion bar , click the mouse button to move the selected effects.

For alternative ways of specifying designs in GDA, see Methods for Specifying Designs.