GLM, GRM, and ANOVA More Results - Custom Tests Tab

Select the Custom tests tab of the GLM More Results or the ANOVA More Results dialogs to access options to perform tests of selected effects with a custom error term. See the Examples section for how this option can be used to analyze designs where not only the observations but also treatment levels are randomized (e.g., in cross-over designs; see, for example, Milliken & Johnson, 1992, Section 32.2). If you selected the Lack of fit check box on the Quick Specs Dialog - Options tab or via the LACKOFFIT keyword in the GLM syntax, then you can also select the Pure error as the error term for the selected effects; the pure error is computed from the sums of squares within each unique combination of treatment levels (for categorical predictors) or values (for continuous predictors).

Effect(s) to test

Select one or more effect to test in the Effect(s) to test list. To select multiple effects, hold down the CTRL key and click on the respective effects; to select lists of consecutive effects, click the first effect and then hold down the SHIFT key when you click the last effect in the consecutive list.

Error term for effect(s)

Use the Error term for effect(s) drop-down list to select only one error term.

Compute tests

Click the Compute tests button to display the Custom multivariate test of significance spreadsheet.

Note: Within-subject (repeated measures) effects. If the design includes within-subject (repeated measures) factors, the resulting ANOVA or MANOVA table will also include the interactions of all within-subject (repeated measures) effects by the selected Effect(s) to test; the error term for those effects will be the appropriate M-transformation of the Error term for effect(s). See also the Introductory Overview for details.

See also GLM - Index.