Collapse Variance Components

Click the Collapse variance components button on the Variance Estimation and Precision ResultsSummary tab to display the Collapse Variance Components dialog box, which contains a table listing all sources of variation in the model and provides an option for collapsing variance components.

Dependent variable. The response variable for the displayed variance components is listed in this field.

Effect. The Effect column lists the sources of variation (i.e., random effects) for the model. Estimated variance components are displayed in the same order in which they enter the model.

Variance Estimate. This field displays the variance estimate for each effect.

Comp DF. This field displays the degrees of freedom for each effect. When variance estimates are calculated using the ANOVA method, degrees of freedom are calculated using the Satterthwaite approximation. For REML estimates, the degrees of freedom for the variance components are computed as described in the Computational Details. When REML variance estimates are combined, the degrees of freedom are computed from the standard error for the combined variance estimate. See Computational Details for more information.

Collapse. To collapse two or more variance estimates, select the effects in the table and click the Collapse button. The Collapsed name dialog box will be displayed enabling you to name the collapsed variance estimate. This newly labeled and formed source of variation will be displayed as a new row in the table. The rows corresponding to the individual components that contributed to the sum will be deleted from the table. Note that you cannot collapse negative variance estimates. For more information on how variance estimates are combined, see the Computational Details.

Output. Click the Output button to create and display a spreadsheet containing the information displayed in the Collapse Variance Components dialog box.

Close. Click Close to return to the Variance Estimation and Precision Results dialog box.