Variance Components and Mixed Model ANOVA/ANCOVA Results - Estimation Tab

Select the Estimation tab of the Variance Components and Mixed Model ANOVA/ANCOVA Results dialog to access the options described here.

Descriptive statistics. Click the Descriptive statistics button to display a spreadsheet with descriptive statistics for the groups that define the effects in the analysis. Included in the spreadsheet are group means, standard deviations, and confidence intervals for the means. Note that weighted means are displayed in this spreadsheet for two or more combined groups with unequal Ns. These are estimates of the weighted marginal means (weighted by the respective sample sizes). The standard errors and confidence intervals that are reported for the means are based on the standard deviation computed for all cases involved in the respective mean.

Marginal means. Click the Marginal means button to display a spreadsheet with marginal means for the groups that define the effects in the analysis. Note that unweighted means are displayed in this spreadsheet for two or more combined groups with unequal N's. These means are called the population marginal means for both the balanced case and the unbalanced case. The standard errors and confidence intervals that are reported for marginal means are based on the mean square error from the current analysis (see Milliken & Johnson, 1992, p. 132-134).

SSCP matrix. Click the SSCP matrix button to display three spreadsheets containing information about the relationships among the variables in the analysis. The first spreadsheet displays the correlations among all variables in the analysis. For fixed and random grouping factors, correlations are shown for each vector in the design matrix for the factors. The second spreadsheet displays the same information in the form of sums of squares and cross products. The third spreadsheet displays the generalized inverse of the sums of squares and cross products matrix for each vector in the design matrix for the factors.

SS & MS Effect. Click the SS & MS Effect button to display a Sums of Squares and Mean Squares spreadsheet which displays information relevant to the ANOVA for the effects in the model. For each dependent variable, the spreadsheet displays whether each effect is a Fixed effect or a Random effect, the Degrees of freedom for the effect, and the Sum of squares and Mean squares accounted for by the effect for each dependent variable. This button is only available if Expected MS (mean squares) has been selected as the Method for estimating variance components on the Advanced tab.

Report all results. Click the Report all results button to send the output produced by clicking the Summary: Components of variance button on the Quick tab to a report. If your output is not set to a report option, when the Report all results button is clicked, you can display the Analysis/Graph Output Manager dialog by clicking Yes in response to the "Would you like to modify your output options?" prompt.

Estimation. The options in the Estimation group box allow you to control the sensitivity of the computations involved in estimating variance components.

Sweep delta: 1.E-. Enter the value in the Sweep delta: 1-E- field to control the sensitivity of a variety of calculations performed by sweep operations, such as computing the generalized inverse of the design matrix and computing the sums of squares for an effect. The displayed value is the exponent to the negative power of 10 (in scientific notation) used to determine the minimum value to be considered nonzero. Setting the value in the Sweep delta: 1-E- field to a small value (e.g., 2) makes the calculations less sensitive, and setting the value in the edit field to a large value (e.g., 12) makes the calculations more sensitive.

Inverse delta: 1.E-. Enter the value in the Inverse delta: 1.E- field to control the sensitivity of the calculations performed in matrix inversion. The displayed value is the exponent to the negative power of 10 (in scientific notation) used to determine the minimum value to be considered nonzero. Setting the value in the Inverse delta: 1.E- field to a small value (e.g., 2) makes the calculations less sensitive, and setting the value in the edit field to a large value (e.g., 12) makes the calculations more sensitive.