Principal Components and Factor Analysis

Click the Pairwise tab of the Review Descriptive Statistics dialog box to access options to review summary statistics for the pairwise deletion of missing data. This tab is only available if a raw data file is being analyzed; the options are only available if the Pairwise option button is selected in the Factor Analysis dialog box and if this dialog box was called immediately after reading the data (these various matrices are no longer needed in the actual factor analysis computations, and are discarded when the factor extraction process begins). Note that in pairwise deletion of missing data, the different correlations in the correlation matrix can be based on unequal numbers of (different) cases. Thus, for each correlation coefficient there can be:

A unique n (valid number of cases);

Unique means for the two respective variables;

Unique standard deviations for the two variables.

If large discrepancies are evident for the values for different correlations (pairs of variables), the different correlations may have been computed from different actual cases. Before reaching final conclusions from the factor analysis, it is advisable to repeat the analysis either with casewise deletion of missing data or mean substitution.

Pairwise means. Click the Pairwise means button to produce a spreadsheet with the pairwise means. The spreadsheet will show the means for the row variables when they are paired (correlated) with the respective column variables.

Pairwise standard deviations. Click the Pairwise standard deviations button to produce a spreadsheet with the pairwise standard deviations. The spreadsheet will show the standard deviations for the row variables when they are paired (correlated) with the respective column variables.

Pairwise N. Click the Pairwise N button to produce a spreadsheet with the pairwise n (valid number of cases).