Model Launch
Factoring method - the method for extracting factors.
The Principal Components method is a computationally efficient method, but it does not allow for hypothesis testing.
The Maximum Likelihood method has desirable properties and enables you to test hypotheses about the number of common factors.
Note: The Maximum Likelihood method requires a positive definite correlation matrix. If your correlation matrix is not positive definite, select the Principal Components method.
Prior Communality - the method for estimating the proportion of variance contributed by common factors for each variable.
Principal Components (diagonals = 1) sets all communalities equal to 1, indicating that 100% of each variable’s variance is shared with the other variables. Using this option with Factoring Method set to Principal Components results in principal component analysis.
Common Factor Analysis (diagonals = SMC) sets the communalities equal to squared multiple correlation (SMC) coefficients. For a given variable, the SMC is the RSquare for a regression of that variable on all other variables.
The Number of factors (or principal components) determined by eigenvalues greater than or equal to 1.0 or from the scree plot where the graph begins to level out.
Note: Alternatively, the Kaiser criterion retains those factors with eigenvalues greater than 1.0. In our example, only factor 1 would be retained for analysis.
The Rotation method to align the factor directions with the original variables for ease of interpretation. The default value is Varimax. See Rotation Methods for a description of the available selections.
Click Go to generate the Factor Analysis report.
Depending on the selected Variance Scaling, the appropriate factor analysis results appear. See Factor Analysis Model Fit Options for details about the contents of the report. The Factor Analysis on Correlations and Factor Analysis on Unscaled reports show the same information.

Help created on 9/19/2017