Multivariate Analysis: Factor Analysis

Method Tab

You can use the Method tab to set options in the analysis. (See Figure 27.4.) Each UI control in the tab corresponds to an option in the FACTOR procedure. The Method tab contains the following controls:

Factoring method

specifies the method used to extract factors or specifies a component analysis. This corresponds to the METHOD= option in the PROC FACTOR statement.

Compute factors from

specifies whether the factors are computed for the correlation matrix or the covariance matrix. This corresponds to the COV option in the PROC PRINCOMP statement. Note: Some methods require a correlation matrix.

Number of Factors

The number of factors retained is determined by the minimum number that satisfies the next three criteria.

Maximum number: specifies how many factors to compute. This corresponds to the N= option in the PROC FACTOR statement. You can type into the Maximum number field; if you want five factors, you can enter 5 even though this is not an option on the list.
Proportion of variance: specifies the proportion of common variance in the retained factors. This value is in the range $(0,1]$ . The option corresponds to the PROPORTION= option in the PROC FACTOR statement.
Minimum eigenvalue: specifies the smallest eigenvalue for which a factor is retained. This corresponds to the MINEIGEN= option in the PROC FACTOR statement.

Prior estimates

specifies a method for computing prior communality estimates. This corresponds to the PRIORS= option in the PROC FACTOR statement. The default method for the principal factor method is to set all priors equal to 1. This results in a principal component analysis. If you want a principal factor analysis, you should select a different method for estimating the prior communalities, as illustrated in the section Example: Reduce Dimensionality through Common Factor Analysis.

Heywood Conditions

specifies how the factor analysis behaves if a communality is greater than 1. The section "Heywood Cases and Other Anomalies about Communality Estimates" in the documentation for the FACTOR procedure describes why this situation might occur.

Do not allow communalities greater than one: specifies that an analysis should stop processing if it encounters a communality greater than one.
Set any communality greater than one to one: specifies that an analysis should set any communality greater than one to one, and then continue. This corresponds to the HEYWOOD option in the PROC FACTOR statement.
Allow communalities greater than one: specifies that an analysis should allow any communality. This corresponds to the ULTRAHEYWOOD option in the PROC FACTOR statement.