Introduction to Structural Equation Modeling with Latent Variables

Comparison of the CALIS and FACTOR Procedures for Exploratory Factor Analysis

Both the CALIS and the FACTOR procedures can fit exploratory factor models. However, there are several notable differences:

  • By default, PROC FACTOR analyzes the correlation matrix, while PROC CALIS analyzes the covariance matrix.

  • PROC FACTOR and PROC CALIS use different parameterizations in the initial factor solution. PROC CALIS uses a lower triangle pattern on the factor loading matrix (a confirmatory factor pattern) in the initial unrotated solution, while PROC FACTOR use certain matrix constraints in the initial unrotated solution. All other things being equal, PROC CALIS and PROC FACTOR might give the same solution after the same factor rotation.

  • Because of the way it parameterizes, PROC FACTOR is usually more efficient computationally. PROC CALIS uses a more general algorithm that might not be computationally optimal for exploratory factor analysis.