A Cautionary Note on the Use of Factor Analysis:

The present paper reanalyzes Borgatta, Kercher, and Stull's data, outlines the difference between factors and components, and reviews several types of factor indeterminancy.

Abstract

A statement by Wilkinson (1986) that “principal component and common factor solutions for real data rarely differ enough to matter” induced Borgatta, Kercher, and Stull (1986) to construct a hypothetical dataset which appeared to contradict Wilkinson's claim. In a later issue of the same journal, Hubbard and Allen (1987) followed with counter-examples to Wilkinson's statement based on real data. Both analyses in reply to Wilkinson were based on mathematical artifacts. The present paper reanalyzes their data, outlines the difference between factors and components, and reviews several types of factor indeterminancy.