FACOM: A library for relating solutions obtained in exploratory factor analysis

Abstract

A library for comparingfactorial solutions obtained in exploratory analyses is presented. The procedure aUows both orthogonal and oblique solutions to be compared by computing several indices. The library is controUed with interactive menus or programs.Over the years, factor analysts have developed a series of methods for comparing factorial solutions obtained in different exploratory analyses. These methods are useful in applied research, such as cross-cultural studies or evaluations offactors previously found in relation to new studies. In particular, two situations will be considered in this paper: (I) comparisons between factors obtained in the same study but with different procedures and (2) comparisons between factors obtained in two different studies.Confirmatory multigroup factor analysis (see, e.g., J6r-eskog, 1971, or Alwin & Jackson, 1981 is the most accurate method of factor comparison when different samples are used. However, hypothesis testing procedures are not appropriate to all situations. For example, a great deal of cross-cultural or survey research requires procedures based on exploratory or descriptive methodology. The methods implemented in the FACOM library will be useful for these cases. Also, apart from substantive research, FACOM can be helpful in comparative studies on what guidelines should be used with various indices of matching factors solutions (see, e.g., Barrett, 1986). Applicability of the FACOM Library and Comparison ProceduresThe FACOM library has been developed to enable factorial solutions to be compared in the following cases: (1) two factorial solutions obtained in the same sample and with the same variables but using different analysis procedures; (2) two orthogonal factorial solutions obtained using the same variables in two different samples; and (3) two oblique factorial solutions obtained using the same variables in two different samples.In the first case, the method applied consists of obtaining the correlation between the estimated factor scores for each pair of factors to be compared. This correlation can be directly computed without computing the factor scores (see Gorsuch, 1974, formula 13.2.3).In the second case, two alternative methods have been included in FACOM. The first compares two solutions ro-

NFACOM: A new program for relating solutions in exploratory factor analysis

“…In addition to these new methods, the old CANONI-CAL and SCORECOMP procedures are still available in NFACOM (see Lorenzo & Ferrando, 1996). However, according to ten Berge (1996), the method of Kaiser, Hunka, and Bianchini (1971) has a dubious theoretical basis, and has therefore been omitted in the new version,…”

“…

(Lorenzo & Ferrando, 1996) was developed for comparing exploratory factor analysis (EFA) solutions in a variety of situations, Since its original presentation, FACOM has undergone considerable modifications which greatly extend its possibilities: Many recently developed procedures have been added, but others, which have proved not to be theoretically correct, have been removed, The purpose of this announcement is to describe the features and facilities of the new FACOM (NFACOM),The new version of the program enables factors to be compared in the following situations, none of which were considered in the previous version:1, Orthogonal comparison versus oblique comparison.

…”