0% found this document useful (0 votes)
283 views2 pages

Exploratory Factor Analysis: Richard L. Gorsuch

Download as pdf or txt
Download as pdf or txt
Download as pdf or txt
You are on page 1/ 2

CHAPTER 6

Exploratory Factor Analysis


Richard L. Gorsuch

1. Introduction to criteria for evaluating possible procedures


is discussed in Section 1.3.
1.1. The Task Section 2 contains discussions of each major
decision point in an exploratory factor
The purpose of this chapter is to provide an
analysis. The discussions examine the possible
overview of exploratory analysis to guide
c~urses of action for their advantages and
users in selecting appropriate procedures.
dIsadvantages. These include the views of
While it will occasionally be necessary to
several paradigms if the paradigmatic
present some new material to aid in selection
differences seem important for that decision.
of appropriate techniques, no new techniques
The whole often has characteristics which
are presented here. The emphasis will be on
are difficult to deduce from the sums of its
the selections to be made and how the results
par~s, and so the use of factor analysis for
may differ depending on those selections. The
typIcal problems occasionally has considera-
result, hopefully, is a guide to the selection
tions only implicit in discussing major factor
and interpretation of appropriate exploratory
analysis decisions. Therefore, Section 3
factor analyses for the researcher familiar with
summarizes four major purposes for which
basic factor-analytic procedures and termi-
factor analysis is often used. The special
nology.
features of each situation are examined and
Decisions regarding factor analysis are
implications for a factor analysis are noted.
partially a function of the perspective from
The result is four major designs for
which factor analysis is approached. In
exploratory factor analysis.
Section 1.2 the major paradigms by which
This chapter relies heavily on Gorsuch
factor analysis has been approached are
(1983). That reference contains extended
summarized. The relevance of the paradigms
discussions which provide detailed back-
ground for this discussion. Primary ref-
erences, for example, are not generally
Richard L. Go~uch .0 Graduate School of Psychology, repeat~d here if they can be found by
Fuller Theological Semtnary, Pasadena, California 91101. consultmg the appropriate pages of Gorsuch
231

J. R. Nesselroade et al. (eds.), Handbook of Multivariate Experimental Psychology


© Plenum Press, New York 1988
232 II • MULTIVARIATE MODELING AND DATA ANALYSIS

(1983). Neither are extensive literature problems from at least one or two paradig-
reviews nor detailed rationales repeated from matic stances.
that source. Instead the emphasis is on
providing guidance to the established rules for
1.2.1. Mathematical versus Scientific
uses of exploratory factor analysis and on
Paradigms
problem areas where decisions are not yet
clear. A major paradigm conflict in the area of
exploratory factor analysis is that of mathe-
matical versus scientific paradigms. In the
1.2. Paradigms in Factor Analysis
mathematical paradigm the stress is upon the
There are several competing paradigms in derivation and exact computation of all the
factor analysis, each of which has its procedures. Procedures which only give
adherents. The degree to which one feels that estimates are much less valued than proce-
one paradigm is "better" than another has dures which can be exactly calculated.
been crucial in relating to factor-analytic Nunnally's (1967, 1978) emphasis upon prin-
technique. These are paradigms in the true cipal components is in keeping with the
Kuhnian sense (Kuhn, 1962), and so they mathematical paradigm.
operate as much or more as philosophies The scientific paradigm is concerned with
guiding the person's approach to factor factor-analyzing data in order to build and test
analysis than they do as principles to guide scientific theories. As such it is more
decisions for a particular case. By making concerned about the congruence of the results
these paradigms explicit, it is hoped that with future data analyses than whether or not
the applicability of each will be more ap- the procedures can be derived or calculated
parent. directly. Hence, estimates which are consis-
With explicit recognition of competing tently given across several sets of data may be
paradigms, it may be possible to move toward more highly valued than an exact computation
a unifying paradigm of which the several which does not generalize as well. And from
techniques are special cases for special the scientific paradigm, the mathematical
circumstances. If the several paradigms were differences between any two possible statisti-
united together into one overarching para- cal procedures are of little importance
digm, the resolution of questions concerning compared to the question of which procedures
which exploratory factor analysis technique to actually produce different scientific conclu-
use would be more empirical than philo- sions. Hence, if two mathematically different
sophical. procedures lead to the same scientific
The use of particular, paradigms will also conclusions, then the scientific paradigm
lead to suggestions for criteria by which considers the mathematical difference be-
current problems of exploratory factor analy- tween those procedures to be trivial. Cattell
sis may be resolved. As Kuhn (1962) notes, (e.g., 1946, 1978) is an example of approach-
conflicting paradigms seriously handicap res- ing factor analysis using the scientific
olution of problems. Each paradigm, of paradigm.
course, has its own criteria and so there is no While the mathematical and scientific
general resolution since each of the different paradigms are in agreement on many issues,
criteria leads to different conclusions. By there are some issues, such as that of factor
emphasizing the differences in paradigms and scores, where the two paradigms diverge.
their approaches to problems of exploratory Then users must decide which paradigm is
factor analysis, it may be possible to most important for their purposes and make
encourage progress toward resolution of the their decisions accordingly.

You might also like