Factor analysis versus PCA
These techniques are typically used
to analyze groups of correlated variables representing one or more common domains;
for example, indicators of socioeconomic status, job satisfaction, health, self-esteem,
political attitudes or family values. Principal components analysis is used to
find optimal ways of combining variables into a small number of subsets, while
factor analysis may be used to identify the structure underlying such variables
and to estimate scores to measure latent factors themselves. The main applications
of these techniques can be found in the analysis of multiple indicators, measurement
and validation of complex constructs, index and scale construction, and data reduction.
These approaches are particularly useful in situations where the dimensionality
of data and its structural composition are not well known.
When an investigator has a set of hypotheses that
form the conceptual basis for her/his factor analysis, the investigator performs
a confirmatory, or hypothesis testing, factor analysis. In
contrast, when there are no guiding hypotheses, when the question is simply
what are the underlying factors the investigator conducts an exploratory
factor analysis. The factors in factor analysis are conceptualized as "real
world" entities such as depression, anxiety, and disturbed thought. This is
in contrast to principal components analysis (PCA), where the components are
simply geometrical abstractions that may not map easily onto real world phenomena.
difference between the two approaches has to do with the variance that is analyzed.
In PCA, all of the observed variance is analyzed, while in factor analysis it
is only the shared variances that is analyzed.
- The SSI homepage for LISREL 8. LISREL is the pre-eminent statistical package
for use in structural equation modeling. This program has been around for
many years, and it can be thought of as the industry standard. Based on my
limited experience, I would recommend LISREL over SAS. I find that I have
much more control over the model than I do in SAS. This page provides links
to FAQ's for LISREL, PRELIS, and SIMPLIS, as well as examples, references,
Return to the Statistics
Last updated 19 June 2010