The
goal of this chapter was to summarize some of the scholarship surrounding
the age-old question of “how large a sample is large enough?”
Unfortunately, the answer is not simple. When data contains a clear
factor structure, the results might be seen in smaller sample sizes.
When the factor structure is not clear, larger samples and subject
to item ratios provide more accurate results. These findings, however,
are of little practical use because the magnitude of item loadings
is often unknown before a project begins and not within the researcher’s
control. In addition, even when the data does have a clear factor
structure and a relatively large subject to item ratio (as was the
case with the SDQ data), the error rate can exceed the standard alpha
level of .05. It is true that larger sample sizes and subject to item
ratios are better and more reliable than small, but there is still
some amount of error in our results! In general, though, we would
support Jöreskog and Sörbom’s (1996) recommendation
of at least 10 participants per parameter estimated; but we would
encourage researchers to try to get closer to 20 participants per
parameter.
The controversy around
sample size again reinforces the point that EFA is exploratory.
It should be used only for exploring data, not hypothesis or theory
testing, nor is it suited to “validation” of instruments.
We have seen many cases where researchers used EFA when they should
have used confirmatory factor analysis. After an instrument has been
developed using EFA and other techniques, it is time to move to confirmatory
factor analysis to answer questions such as “does an instrument
have the same structure across certain population subgroups?”
Based on the data presented in this chapter, we think it is safe to
conclude that researchers using large samples and making informed
choices from the options available for data analysis are the ones
most likely to accomplish their goal: to come to conclusions that
will generalize beyond a particular sample to either another sample
or to the population (or a population) of interest. To do less is
to arrive at conclusions that are unlikely to be of any use or interest
beyond that sample and that analysis.