1: At this time, there were also
extended arguments about whether it was even proper to compute factor
scores, due to something called “indeterminacy”—essentially
meaning that there are more unknowns than equations being estimated
in EFA (for an excellent overview of the issue, see Grice, 2001).
The same individuals could be ranked multiple ways, leaving their
relative ranking indeterminate. This is a concern that remains today.
[return]
2: Refined techniques are actually
a bit more complex, taking into account more information than just
factor loading—for example, the correlation between factors.
However, as the point of this chapter is to discourage the use of
factor scores, we will refrain from providing more detail.
[return]
3: See
the accompanying SAS syntax for this chapter, available on the book
website, for a step-by-step example of how SAS computes the factor
scores.
[return]
4:Keep in mind that others might
use the pattern matrix loadings instead of the standardized scoring
coefficients when estimating factor scores. See DiStefano et al. (2009)
or (Grice, 2001) for more information about these methods.
[return]
5: Of course, this sample size is
entirely too small, but 10:1 ratio is about average for most EFAs
reported in the literature.
[return]