Conventional wisdom in the literature
and many texts advise researchers to use orthogonal rotation because
it produces more easily interpretable results, but this might be a
flawed argument. In the social sciences (and many other sciences),
we generally expect some correlation among factors, particularly scales
that reside within the same instrument or questionnaire (i.e., shared
method variance will generally produce nonzero correlations). In practice,
even when we create factors using an orthogonal method, the factor
scores (scores derived from the factor structure; see Chapter 9) are
often correlated despite the orthogonal nature of the factors. Therefore,
using orthogonal rotation results in a loss of valuable information
if the factors are really correlated, and oblique rotation should
theoretically render a more accurate, and perhaps
more reproducible, solution. Further, in the unlikely event
that researchers manage to produce truly uncorrelated factors, orthogonal
and oblique rotation produce nearly identical results, leaving oblique
rotation a very low-risk, potentially high-benefit choice.
The two sets of methods—orthogonal
and oblique—do, however, differ in ease of interpretation.
When using orthogonal rotation, researchers have only one matrix to
interpret. When using oblique rotations, there are two matrices of
results to review (described in the next section). In our experience—and
in many of our examples—the two matrices tend to parallel each
other in interpretation. So again, in our mind this does not create
an insurmountable barrier.