I would appreciate advice or information
on whether it is possible to perform a
subjective/manual rotation for
theoretical purposes when conducting
factor analysis with Stata 9.2. I'll
try to explain my purpose with a
hypothetical example.
Assume I have a correlation matrix of 10
attitudinal items and that in previous
research Item 1 has been identified as
highly representative (an ideal type) of
a certain theoretical trait and that
Item 10 has been identified as highly
representative (a different ideal type)
of a different (but not
necessarily orthogonal) theoretical
trait. If I were to factor analyze the
correlation matrix and observe the
factors on which each of these two items
loads highest after some traditional
orthogonal or oblique rotation, I would
expect that Item 1 would have a
significant and higher (or bi-polar,
negative) loading than Item 10 on one
factor and that the reverse would hold
for the relationship of the two items on
a second factor. It is possible,
however, given sampling and other
issues, that neither Item 1 nor Item 10
would actually be the highest loaded
items on their respective factors in my
sample after traditional factor rotation
strategies designed to achieve simple
structure. I would like, therefore, to
manually rotate the factors to achieve
this objective of maximizing the
loadings of Items 1 and 10 on their
respective factors to the extent
possible.
That is, I would like, for theory
testing purposes, to be able to rotate
the factors manually so that I can find
a factor solution in which the loadings
of Item 1 and Item 10 are maximized as
much as possible on their respective
factors. Is such a rotation strategy
available in Stata?
Thanks for any information or advice.
George
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
