Hi all,

I've been asked to do some cross-validation for a CFA that I recently ran.
It is a sample of several hundred individuals who completed a psychological
instrument.  We then fit several proposed factor structures for this
instrument (5 in total).  The reviewer cited a couple of sources and I've
been able to locate the oldest of the two.  The article I read (Floyd &
Widaman. Psychological Assessment 1995;7(3):286-299), indicates that
cross-validation is important for CFA but doesn't really say much more than
that for CFA.  Is cross-validation important/appropriate for a CFA that is
comparing a sample to 5 different potential models?



I've read several CFA articles and none of them seem to mention true
cross-validation.  They seem to reference the ECVI statistic as a single
sample cross-validation statistic.  Is that sufficient to address
cross-validation?



David Neal