Two major drawbacks of a recently proposed F-test for abstract factor analy
sis (AFA) are discussed. First, heteroscedasticity in the data is not handl
ed by a proper data pre-treatment. Second, negative F-values and associated
degrees of freedom are encountered. It is proposed to pre-treat the data p
rior to AFA. Copyright (C) 2000 John Wiley & Sons, Ltd.