Covariance of baseline symptom scores in prediction of future symptom scores: A methodological note

Abstract

It is now standard practice in psychopathology research to longitudinally assess the predictive effects of putative etiological factors on symptoms. A common statistical technique employed in the implementation of such research is multiple regression/correlation, with a baseline symptom measure used as a covariate in an analysis of the association between etiological factors and future symptoms. This report highlights the importance of testing the assumption of homogeneity of covariance in such analyses, not only because it is statistically required, but also because it provides theoretically relevant information regarding onset, exacerbation, or remission of symptoms.