To quantitatively review all presently available evidence about the in
terrelations between positive and negative schizophrenic symptoms, we
created an aggregate matrix of the intercorrelations among schizophren
ic symptoms by combining data from 28 independent samples using meta-a
nalytic procedures (net bivariate dfs ranging from 683 to 1657). Using
confirmatory factor analyses, we then statistically compared four the
oretically derived models of the structure of schizophrenic symptoms.
Although a three-factor model (Liddle, 1987) best fit the data, result
s suggest that either more factors or different symptoms are required
to account well for the latent structure underlying schizophrenic symp
tomatology. The nature of such augmented approaches, the opportunities
and constraints inherent to multifactorial models, and the limitation
s of current instruments are discussed. (C) 1998 Elsevier Science B.V.
All rights reserved.