Discerning the latent structure of hypothetical psychosis proneness through admixture analysis

The application of psychometric procedures to a normal population sample to detect individuals with increased liability for schizophrenia is a useful methodological adjunct to the traditional genetic high-risk strategy. A necessary and reasonable step in the process of establishing the utility of a viable psychometric index of schizotypy is the formal investigation of the latent structure of psychometric values. The present study used admixture analysis to examine the distribution of scores on the Perceptual Aberration Scale (PAS), an objective measure of hypothetical psychosis-proneness, in a randomly ascertained sample of 18-year-old university students (n = 707). We applied parametric methods that assumed normally distributed component distributions; viewed in this context, our results are hypothesis-generating and not definitive confirmation of specific hypotheses. Within our methodological framework, the results provide strong evidence for the commingling of normal distributions, even after allowing for unequal variances across components and after removing skewness. The overall distribution of power-transformed data is consistent with the existence of three qualitatively distinct classes of PAS responders. We discuss our results in light of Meehl's model of schizotaxia, a "mixed model" of inheritance of liability to schizophrenia, and we review the methodological implications of our findings for future research.