Validation of the importance of continua in representing delusional ideation in the general population

Previous studies have examined the distribution of psychotic like experiences (PLE) with the aim of informing the debate concerning the categorical versus continuous nature of psychosis. We extend this research by subjecting a number of competing models of delusional ideation to validation analysis to further examine previous findings.

We constructed latent variable models representing the factor structure of delusional ideation reported previously, using self-reported delusional ideation (Peter's Delusional Inventory; PDI) at age 21 in a general population prospective birth cohort study. After firstly eliminating models which exhibited poor fit we performed a longitudinal validation analyses among the competing models to investigate whether increasing levels of ideation were associated with developmental antecedents, correlates and distal indicators of psychotic disorder.

Four latent variable models were found to adequately represent the delusional ideation data, two comprised exclusively of continua (a multidimensional 5 factor model and a bifactor model with 1 general and 4 specific factors), and two which included both categories and continua (two factor mixture models, each with 3 classes and 1 factor per class, but with varying levels of parameter restrictions). Exclusively categorical latent models obtained poor fit and the categorical components of hybrid models failed to discriminate on psychotic illness, while among the models incorporating continuous latent factors validation analyses did not clearly identify any model as better than the others.

We provide novel evidence of the importance of continua in adequately and validly representing delusional ideation in the general population. Beyond this, our data suggests it is not possible to further refine the structure of delusional ideation in the general population.