An application of item response mixture modelling to psychosis indicators in two large community samples

Previous research has suggested that psychosis is better described as a continuum rather than a dichotomous entity. This study aimed to describe the distribution of positive psychosis-like symptoms in two large community samples using an item response mixture model.

An item response mixture model was used to explain the pattern of psychosis-like symptom endorsement. This model incorporated two elements. First, a continuous non-normal latent variable to explain the observed pattern of data. Second, a categorical latent variable to explain the variation in the continuous non-normal latent variable.

For both samples, representing broadly and narrowly defined psychosis, the best fitting model was a four-class solution. In both cases, the classes differed quantitatively rather than qualitatively.

The analysis showed that psychosis-like symptoms at the population level could be best explained by four classes that appeared to represent an underlying continuum.