Latent Profile Analysis of Schizotypy and Paranormal Belief: Associations with Probabilistic Reasoning Performance

Andrew Denovan, Neil Dagnall, Kenneth Drinkwater, Andrew Parker

Research output: Contribution to journalArticlepeer-review

25 Citations (Scopus)


This study assessed the extent to which within-individual variation in schizotypy and paranormal belief influenced performance on probabilistic reasoning tasks. A convenience sample of 725 non-clinical adults completed measures assessing schizotypy (Oxford-Liverpool Inventory of Feelings and Experiences; O-Life brief), belief in the paranormal (Revised Paranormal Belief Scale; RPBS) and probabilistic reasoning (perception of randomness, conjunction fallacy, paranormal perception of randomness, and paranormal conjunction fallacy). Latent profile analysis (LPA) identified four distinct groups: class 1, low schizotypy and low paranormal belief (43.9% of sample); class 2, moderate schizotypy and moderate paranormal belief (18.2%); class 3, moderate schizotypy (high cognitive disorganization) and low paranormal belief (29%); and class 4, moderate schizotypy and high paranormal belief (8.9%). Identification of homogeneous classes provided a nuanced understanding of the relative contribution of schizotypy and paranormal belief to differences in probabilistic reasoning performance. Multivariate analysis of covariance revealed that groups with lower levels of paranormal belief (classes 1 and 3) performed significantly better on perception of randomness, but not conjunction problems. Schizotypy had only a negligible effect on performance. Further analysis indicated that framing perception of randomness and conjunction problems in a paranormal context facilitated performance for all groups but class 4.
Original languageEnglish
Article number35
Number of pages18
JournalFrontiers in Psychology
Publication statusPublished - 26 Jan 2018
Externally publishedYes


Dive into the research topics of 'Latent Profile Analysis of Schizotypy and Paranormal Belief: Associations with Probabilistic Reasoning Performance'. Together they form a unique fingerprint.

Cite this