Discovery Lab Answers

Kenneth Drinkwater, PhD is a psychologist at Manchester Metropolitan University whose research sits at the intersection of parapsychology and mainstream cognitive psychology. His work focuses primarily on the psychological architecture of paranormal belief rather than on direct psi experimentation — examining how individual differences in reasoning, personality, and cognition shape anomalous belief and experience. The sources retrieved for this question cover his collaborative output, predominantly with Neil Dagnall and Andrew Denovan, across several active research lines.

Experiments

Drinkwater’s empirical program spans at least four identifiable lines of work, all conducted collaboratively.

Paranormal belief structure. A core contribution is large-scale psychometric analysis of the Revised Paranormal Belief Scale (RPBS). The ESP-Nexus page for this line of work notes that Drinkwater led a study aggregating data from 3,764 participants to test competing factorial models of the RPBS, finding that a seven-factor bifactor solution provided the best fit — indicating that paranormal belief functions simultaneously as a general factor and as a set of specific subdimensions ^[2][4].

Belief, conspiracy endorsement, and wellbeing. Two recent studies examine how paranormal belief and conspiracy theory endorsement relate to positive wellbeing and meaning in life. One, with Dagnall, Denovan, and Escolà-Gascón, used network analysis on a sample to map associations among paranormal belief, conspiracy endorsement, and wellbeing constructs ^[2]. A companion longitudinal study (N = 1,158) used structural equation modeling and latent profile analysis to examine adaptive functions and wellbeing variations ^[4]. A further mediation study (N = 647) tested whether creativity and self-esteem sequentially mediate the path from paranormal and conspiracist belief to meaning in life ^[5].

Cognitive styles of skepticism and paranormal belief. A 2026 study with Dagnall, Denovan, Murphy-Morgan, Powell, and Neave used latent profile analysis to identify subgroups — distinguishing a “Higher Evidence-based Thinking” profile (55% of the sample, n = 165) from a contrasting profile — and examined how these differ across paranormal belief, new age philosophy, and belief in science ^[1].

Anomalous experience and haunting-type reports. Drinkwater has contributed to research on ghostly episodes and haunt-type phenomena. A 2025 multi-author study with Massullo, Houran, Escolá Gascón, O’Keeffe, and Dagnall quality-checked a “fact sheet” on ghostly episodes, evaluating its perceived accessibility, usefulness, and overall impression across four respondent groups: ghost hunters, clinicians, lay percipients, and lay non-percipients ^[6]. Earlier work cited across the sources includes VAPUS model illustrations for ghost narratives and related haunt phenomenology.

Illusory health beliefs and psychometric validation. Drinkwater co-authored a validation study of the Illusory Health Beliefs Scale using exploratory structural equation modeling and multidimensional Rasch analysis, connecting illusory health belief to related paranormal constructs including psi, precognition, and superstition ^[3].

Methodology

Drinkwater’s studies consistently employ large online survey samples with established psychometric instruments. Key design features across the retrieved work include:

• Instrument choice: The RPBS (seven subscales including traditional religious belief, psi, witchcraft, spiritualism, superstition, extraordinary life forms, and precognition) is the primary measure of paranormal belief across studies ^[2][4][5]. The Generic Conspiracist Beliefs Scale (GCBS or its short form GCB-5) is used to operationalize conspiracy endorsement ^[4][5]. Schizotypy is measured with the Schizotypal Personality Questionnaire — Brief (SPQ-B) ^[5].
• Latent variable and profile methods: Multiple studies use latent profile analysis (LPA) to identify belief subgroups rather than treating belief as a continuous variable. LPA model selection follows standard fit indices (AIC, BIC, ssaBIC, entropy, Lo-Mendell-Rubin Adjusted Likelihood Ratio Test) ^[1][4][5].
• Structural equation modeling: Path and mediation analyses use bootstrapped confidence intervals (1,000 resamples) with bias-corrected 95% CIs, following Preacher and Hayes (2008). Model comparison uses the Satorra-Bentler chi-square test ^[4][5].
• Longitudinal designs: At least one study employs a three-wave longitudinal design, permitting examination of temporal stability and prospective relationships among belief, coping, and wellbeing ^[4].
• Psychometric validation: The Illusory Health Beliefs Scale study used both exploratory structural equation modeling (ESEM) and multidimensional Rasch analysis in combination, addressing limitations of single-method approaches to scale validation ^[3].
• Common method bias checks: Studies consistently apply the Harman single-factor test to assess common method bias. In the IHBS validation, a single factor accounted for 20.98% of variance, below the 50% threshold ^[3].
• Expert panel ratings for haunt research: The ghostly episodes fact-sheet study used structured ratings by four respondent groups across multiple quality dimensions, with Bayes factors computed alongside frequentist tests to assess the strength of evidence for group differences ^[6].

Data

The retrieved sources do not present direct psi-effect outcomes (hit rates, z-scores in forced-choice trials, or similar) because Drinkwater’s program is oriented toward the psychology of paranormal belief, not toward experimental psi testing. The quantitative outputs are therefore psychometric and structural rather than parapsychological-effect statistics.

Selected findings from the retrieved sources, in tabular form where the surrounding text is intact and attribution is unambiguous:

On the belief–conspiracy relationship: across multiple studies, paranormal belief and conspiracist belief correlate positively and consistently, a finding Drinkwater et al. have characterized as replicating across different measurement instruments ^[4]. The network analysis study found satisfaction with life negatively associated with conspiracy endorsement (r = −0.14 to −0.30 range across conditions) and positively associated with other wellbeing indicators ^[2].

The haunt fact-sheet study found that, for most quality dimensions (accessibility, usefulness), differences among ghost hunters, clinicians, lay percipients, and lay non-percipients were not statistically significant — with Bayes factors supporting the null in several cases — while Overall Impression showed two significant pairwise contrasts of moderate size ^[6].

The retrieved excerpts do not include full descriptive statistics for all samples; where sample sizes for specific studies are not recoverable from the excerpts without ambiguity, they are not stated above.

Skeptical critiques

The sources retrieved for this question do not include published external criticism directed specifically at Drinkwater’s methodology or findings. What the sources do record are the authors’ own acknowledged limitations, which are worth noting as the closest available critical engagement.

Dagnall et al. (2026) ^[1] note that latent profile analysis profiles are statistical constructs dependent on the specific variables entered, and that the study’s two-profile solution, while superior by fit indices, may not exhaust conceptual possibilities. They also note reliance on self-report and established measures, which constrain precision in operationalizing skepticism.

Dagnall et al. (2025) ^[4] acknowledge that the longitudinal design, while an improvement over cross-sectional work, does not establish causation, and that the sample’s mean age and online recruitment mode may limit generalizability. The use of the RPBS and GCBS — though defended as established instruments with known psychometric properties — is noted as carrying conceptual alignment assumptions that may not hold across all populations.

Massullo et al. (2025) ^[6] explicitly limit their conclusions to perceived quality of the ghostly episodes fact sheet rather than its educational or clinical impact, and note that inter-rater comparisons may be marginal given the sample size and that effect sizes based on explained variance were modest (maximum 12%).

Independent external replication of Drinkwater’s specific psychometric findings — the seven-factor bifactor RPBS solution, the two-profile cognitive-style typology — by research groups outside the Manchester Metropolitan University / Liverpool John Moores University collaboration has not been documented in the sources retrieved for this question.

For a fuller overview of Drinkwater’s research program, including his work on schizotypy and reasoning biases, see his profile at Kenneth Drinkwater, PhD — ESP-Nexus and the spoke page on paranormal belief structure and psychological dimensions.