ArticlePDF Available

Abstract and Figures

This study examined the degree to which within-individual variations in paranormal experience were related to belief in the paranormal, preferential thinking style, and delusion formation. A sample of 956 non-clinical adults completed measures assessing experience-based paranormal indices (i.e., paranormal experience, paranormal practitioner visiting, and paranormal ability), paranormal belief, belief in science, proneness to reality testing deficits, and emotion-based reasoning. Latent profile analysis (LPA) combined the experience-based indices to produce six underlying groups. Inter-class comparison via multivariate analysis of variance (MANOVA) indicated that both breadth and intensity of experiential factors were associated with higher belief in in the paranormal, increased proneness to reality testing deficits, and greater emotion-based reasoning. Belief in science, however, was less susceptible to experiential variations. Further analysis of reality testing subscales revealed that experiential profiles influenced levels of intrapsychic activity in subtle and intricate ways, especially those indexing Auditory and Visual Hallucinations and Delusional Thinking. Collectively, identification of profiles and inter-class comparisons provided a sophisticated understanding of the relative contribution of experiential factors to differences in paranormal belief, belief in science, proneness to reality testing deficits, and emotion-based reasoning.
Content may be subject to copyright.
ORIGINAL RESEARCH
published: 28 June 2021
doi: 10.3389/fpsyg.2021.670959
Frontiers in Psychology | www.frontiersin.org 1June 2021 | Volume 12 | Article 670959
Edited by:
Roumen Kirov,
Bulgarian Academy of Sciences
(BAS), Bulgaria
Reviewed by:
Emmanuelle Charlotte Sophie
Bostock,
University of Tasmania, Australia
Christine Anne Simmonds-Moore,
University of West Georgia,
United States
*Correspondence:
Kenneth Graham Drinkwater
k.drinkwater@mmu.ac.uk
Specialty section:
This article was submitted to
Consciousness Research,
a section of the journal
Frontiers in Psychology
Received: 22 February 2021
Accepted: 03 June 2021
Published: 28 June 2021
Citation:
Drinkwater KG, Dagnall N, Denovan A
and Williams C (2021) Paranormal
Belief, Thinking Style and Delusion
Formation: A Latent Profile Analysis of
Within-Individual Variations in
Experience-Based Paranormal Facets.
Front. Psychol. 12:670959.
doi: 10.3389/fpsyg.2021.670959
Paranormal Belief, Thinking Style and
Delusion Formation: A Latent Profile
Analysis of Within-Individual
Variations in Experience-Based
Paranormal Facets
Kenneth Graham Drinkwater*, Neil Dagnall, Andrew Denovan and Christopher Williams
Department of Psychology, Manchester Metropolitan University, Manchester, United Kingdom
This study examined the degree to which within-individual variations in paranormal
experience were related to belief in the paranormal, preferential thinking style, and
delusion formation. A sample of 956 non-clinical adults completed measures assessing
experience-based paranormal indices (i.e., paranormal experience, paranormal
practitioner visiting, and paranormal ability), paranormal belief, belief in science,
proneness to reality testing deficits, and emotion-based reasoning. Latent profile
analysis (LPA) combined the experience-based indices to produce six underlying groups.
Inter-class comparison via multivariate analysis of variance (MANOVA) indicated that both
breadth and intensity of experiential factors were associated with higher belief in in the
paranormal, increased proneness to reality testing deficits, and greater emotion-based
reasoning. Belief in science, however, was less susceptible to experiential variations.
Further analysis of reality testing subscales revealed that experiential profiles influenced
levels of intrapsychic activity in subtle and intricate ways, especially those indexing
Auditory and Visual Hallucinations and Delusional Thinking. Collectively, identification
of profiles and inter-class comparisons provided a sophisticated understanding of the
relative contribution of experiential factors to differences in paranormal belief, belief in
science, proneness to reality testing deficits, and emotion-based reasoning.
Keywords: paranormal belief/experience, reality testing, emotion-based reasoning, belief in science, delusion
proneness, latent profile analysis
INTRODUCTION
National surveys report that belief in the paranormal remains widespread within contemporary
Western societies (i.e., United Kingdom, Ipsos, 1998,Ipsos, 2003, United States, Gallup: Newport
and Strausberg, 2001; Moore, 2005). Indeed, a 2005 Gallup poll (Moore, 2005) observed that
three in four Americans acknowledged at least one paranormal belief. Concomitant with belief,
the reporting of paranormal experiences is also relatively common (e.g., Schmied-Knittel and
Schetsche, 2005; Castro et al., 2014; Dagnall et al., 2016). For instance, Dagnall et al. (2016) noted
that 42% of a British university-based sample experienced at least one paranormal occurrence.
Studies from other geographical areas have produced comparable findings (e.g., Germany,
Schmied-Knittel and Schetsche, 2005; America, McCready and Greeley, 1976; and Latin American,
Drinkwater et al. Latent Profile Analysis
Montanelli and Parra, 2002; Marks, 2021). Consideration of
literature reveals also that experiencers frequently report multiple
occurrences (Castro et al., 2014; Dagnall et al., 2016).
Irwin et al. (2013) postulate that discernment of a paranormal
experience comprises two fundamental processes. Perception of
an anomaly (inexplicable stimulus), and subsequent ascription
of causation to paranormal entities or powers (Lange et al.,
2019). Thus, belief provides an interpretative lens that structures
comprehension of anomalous phenomena. Explicitly, it provides
a cognitive framework for organising life events, so that
they appear intellectually coherent, and experiences represent
internal misattributions to external (“paranormal”) forces (Irwin,
1994; Dagnall et al., 2020). Congruent with this notion,
several studies report a positive association between belief in
and experience of the paranormal (Glicksohn, 1990; Dagnall
et al., 2016). Noting the interpretative nature of this process,
researchers often describe outcomes as subjective paranormal
experiences (SPEs) (see Neppe, 1990; Dagnall et al., 2016;
Drinkwater et al., 2020). Similarly, when faced with anomalous
phenomena, disbelief can facilitate rejection of paranormal
explanations in favour of conventional elucidations (Dagnall
et al., 2017a).
This constructionist view accords with sociological and
psychological research. Sociologically, the cultural source
hypothesis depicts paranormal experiences as products of
tradition, or imaginary happenings created, and shaped by social
custom (Hufford, 1982; McClenon, 1994). At an individual
cognitive level, this aligns with the psychological concept
of worldview, the idea that overarching expectations and
assumptions (beliefs) about the world shape understanding of
reality and existence (Overton, 1991; Miller and West, 1993;
Koltko-Rivera, 2004; Dagnall et al., 2015b). In this context, direct
(i.e., personal encounters) and indirect experience (second-
hand accounts, academic knowledge, media, etc.) provide
confirmatory evidence for beliefs.
Personal life events can also promote supernatural credence
(Glicksohn, 1990). This viewpoint is embodied within
experiential source theory (Hufford, 1982), which advocates
that a significant portion of traditional supernatural belief is
associated with accurate observations, interpreted rationally.
Explicitly, that certain types of anomalous experience provide
a source of recurring beliefs (Hufford, 1982, 2005; Bennett,
1987; McClenon, 1994). Consistent with experiential source
theory, the exchange and evaluation of information about
anomalous occurrences, in part, contributes to the creation
of folk religions. An example of this is the Night or Old Hag
tradition, where reference to a communal, paranormal entity
is used to explain elements of sleep paralysis, such as feeling
immobilised by a malevolent presence (Hufford, 2005). Other
studies provide affirmative evidence for the experiential source
hypothesis. Pertinent to the present paper, these reference
the observation of “common elements” in reported cases of
spontaneous extrasensory perception, precognitive dreams,
apparitional experiences, and contact with the dead (see Rhine,
1981; Emmons, 1982; McClenon, 1990). Similarly, in support
of the experiential source theory, several authors note that
individuals frequently cite personal paranormal encounters as
the motivation for belief(s) (McClenon, 1982, 2000; Blackmore,
1984; Irwin, 1991).
In rational terms, since individuals can explain anomalous
stimuli via a range of non-supernatural elucidations, experiential
factors are less likely to generate paranormal attributions
than beliefs. Acknowledging this, the tendency to ascribe
paranormality to experience is heightened under specific
conditions. For example, when a personal event creates
uncertainty and/or anxiety, the psychological desire for
understanding and control, can truncate objective decision-
making and encourage an overreliance on self-generated
(internal/subjective) data (Frenkel-Brunswik, 1949, 1951).
This includes the endorsement of alternative, scientifically
unsubstantiated (paranormal) beliefs (see Williams and Irwin,
1991; Houran and Williams, 1998; Hart et al., 2013). Moreover,
experience and belief can function in an interactive, reciprocal
manner so that experiences stimulate interest and belief in the
supernatural (van Elk, 2017), and beliefs encourage the search
for confirmatory personal paranormal occurrences (Dagnall
et al., 2015a; Drinkwater et al., 2020). This dynamic synergy
is an inherent feature of Van Leeuwen and van Elk’s (2019)
Interactive Religious Experience Model (IREM). Within the
IREM, general belief causes individuals to seek situations that
activate corresponding experiences. In the case of religion, these
often take the form of low-level agency-intuitions (i.e., feelings
of presence). These are important since they reflect and influence
belief formation.
In terms of previous research, the observed co-occurrence
of experience(s) and beliefs supports the notion that the
constructs are often connected within individuals. Indeed, studies
generally report a moderate-positive correlation (Cohen, 1992)
(e.g., Glicksohn, 1990; Musch and Ehrenberg, 2002; Dagnall
et al., 2016). Regarding the observed relationship, there are
important points to note. Firstly, variations in association
strength arise from the employment of different measurement
instruments. There is no commonly agreed index of experience,
and researchers have historically assessed belief using a variety
of scales, which encompass diverse but related content (i.e.,
dimensions) (see Dagnall et al., 2010a). Secondly, studies using
the Australian Sheep Goat Scale (ASGS; Thalbourne and Delin,
1993), a widely employed index of belief, require cautious
interpretation as the scale conflates belief, experience, and
ability. Particularly, items sample all three constructs (e.g., “I
believe I have had personal experience of ESP”). In cases where
experiences overlap with elements present within the ASGS this
likely results in correlation inflation (i.e., ESP, psychokinesis, and
life after death, see Drinkwater et al., 2018b).
Thirdly, since the RPBS assesses a wide breadth of construct
content (i.e., Psi, Witchcraft, Superstition, Spiritualism,
Extraordinary Life Forms, and Precognition) that varies in
frequency of perceived occurrence and plausibility, the scale
may not accurately assess the general relationship between
paranormal belief and experience. This notion is supported by
Dagnall et al. (2016), who observed considerable variation in
experience(s) as a function of type. For instance, extrasensory
perception was commonly reported, whereas psychokinesis was
rarely acknowledged. There is also an important conceptual
Frontiers in Psychology | www.frontiersin.org 2June 2021 | Volume 12 | Article 670959
Drinkwater et al. Latent Profile Analysis
misalignment between haunting encounters and RPBS
dimensions. The nearest indirect correspondences are the
Traditional Religious Belief (i.e., soul, Devil, God, and heaven
and a hell) and Spiritualism (i.e., astral projection, out-of-body
experiences, reincarnation, and communicate with the dead)
subscales. Hence, although haunt-related experiences are
relatively commonly reported (i.e., ghosts, poltergeists, and
apparitions) there is no direct referent within the RPBS.
Thirdly, the reported relationship between belief and
experience is more compelling when Gignac and Szodorai’s
(2016) normative guidelines for interpreting correlation effect
sizes are applied (i.e., relatively large, r>0.39). Finally, the two
constructs are not mutually inclusive. Hence, not all believers
have had a corresponding experience, nor are all experiencers
high in paranormal belief (Drinkwater et al., 2017a; Lange
et al., 2019; Wahbeh et al., 2019). Indeed, individuals can label
paranormal events as “supernatural” without any great faith or
conviction (Drinkwater et al., 2013, 2017a). Thus, SPEs represent
just one ontological facet that informs or reinforces belief.
The relatively small proportion of variance shared by the two
constructs demonstrates the relative independence of experience
and belief. Moreover, experience and belief are differentially
related to psychological variables (Rattet and Bursik, 2001).
Collectively, these factors help to explain why research on the
experiential basis of belief has generated inconsistent findings
(Castro et al., 2014).
Limitations of Previous Work
Noting these issues, it is apposite to conclude that due
to methodological limitations, previous work examining the
relationship between paranormal experience and belief has
provided useful, but restricted insights. The major limitation
being overreliance on a narrow index of person-centred
paranormal experience, direct reported encounters (i.e., SPEs).
This focus ignores the fact that involvement in the paranormal
takes many forms, such as actively engaging with practitioners
(i.e., Mediums, Psychics, Fortune-Tellers, and Spiritualists), and
normative influences (e.g., family and peers) (Hill et al., 2018;
Drinkwater et al., 2019).
Additionally, there are potentially important differences
between people who report single vs. multiple instances, and
those that believe they possess psychic abilities. Consistent
with this supposition, Zingrone et al. (2009) observed positive
relationships between aura viewers (i.e., perceiving lights,
glimmers, or force fields around the human body) and reports of
other psychic experiences (extrasensory perception, apparitions,
and out-of-body experiences). This suggests that multiple
experiencers have increased openness to “other” paranormal
encounters and perhaps in some instances self-perceived abilities.
Consequently, they may differ from individuals who experience
one off phenomenon, or just one type of paranormal experience.
Furthermore, it is possible to conceptualise experiences as
predominately dispositional (internally generated by an inherent,
personal quality, e.g., precognition) vs. situational (externally
created by context/situation, e.g., ghost sighting). From this
perspective, perceived possession of abilities infers differences in
type and frequency of reported paranormal experience(s).
Considerations such as these suggest that the measurement of
specific phenomena (i.e., personal encounters) alone is unlikely
to offer full insights into the effects of paranormal experience on
individual beliefs and cognitions. The importance of including
an assortment of experiential facets is not a new idea as
demonstrated by the Anomalous Experiences Inventory (AEI;
Gallagher et al., 1994). The AEI contains subscales measuring
experiences, abilities, fear, and beliefs. The AEI, however, is rarely
used in contemporary research since the Revised Paranormal
Belief Scale (RPBS; Tobacyk, 2004) and Australian Sheep Goat
Scale (ASGS; Thalbourne and Delin, 1993) have become the
mostly widely used measures of belief (Drinkwater et al., 2017b).
Another limitation derives from researchers’ use of variable-
based analytical methods, such as path models and regression.
For example, Dagnall et al. (2016) employed correlation analyses
when assessing indices of paranormal experience. A variable-
centred approach anticipates that findings are an estimate of
associations among discrete variables averaged across the whole
population, correspondingly assumed to be homogeneous (Orri
et al., 2017). This variable-centred approach is problematic
because it fails to examine how indices of experience relate within
individuals and ignores the observation that they interrelate
(interact) in complex ways.
Addressing Limitations
Noting these limitations, the present paper used a range
of experience-based indices (i.e., paranormal experience,
paranormal practitioner visiting, and paranormal ability). From
a methodological perspective, the approach of including and
amalgamating multiple measures was advantageous since it
sampled greater construct breadth and acknowledged intra-
respondent (person-centred) variations within participants.
Latent profile analysis (LPA) was used to combine experience-
based indices. LPA categorises individuals into configural profiles
based on varying degrees of probabilities (for a review of LPA see
Spurk et al., 2020).
The emergent composite measure, like the AEI, included
direct experience and self-professed ability. Regarding the AEI
dimension of Fear, this was replaced by paranormal practitioner
visiting. This decision was informed by the rationale that
visiting reflected an “active” desire to seek out paranormal
experience, whereas Fear was likely to promote “avoidance.”
These factors together encompassed, theoretically important
experiential aspects. It was necessary to develop new measures
because preceding research has typically employed limited
measures of paranormal experience (e.g., SPE), which focus on
restricted aspects of direct encounters.
In this context, the use of LPA to combine person-centred
factors represented an important conceptual development.
Particularly, the use of profiles recognised the heterogeneous
nature of individual paranormal experience and facilitated the
identification of subtle differences between category members.
Previous work using LPA, has provided useful insights into how
interactions between belief in the paranormal and schizotypy
are related to differential performance on probabilistic reasoning
tasks (Denovan et al., 2018). Similarly, research using cluster
analysis has contributed greatly to academic understanding
Frontiers in Psychology | www.frontiersin.org 3June 2021 | Volume 12 | Article 670959
Drinkwater et al. Latent Profile Analysis
of paranormal experience and belief. Clustering algorithms
partition data into subsets based on similarity or dissimilarity
(Frades and Matthiesen, 2010). For instance, Goulding (2004,
2005) used cluster analysis to investigate how relationships
between schizotypal factors were associated with belief in
the paranormal.
The Present Paper
Observing the potential of LPA, the current paper examined
the degree to which profile membership was associated
with differences in paranormal belief and cognitive-perceptual
information processing; specifically, preferential thinking style
(i.e., intuitive vs. experiential) and factors related to delusion
formation in general populations (i.e., reality testing and
emotion-based reasoning). This builds upon the body of
academic work that has investigated the extent to which
preferential thinking style predicts endorsement of scientifically
unsubstantiated beliefs and mediates related cognitive-perceptual
processes (i.e., schizotypy) (Dagnall et al., 2017a; Barron
et al., 2018; Denovan et al., 2020). Although the relationship
between belief in the paranormal and intuitive thinking is well-
established, relatively few studies have examined whether this
applies also to experiences (Irwin and Wilson, 2013). Those that
have, posit a similar positive association between paranormal
experiences and intuitive-experiential processing (Wolfradt and
Watzke, 1999; Wolfradt et al., 1999; Irwin and Wilson, 2013).
Researchers often adopt a dual processing approach to
examine differences in preferential thinking style. This refers
to the notion that two independent, but parallel operating
systems influence decision-making (Epstein et al., 1996).
Although dual processing conceptualisations vary across
models, theorists generally agree on the importance of the
distinction between automatic (implicit) and controlled
(explicit) processes (see Shirzadifard et al., 2018). Consistent
with this perspective, preceding research has drawn heavily
on the intuitive-experiential vs. analytical-rational dichotomy
encapsulated within Cognitive-Experiential Self-Theory (CEST;
Epstein, 1983, 1990).
CEST advocates that experiential processing is automatic,
rapid, unconscious, and holistic, whereas rational processing is
deliberate, slow, and conscious. Correspondingly, experiential
processing draws on feelings and emotional experiences
(subjective, internal mental activity) (Epstein et al., 1996), while
rational processing seeks to validate meaning by referring to
impartial evidence (objective, external data) (Epstein, 1994).
Although these processes contribute jointly to reasoning and
operate in parallel, one system typically preponderates.
Consistent with earlier work, indirect, proxy measures
assessed differences in thinking style (e.g., Dagnall et al.,
2017a; Denovan et al., 2017; Barron et al., 2018). Congruently,
the reality testing sub-scale of the Inventory of Personality
Organisation (IPO-RT; Lenzenweger et al., 2001) assessed
intuitive-experiential thinking, and the Belief in Science Scale
(BISS; Farias et al., 2013) indexed rational-analytical thinking.
The IPO-RT assesses the ability to differentiate self from non-self,
intrapsychic from external stimuli, and the capacity to maintain
empathy with ordinary social criteria of reality (Kernberg,
1996). This conceptualisation is consistent with the information
processing approach to belief generation proposed by Langdon
and Coltheart (2000).
This model explains delusions in terms of impairments within
the cognitive belief system. Delusions with ordinary content,
arise because of an extreme (but normal) attentional bias.
Precisely, the failure to critically evaluate hypotheses based
on misperceptions and misinterpretations of ambiguous first-
person experience. In the case of bizarre delusions, two deficits
occur. Firstly, damage to sensory and/or attentional-orienting
mechanisms creates an aberrant perception. Secondly, there
is a failure in belief evaluation. This explains why bizarre
delusions both contain unusual content and are implausible.
The implausibility arises from an inability to suspend the
natural tendency to favour direct first-person evidence. This bias
prevents objective critical evaluation of information.
Accordingly, researchers use the IPO-RT to measure
inclination to engage in intuitive-experiential processing
(Dagnall et al., 2010a; Denovan et al., 2017), proneness to
reality testing deficits (Irwin, 2004; Dagnall et al., 2017a), and
delusional thinking (Irwin et al., 2012; Dagnall et al., 2017b).
This reflects intersection between these constructs. Explicitly, the
delineation of reality testing as the ability to assess the validity of
beliefs and suppositions via reference to external data sources.
The application to delusions derives from the supposition that
delusion formation within non-clinical populations is associated
with proneness to reality testing deficits and emotion-based
reasoning (EBR) (i.e., the extent which judgements are based
on affective responses) (Irwin et al., 2012; Dagnall et al., 2017a;
Drinkwater et al., 2021).
From this viewpoint, proneness to reality testing deficits
is the failure to rigorously assess and continuously evaluate
information, and EBR denotes preference for emotional (vs.
logical) data. These psychological constructs align closely with
the contemporary definition of delusion(s) (Drinkwater et al.,
2021). This no longer references falsity (Dagnall et al., 2017b),
but instead views delusions as beliefs founded on insufficient
scrutiny of evidence. These are persistently held in the face of
conflicting evidence and are accepted for their emotional appeal
than for logical coherence (American Psychiatric Association,
2013). Although, emotional significance in paranormal beliefs
is typically lower than with psychotic delusions (Cella et al.,
2012), research with delusional patients has confirmed that EBR
generally plays an important role in delusion formations and
maintenance (e.g., Beck et al., 2008).
The shift in emphasis from falsity was beneficial to the
conceptual understanding of belief in the paranormal because,
unlike baseless psychotic delusions, it is not possible to
definitively disprove the existence of supernatural phenomena.
Thus, the inclusion of inadequate critical evaluation and
emotional emphasis as defining cognitive processes legitimises
the generalisation of findings from clinically defined delusions,
to paranormal beliefs in non-clinical populations. This concurs
with the clinically informed notions of delusions as beliefs arising
from faulty interpretation of anomalous experiences (Garety
and Freeman, 1999) and/or inadequate evidence (Irwin, 2009;
Coltheart et al., 2010; Irwin et al., 2012). From this perspective,
Frontiers in Psychology | www.frontiersin.org 4June 2021 | Volume 12 | Article 670959
Drinkwater et al. Latent Profile Analysis
paranormal beliefs within general populations represent non-
psychotic delusions (Irwin et al., 2012).
In support of this supposition, EBR and the tendency to
suspend reality testing predicted scores on the two dimensions
of the Survey of Anomalous Experiences (SAE) (Irwin et al.,
2013). The SAE measures proneness to anomalous experiences
(the tendency to experience anomalous or uncanny experiences)
and proneness to paranormal attributions (the degree to
which respondents ascribe experiences to specific paranormal
process). In addition to this, EBR and proneness to reality
testing deficits predicted intuitive–experiential thinking style
and were not associated with rational thinking (Irwin and
Wilson, 2013). These results concurred with studies that
found EBR was a prognosticator paranormal belief intensity
(e.g., Irwin et al., 2012).
These findings concur with the postulation that individuals
endorse paranormal beliefs based on emotional appeal, and that
this process is the foundation for enduring beliefs (Irwin et al.,
2012). This outcome aligned with Sappington (1990), who found
that potential to explain personal experiences heighten emotion-
based reasoning, and increased judgment of phenomena as
paranormal. These findings applied to experience, suggest that
scores on belief in the paranormal, proneness to reality testing
deficits, belief in science and EBR will vary as a function of
profile membership.
In conclusion, LPA examined the degree to which within-
individual variations in experience were related to belief in the
paranormal, preferential thinking style, and delusion formation.
Congruent with preceding research, it was predicted that profiles
with greater levels of paranormal experience would score higher
on belief in the paranormal and the measures of intuitive
thinking/delusion formation (i.e., IPO-RT and EBR), and lower
on critical thinking (i.e., BISS).
METHODS
Design
To identify heterogeneous, experience-based latent profiles a
cross-sectional design was used. Profiles comprised paranormal
experience, paranormal practitioner visiting, and paranormal
ability. Analysis then examined relationships between emergent
profiles and belief in the paranormal, preferential thinking style,
and delusion formation.
Respondents
The sample comprised 956 respondents, (Mean age, M)=
33.02 years, SD =14.64, range 18–83. There were 342 males
(36%), M=37.16 years, SD =15.51, range 18–82; and
614 females (64%), M=30.71 years, SD =13.61, range
18–83. For all variables, skewness and kurtosis values were
within the recommended range of2.0 to +2.0 (Byrne, 2010).
Respondent recruitment was via Bilendi, an online multi-
channel management platform for data collection. Bilendi is an
established provider of good quality representative samples (see
Häusermann et al., 2020; Salak et al., 2021).
The researchers requested a sample of UK-based respondents
aged 18 years and over. This was the only exclusion criteria. Data
accessed via respondent recruitment panels are generally more
diverse and far reaching than traditional student samples. These
advantages are not detrimental to quality and are commensurate
with traditional samples in terms of demographics and responses
to established surveys (Kees et al., 2017; Miller et al., 2017).
Measures
Experiential Paranormal Factors
The survey contained a section assessing experiential paranormal
factors. This was subdivided into paranormal experiences,
paranormal practitioner visiting, and perceived ability. These
measures were adapted from (see Dagnall et al., 2016; Drinkwater
et al., 2018a, 2021).
Paranormal Experience
Items asked respondents whether they had experienced a range of
psychic phenomena. Presented experiences were associated with
psi and life after death, and included communication with the
dead, psychic occurrence, mediumship, spiritualism, telepathy,
precognition, premonition, and remote viewing. These items
indexed frequently reported paranormal experiences (see Dagnall
et al., 2016), and represent core receptive elements of belief in the
paranormal (Drinkwater et al., 2018b).
Each psychic phenomenon was accompanied by a clear
delineation. The use of definitions ensured that participants were
responding to conceptual classifications, rather than personal
interpretations. For instance, “Precognition is paranormal
awareness (knowing) that an event in the future will occur.
In the context of this definition, have you ever personally
experienced precognition?” Summation of experiences produced
a total ranging from 0 (no experience) to 8 (experienced
all phenomena). A frequency scale followed each experience
item (EXP) (“On how many occasions? Once, Between 2 and
5, or More than 5”). This approach to the measurement of
self-reported paranormal experiences is well-established (see
Dagnall et al., 2016; Drinkwater et al., 2020, 2021). These scales
were dichotomous and demonstrated a good level of reliability
according to the Kuder-Richardson-20 (KR-20) coefficient, EXP
KR-20 =0.799; FREQ KR-20 =0.839.
Paranormal Practitioner Visiting
A further item set, using a dichotomous format (yes vs. no), asked
respondents whether they had visited paranormal practitioners
associated with psi and life after death. Designated categories
centred on main industries (i.e., Mediums, Psychics, Spiritualists,
and Fortune-Tellers). The reliability for this measure was
satisfactory, KR-20 =0.713. For each category, if respondents
provided an affirmative response, a further item assessed
frequency of visits (either once, between 2 and 5, or more than 5).
Reliability for this was good, KR-20 =0.849. A final item asked
how accurate (in percentage terms, 0-100) was the information
provided. Alpha and omega reliability for this continuous scale
was good, α=0.824; ω=0.826.
Frontiers in Psychology | www.frontiersin.org 5June 2021 | Volume 12 | Article 670959
Drinkwater et al. Latent Profile Analysis
Perceived Personal Ability
This section asked respondents about their perceived personal
abilities (i.e., mediumship, psychic, spiritualism, and fortune-
telling); e.g., “To what extent (in percentage terms, 0-100) do
you believe that you possess mediumship abilities?”. Internal
consistency for this was excellent, α=0.912; ω=0.917. A
final item asked respondent whether they were paranormal
practitioners (yes vs. no).
Belief in the Paranormal
The Manchester Metropolitan University New (MMU-N) scale
assessed belief in the paranormal. This measure has featured in
published studies (Dagnall et al., 2014, 2020; Drinkwater et al.,
2021) and provides total and dimensional scores (hauntings,
superstition, religious belief, alien visitation, ESP, PK, astrology,
and witchcraft) (Dagnall et al., 2010a,b,c). The MMU-N
comprises 50-items presented as statements (e.g., “Some people
have visions of the future, which come true”). Participants
respond to these using a seven-point Likert scale (ranging from
1, strongly disagree, to 7, strongly agree). Overall, the MMU-N is
conceptually coherent and psychometrically robust. Particularly,
the scale and subfactors possess good face validity and excellent
internal reliability (Dagnall et al., 2010a). Furthermore, the
measure has demonstrated good concurrent validity (Dagnall
et al., 2014). In the present report, excellent alpha and omega
reliability existed, α=0.963; ω=0.965.
Reality Testing
The IPO-RT contains 20-items presented as statements (e.g., “I
have heard or seen things when there is no apparent reason
for it”). Respondents indicate agreement on a five-point Likert
scale, with responses ranging from 1 =never true, to 5 =always
true. Totalling items produces scores between 20 and 100. Higher
scores are indicative of proneness to reality testing deficits and
greater reliance on intrapsychic activity (intuitive-experiential
thinking) (Dagnall et al., 2018). The IPO-RT has demonstrated
good internal and external reliability, and construct validity
(Lenzenweger et al., 2001; Dagnall et al., 2018).
The IPO-RT is a superior measure of intuitive-experiential
processing because it assesses extensive construct content. This
includes cognitive, perceptual, emotional, and social components
of internal and self-orientation (Dagnall et al., 2018). Indeed,
Dagnall et al. (2017a) supported this notion via verifying the
existence of four distinct but related subfactors. Auditory, and
visual hallucinations, delusional thinking (possessing beliefs
contrary to reality), social deficits (difficulties reading social
cues) and confusion (inability to understand feelings and
sensations). Other widely used measures, such as the Faith
in Intuition subscale of the Rational-Experiential Inventory
(REI; Epstein et al., 1996), focus only on the role of feelings,
instincts, and emotions in decision-making (Pennycook et al.,
2016). Therefore, the IPO-RT has become an established index
of individual inclination to engage in intuitive-experiential
processing (subjective thinking). In this study, excellent reliability
emerged for the overall measure, α=0.925; ω=0.926. The
subfactors of auditory and visual hallucinations (α=0.848; ω=
0.850), delusional thinking (α=0.858; ω=0.860), social deficits
(α=0.755; ω=0.763), and confusion (α=0.682; ω=0.691)
demonstrated good to acceptable internal consistency.
Belief in Science
The Belief in Science Scale (BISS; Farias et al., 2013) assesses
the degree to which respondents endorse the virtues of science.
The measure is composed of 10 statements (e.g., “Science is the
most efficient means of attaining truth”). Respondents indicate
their level of agreement via a six-point Likert scale (1 =Strongly
Disagree, to 6 =Strongly Agree). Researchers can produce
total scores by summing items (10 to 60) and can average
these to produce scores ranging from 1.0 to 6.0. The BISS has
high internal consistency and validity (Farias et al., 2013). The
measure also has demonstrated invariance (i.e., gender, form,
factor structure), and item intercepts for the unidimensional
structure (Dagnall et al., 2019). In the current study, internal
reliability was excellent, α=0.920; ω=0.922.
Emotion-Based Reasoning
The Emotion-Based Reasoning (EBR) subscale of the Cognitive
Biases Questionnaire (Peters et al., 2014) is a 6-item measure
of the degree to which decision-making is based upon affective
reactions. Items are framed in the context of a short vignette
and respondents select one of three available options that best
describes their feelings about the outlined situation. Each item
has a three-point scale (1 =absence of bias; 2 =presence of
bias with some qualification; and 3 =presence of bias). EBR
is computed as a total across the items, hence scores range
from 6 to 18. The Cognitive Biases Questionnaire possesses good
psychometric properties (Cronbach alpha, α=0.89; test-retest
reliability, r=0.92) (Peters et al., 2014). In the current research,
alpha was acceptable (α=0.630; ω=0.639). (see Taber, 2018).
This result aligned with previous research (e.g., Drinkwater et al.,
2018b), additionally the subscale possessed a satisfactory mean
inter-item correlation of 0.227.
Procedure
Respondents accessed materials via a web-link. Prior to
item presentation, potential respondents received background
information about the research project. This outlined the nature
of the investigation and details about ethics. To progress, it
was necessary for respondents to indicate informed consent.
Respondents then provided basic demographic information (i.e.,
age and preferred gender) before receiving the measures. Within
these, guidelines asked respondents to carefully read and answer
all questions, take their time, and respond in an open and
honest manner. Items were organised into sections: paranormal
experience and abilities, MMU-N, BISS, IPO-RT, and EBR.
To prevent order effects section presentation rotated across
respondents. Respondents worked through the items at their own
pace until they reached the end of survey, at which point they
received the debrief.
As aforementioned, this study used a cross-sectional design,
data collection occurred at one point in time. A frequent
criticism of this method is that it is prone to common method
variance (CMV) (Spector, 2019). To remedy this the researchers
employed procedural devices (Krishnaveni and Deepa, 2013).
Frontiers in Psychology | www.frontiersin.org 6June 2021 | Volume 12 | Article 670959
Drinkwater et al. Latent Profile Analysis
Explicitly, section instructions created psychological distance
between scales by accentuating differences between constructs
and measures (Podsakoff et al., 2003). Furthermore, instructions
reduced the potential for evaluation apprehension and social
desirability effects by telling respondents that there were
no right or wrong responses, and that they should answer
questions honestly.
Ethics Statement
Ethical approval was granted for a series of studies examining
psychological and neuropsychological factors associated with
self-professed psychic ability/mediumship by the Manchester
Metropolitan University Faculty of Health, Psychology and Social
Care Ethics Committee (October 2018).
Analysis
Aside from latent profile analysis (LPA), which required
Mplus version 7 (Muthén and Muthén, 2012), analyses utilised
SPSS 26. Preliminary analysis examined descriptive statistics.
Subsequently, exploratory LPA established latent group affiliation
based on indices of paranormal experiences and abilities
(Paranormal Experience, Paranormal Practitioner Visiting, and
Paranormal Ability). Assessment of model fit involved evaluating
a 1-class model, followed by examining models with an
increasing quantity of latent profiles until the inclusion of
additional profiles was not justified.
Consultation of the following indices determined the optimal
quantity of latent profiles: the Akaike Information Criterion
(AIC; Akaike, 1987), the Bayesian Information Criterion (BIC;
Schwarz, 1978), the sample-size adjusted BIC (ssaBIC; Sclove,
1987), the Lo-Mendell-Rubin-adjusted likelihood ratio test
(LMR-A-LRT; Lo et al., 2001), and a standardised measure of
entropy (Ramaswamy et al., 1993). Lower values suggest superior
fit for AIC, BIC, and ssaBIC. The LMR-A-LRT includes a p-
value and determines statistical significance (or otherwise) in fit.
Entropy values reflect the classification quality of participants,
with values above 0.8 indicative of a sound separation of
identified profiles relative to the data (Ramaswamy et al., 1993).
Resultant latent profiles represented a group variable
(independent variable) for assessing differences in relation
to Paranormal Belief, Reality Testing, Belief in Science, and
Emotion-based Reasoning. Further analysis investigated
differences on IPO-RT (Reality Testing) subscales.
RESULTS
Descriptive Statistics
Prior to analysis, data screening removed outliers. Four z-
scores marginally greater than 3.25 were transformed to the next
highest score (Tabachnick and Fidell, 2001). As Table 1 indicates,
indices of paranormal experiences and abilities (Paranormal
Experience, Paranormal Practitioner Visiting, and Paranormal
Ability) were positively intercorrelated. Additionally, Belief in the
Paranormal, Proneness to Reality Testing Deficits, and Emotion-
Based Reasoning correlated positively, whereas Belief in Science
was negatively associated with paranormal experiences and
abilities. Although, several of the relationships were significant,
they were in the small range (r=0.10 to 0.30) (see Cohen, 1992).
These represent more meaningful associations when interpreted
using the guidelines of Gignac and Szodorai (2016) (i.e., small, r
=0.11; medium, r=0.19, and large, r=0.29).
Latent Profile Analysis
For LPA model comparisons see Table 2. Initial assessment of 1-
class and 2-class models was undertaken. AIC, BIC, and ssaBIC
indices revealed the superior fit of the 2-class model. This
was supported by the LMR-A-LRT, which indicated significant
improvement over the 1-class model. Evaluation of 2-class and 3-
class solutions found that the 3-class solution was superior, due
to lower AIC, BIC, ssaBIC statistics, higher entropy (0.958 vs.
0.946), and a significant LMR-A-LRT p-value.
Next, the 4-class solution demonstrated superior fit to the
3-class solution; lower AIC, BIC, ssaBIC statistics, higher
entropy (0.975 vs. 0.958), and a significant LMR-A-LRT p-value.
Iteratively, the 5-class model indicated a significant improvement
over the 4-class solution, possessing lower AIC, BIC, ssaBIC
statistics, higher entropy (0.978 vs. 0.975), and a significant
LMR-A-LRT p-value. Subsequent analysis of a 6-class model
suggested superiority to the 5-class model, with lower AIC,
BIC, ssaBIC statistics, higher entropy (0.981 vs. 0.978), and a
significant LMR-A-LRT p. Finally, the 7-class solution indicated
no significant improvement over the 6-class model; hence,
the optimal solution was identified and there was no further
consideration of solutions.
Figure 1 provides a visual representation of relative scores on
Paranormal Experience, Paranormal Practitioner Visiting, and
Paranormal Ability. Table 3 displays the profiles organised in
sequence from higher to lower overall scores.
Average latent class probabilities for most likely latent class
membership was 0.972 for class 1, 0.997 for class 2, 0.971 for
class 3, 0.997 for class 4, 0.966 for class 5, and 0.999 for class 6,
indicating good overall discrimination.
Association of Latent Profiles With
Paranormal Belief, Reality Testing, Belief in
Science, and Emotion-Based Reasoning
Multivariate analysis of variance (MANOVA) examined the effect
of latent profile membership on the following outcome variables:
Belief in the Paranormal, Belief in Science, Proneness to Reality
Deficits, and Emotional-Based Reasoning (see Table 4). Analysis
revealed a significant main effect of group, Pillai’s trace =0.316,
F(20,3800) =16.314, p<0.001, η2=0.079 (medium effect size).
Significant effects for group were observed in relation to all
outcome variables.
Post-hoc pairwise comparisons with Bonferroni correction
(Table 4) revealed that the class 1 group (High Paranormal
Experience, intermediate Paranormal Practitioner Visiting
and Paranormal Ability) demonstrated high Belief in the
Paranormal, Proneness to Reality Testing, and lowest Belief in
Science. Correspondingly, class 6 (Low Paranormal Experience,
Paranormal Practitioner Visiting and Paranormal Ability)
reported considerably lower levels of Belief in the Paranormal,
Proneness to Reality Testing and Emotion-based reasoning, and
Frontiers in Psychology | www.frontiersin.org 7June 2021 | Volume 12 | Article 670959
Drinkwater et al. Latent Profile Analysis
TABLE 1 | Descriptive statistics and intercorrelations among all study variables.
Variable Mean SD 1 2 3 4 5 6 7 8 9 10 11
1. Paranormal Experience 0.55** 0.41** 0.48** 0.40** 0.24** 0.18** 0.38** 0.41** 0.24** 0.18**
2. Paranormal Practitioner Visiting 0.26** 0.34** 0.20** 0.13** 0.10* 0.19** 0.21** 0.13** 0.07*
3. Paranormal Ability 0.44** 0.45** 0.30** 0.15** 0.40** 0.45** 0.35** 0.19**
4. Paranormal Belief 174.48 54.85 0.52** 0.37** 0.34** 0.49** 0.52** 0.30** 0.33**
5. Reality Testing 42.46 14.22 0.51** 0.13** 0.91** 0.93** 0.81** 0.65**
6. Emotion-based Reasoning 8.39 2.23 0.23** 0.44** 0.51** 0.42** 0.26**
7. Belief in Science 38.82 11.37 0.14** 0.15** 0.06 0.03
8. Auditory and Visual Hallucinations 13.23 5.00 0.76** 0.64** 0.63**
9. Delusional Thinking 13.99 5.82 0.71** 0.48**
10. Social Deficits 7.61 3.22 0.40**
11. Confusion 8.41 2.65
*indicates p <0.05; **indicates p <0.001.
TABLE 2 | Fit of competing latent profile models.
Model AIC BIC ssaBIC LMR-A LMR-A p-value Entropy
1-class 10202.221 10231.403 10212.346
2-class 9180.811 9229.439 9197.679 693.230 <0.001 0.946
3-class 8748.389 8816.468 8772.004 424.942 0.008 0.958
4-class 8335.268 8422.798 8365.631 406.319 <0.001 0.975
5-class 7952.878 8079.309 7996.734 251.743 <0.001 0.978
6-class 7452.631 7559.612 7489.740 151.327 <0.001 0.981
7-class 7593.278 7739.160 7643.881 306.919 0.770 0.989
AIC, Akaike Information Criterion; BIC, Bayesian Information Criterion; ssaBIC, sample-
size adjusted BIC; LMR-A, Lo-Mendell-Rubin-adjusted likelihood ratio test.
higher Belief in Science. Across remaining profiles (classes 2, 3,
4, and 5), scores were moderate to intermediate. An exception
to this trend was Emotion-based reasoning, where class 2
(Moderately high Paranormal Experience and Paranormal
Ability, intermediate Paranormal Practitioner Visiting), and
class 4 (Moderately low Paranormal Experience, low Paranormal
Practitioner Visiting, and moderately high Paranormal Ability)
scored highest, though not in both instances significantly
different from class 1 (High Paranormal Experience, intermediate
Paranormal Practitioner Visiting and Paranormal Ability).
IPO-RT Subscale Analysis
Noting overall effects for IPO-RT, additional analysis using
MANOVA investigated subscales differences (Auditory and
Visual Hallucinations, Delusional Thinking, Social Deficits, and
Confusion) (Table 5). Analysis revealed a significant main effect
of group, Pillai’s trace =0.242, F(20,3800) =12.252, p<0.001,
η2=0.061 (medium effect size). Significant group effects were
also observed. Post-hoc pairwise comparisons with Bonferroni
correction (Table 4) indicated that class 6 (Low Paranormal
Experience, Paranormal Practitioner Visiting and Paranormal
Ability) scored significantly lower on Auditory and Visual
Hallucinations, Delusional Thinking, and Social Deficits. The
subfactor of Confusion was an exception to this trend.
Class 1 (High Paranormal Experience, intermediate
Paranormal Practitioner Visiting and Paranormal Ability)
reported significantly higher Auditory and Visual Hallucination
scores in comparison with all classes/groups but class 2
(Moderately high Paranormal Experience and Paranormal
Ability, intermediate Paranormal Practitioner Visiting), and
class 4 (Moderately low Paranormal Experience, low Paranormal
Practitioner Visiting, and moderately high Paranormal Ability).
In contrast with the general reality testing factor the highest
scores across the remaining subfactors occurred for class 2
(Moderately high Paranormal Experience and Paranormal
Ability, intermediate Paranormal Practitioner Visiting) and
class 4 Moderately low Paranormal Experience, low Paranormal
Practitioner Visiting, and moderately high Paranormal Ability)
(Social Deficits and Confusion).
DISCUSSION
Latent profile analysis (LPA) identified discrete classes that
categorised important variations in paranormal experience
and ability. These represented common differentiations
in the frequencies of Paranormal Experience, Paranormal
Practitioner Visiting, and Paranormal Ability. Accordingly,
each profile grouped individuals based on mutually exclusive
relationships between experiential indices. For instance, not
all experiencers visited paranormal practitioners, nor did they
profess supernatural ability. In addition, individuals reported
multiple experiences and visited paranormal practitioners, but
claimed little or no paranormal ability. Thus, classes provided
a nuanced categorisation of sample subpopulations based
on heterogeneous paranormal histories. This approach was
theoretically important because it acknowledged that people
accrue experience in quantitatively and qualitatively different
ways. An additional advantage of LPA was the ability to compare
emergent classes on levels of paranormal belief and measures of
thinking style.
Consistent with this notion that paranormal experience has
a broad and varied basis, zero-order correlations indicated that
Paranormal Experience, Paranormal Practitioner Visiting, and
Frontiers in Psychology | www.frontiersin.org 8June 2021 | Volume 12 | Article 670959
Drinkwater et al. Latent Profile Analysis
FIGURE 1 | Pattern of mean scores for Paranormal Experience, Paranormal Practitioner Visiting, and Paranormal Ability as a function of latent profile.
TABLE 3 | Profiles organised in sequence from higher to lower overall scores.
Grouping Percentage of
sample
Description
Class 1 3.6% (n=34) High Paranormal Experience, intermediate
Paranormal Practitioner Visiting and Paranormal
Ability
Class 2 4.6% (n=44) Moderately high Paranormal Experience and
Paranormal Ability, intermediate Paranormal
Practitioner Visiting
Class 3 7.1% (n=68) Intermediate Paranormal Experience and
Paranormal Practitioner Visiting, low Paranormal
Ability
Class 4 15.0% (n=144) Moderately low Paranormal Experience, low
Paranormal Practitioner Visiting, and moderately
high Paranormal Ability
Class 5 15.5% (n=148) Moderately low Paranormal Experience and
Ability, low Paranormal Practitioner Visiting
Class 6 54.2% (n=518) Low Paranormal Experience, Paranormal
Practitioner Visiting and Paranormal Ability
Paranormal Ability were related but relatively distinct indices.
Specifically, shared variance was relatively low: Experience
and Ability, 16%; Visiting and Ability, 19%; and Experience
and Visiting, 30%. This explains why, when the experiential
indices were combined, they produced a discrete series of
complex interactions. Acknowledging the phenomenological
sophistication of paranormal experiences is an important
academic contribution to the area because it extends the
conceptualisation of experience. Traditionally, research has
focused exclusively on the paranormal experience index, which
is defined as the personal ascription of supernatural powers
or forces to direct observations or conscious occurrences.
These characteristics are inherent within the standard research
operationalisation of subjective paranormal experiences (SPEs),
as the willingness to attribute supernatural causation to an event
or occurrence (Glicksohn, 1990). The fact that SPEs index only
this single experiential aspect explains, in part, why they share
only a small proportion of variance with belief in the paranormal.
This study demonstrated that supernatural credence is informed,
shaped, and reinforced by myriad life events beyond the restricted
remit of SPEs.
The finding that Paranormal Experience and Belief in the
Paranormal shared only 23% common variance illustrates this.
Other studies using a variety of equivalent belief measures have
reported similar figures (see Dagnall et al., 2016; Drinkwater et al.,
2020). Despite providing only a limited assessment of paranormal
experience, it is important to recognise that the study of SPEs
has historically acted as a useful indicator of individual proclivity
to interpret personal events as supernatural (Schouten, 1983,
1986; Neppe, 1984; Persinger and Valliant, 1985; Palmer and
Neppe, 2004; Schmied-Knittel and Schetsche, 2005; Simmonds-
Moore, 2016). Therefore, findings recommend the continued use
of SPEs in combination with other “additive” indices to produce
a broader measure of experience. In this context, an important
research development is the utilisation of profiles that classify a
range of heterogeneous paranormal phenomena.
Another related limitation of previous research is the use
of simplified measures. For example, Dagnall et al. (2016)
used dichotomies (i.e., experience vs. non-experience), and
categorical distinctions (i.e., single vs. multiple experiences 2-5
vs. more than 5). Relative to class profiles, these provide only
Frontiers in Psychology | www.frontiersin.org 9June 2021 | Volume 12 | Article 670959
Drinkwater et al. Latent Profile Analysis
TABLE 4 | The effects of group (latent profile) in relation to Paranormal Belief, Reality Testing, Belief in Science, and Emotion-based Reasoning.
Dependent variable
Paranormal belief Reality testing Belief in science Emotion-based reasoning
ANOVA MANOVA
Fdf (Sig.;η2)Fdf (Sig.;η2)Fdf (Sig.;η2)Fdf (Sig.;η2) Pillai’s trace Fdf (Sig.)η2
Variable
Group 62.50 5,950 (<0.001; 0.24) 2.89 5,950 (<0.001; 0.19) 79.04 5,950 (<0.001; 0.03) 53.01 5,950 (<0.001; 0.09) 0.31 16.31 20,3800 (<0.001) 0.07
Pairwise comparisons (mean differences) between classes
Class contrast Mean diff. (Sig.) Mean diff. (Sig.) Mean diff. (Sig.) Mean diff. (Sig.)
Class 1 vs. Class 2 5.26 (1.00) 0.67 (1.00) 2.49 (1.00) 0.35 (1.00)
Class 1 vs. Class 3 30.20 (0.034) 9.72 (0.004) 4.14 (1.00) 0.65 (1.00)
Class 1 vs. Class 4 22.15 (0.209) 0.31 (1.00) 2.63 (1.00) 0.44 (1.00)
Class 1 vs. Class 5 44.09 (<0.001) 7.32 (0.035) 4.94 (0.285) 0.29 (1.00)
Class 1 vs. Class 6 78.49 (<0.001) 15.59 (<0.001) 6.97 (0.006) 1.31 (0.006)
Class 2 vs. Class 3 24.93 (0.107) 9.05 (0.004) 1.65 (1.00) 1.00 (0.220)
Class 2 vs. Class 4 16.88 (0.629) 0.36 (1.00) 0.14 (1.00) 0.09 (1.00)
Class 2 vs. Class 5 38.82 (<0.001) 6.64 (0.040) 2.45 (1.00) 0.64 (1.00)
Class 2 vs. Class 6 73.22 (<0.001) 14.91 (<0.001) 4.48 (0.179) 1.67 (<0.001)
Class 3 vs. Class 4 8.04 (1.00) 9.41 (<0.001) 1.50 (1.00) 1.09 (0.006)
Class 3 vs. Class 5 13.89 (0.662) 2.40 (1.00) 0.79 (1.00) 0.35 (1.00)
Class 3 vs. Class 6 48.29 (<0.001) 5.86 (0.005) 2.83 (0.716) 0.66 (0.222)
Class 4 vs. Class 5 21.93 (0.001) 7.01 (<0.001) 2.30 (1.00) 0.73 (0.043)
Class 4 vs. Class 6 56.33 (<0.001) 15.28 (<0.001) 4.33 (0.001) 1.76 (<0.001)
Class 5 vs. Class 6 34.40 (<0.001) 8.27 (<0.001) 2.03 (0.747) 1.02 (<0.001)
Class 1 =High Paranormal Experience, intermediate Paranormal Practitioner Visiting and Paranormal Ability (n =34). Class 2 =Moderately high Paranormal Experience and Paranormal
Ability, intermediate Paranormal Practitioner Visiting (n =44). Class 3 =Intermediate Paranormal Experience and Paranormal Practitioner Visiting, low Paranormal Ability (n =68). Class
4=Moderately low Paranormal Experience, low Paranormal Practitioner Visiting, and moderately high Paranormal Ability (n =144). Class 5 =Moderately low Paranormal Experience
and Ability, low Paranormal Practitioner Visiting (n =148). Class 6 =Low Paranormal Experience, Paranormal Practitioner Visiting and Paranormal Ability (n =518).
snapshots of phenomenological influences. Furthermore, the use
of these measures as distinct indices ignores the observation
that they interrelate in complex ways. In the current paper, the
effects of Paranormal Experience were qualified by Paranormal
Practitioner Visiting and Paranormal Ability. This was evidenced
by differences between class profiles on observed measures.
Due to the number of comparisons, subsequent discussion
focuses on the main trends. Class 5 (Moderately low Paranormal
Experience and Ability, low Paranormal Practitioner Visiting)
scored higher than class 6 (Low Paranormal Experience,
Paranormal Practitioner Visiting and Paranormal Ability) on
Belief in the Paranormal, Proneness to Reality Testing Deficits,
and Emotion-Based Reasoning; there was no difference for
Belief in Science. This outcome suggests that even moderate
levels of intra class variation can heighten scores. Similarly,
class 4 (Moderately low Paranormal Experience, low Paranormal
Practitioner Visiting, and moderately high Paranormal Ability)
scored higher than class 5 on Belief in the Paranormal, Proneness
to Reality Testing Deficits, and Emotion-Based Reasoning.
Other comparisons revealed also noteworthy outcomes.
Class 4 scored higher Proneness to Reality Testing Deficits
and Emotion-Based Reasoning than class 3 (Intermediate
Paranormal Experience and Paranormal Practitioner Visiting,
low Paranormal Ability), and class 2 (Moderately high
Paranormal Experience and Paranormal Ability, intermediate
Paranormal Practitioner Visiting) demonstrated higher scores
on Proneness to Reality Testing Deficits than class 3. Finally,
no differences were observed between the two highest intensity
experience groups: Class 1 (High Paranormal Experience,
intermediate Paranormal Practitioner Visiting and Paranormal
Ability) vs. class 2. Collectively, results established that profile
membership subtly influenced scores on observed variables.
Focusing on outcome measures (see Table 4), significant
differences were typically observed for Proneness to Reality
Testing Deficits (11 out of 15 comparisons) and Belief in the
Paranormal (9 out of 15 comparisons). Significant differences
were observed also for Emotion-Based Reasoning (6 out of 15
comparisons). In contrast, Belief in Science comparisons revealed
only two significant differences. These findings suggest that
variations in the composition of paranormal experience profiles
are most likely to manifest as differences in Proneness to Reality
Testing deficits and Belief in the Paranormal. The observed
differences in Emotion-Based Reasoning, tentatively indicated
increased levels of affective driven decision-making in experience
groups (classes 5, 4, 2, and 1) relative to the low paranormal
experience group (class 6). However, this trend for Emotion-
Based Reasoning was less pronounced and therefore requires
cautious interpretation.
Collectively, findings supported the notion that greater
breadth and intensity of experiential factors were associated with
Frontiers in Psychology | www.frontiersin.org 10 June 2021 | Volume 12 | Article 670959
Drinkwater et al. Latent Profile Analysis
TABLE 5 | The effects of group (latent profile) in relation to Reality Testing subfactors.
Dependent variable
Auditory and visual
hallucinations
Delusional thinking Social deficits Confusion
ANOVA MANOVA
Fdf (Sig.;η2)Fdf (Sig.;η2)Fdf (Sig.;η2)Fdf (Sig.;η2) Pillai’s trace Fdf (Sig.)η2
Variable
Group 38.72 5,950 (<0.001; 0.17) 48.12 5,950 (<0.001; 0.20) 23.69 5,950 (<0.001; 0.11) 7.44 5,950 (<0.001; 0.04) 0.24 12.25 20,3800 (<0.001) 0.06
Pairwise comparisons (mean differences) between classes
Class contrast Mean diff. (Sig.) Mean diff. (Sig.) Mean diff. (Sig.) Mean diff. (Sig.)
Class 1 vs. Class 2 1.31 (1.00) 0.61 (1.00) 0.11 (1.00) 0.07 (1.00)
Class 1 vs. Class 3 3.44 (0.004) 4.10 (0.002) 1.54 (1.00) 0.54 (1.00)
Class 1 vs. Class 4 0.59 (1.00) 0.08 (1.00) 2.63 (0.221) 0.26 (1.00)
Class 1 vs. Class 5 2.83 (0.015) 2.62 (0.111) 1.30 (0.338) 0.05 (1.00)
Class 1 vs. Class 6 5.59 (<0.001) 6.15 (<0.001) 2.37 (<0.001) 0.95 (0.555)
Class 2 vs. Class 3 2.12 (0.251) 4.71 (<0.001) 1.65 (0.078) 0.46 (1.00)
Class 2 vs. Class 4 0.72 (1.00) 0.52 (1.00) 0.27 (1.00) 0.34 (1.00)
Class 2 vs. Class 5 1.51 (0.827) 3.23 (0.005) 1.42 (0.107) 0.13 (1.00)
Class 2 vs. Class 6 4.27 (<0.001) 6.76 (<0.001) 2.49 (<0.001) 0.87 (0.524)
Class 3 vs. Class 4 2.85 (<0.001) 4.18 (<0.001) 1.92 (<0.001) 0.80 (0.501)
Class 3 vs. Class 5 0.60 (1.00) 1.47 (0.752) 0.23 (1.00) 0.60 (1.00)
Class 3 vs. Class 6 2.15 (0.003) 2.05 (0.031) 0.83 (0.476) 0.40 (1.00)
Class 4 vs. Class 5 2.24 (<0.001) 2.70 (<0.001) 1.69 (<0.001) 0.20 (1.00)
Class 4 vs. Class 6 5.00 (<0.001) 6.23 (<0.001) 2.76 (<0.001) 1.21 (<0.001)
Class 5 vs. Class 6 2.76 (<0.001) 3.53 (<0.001) 1.07 (0.002) 1.01 (<0.001)
Class 1 =High Paranormal Experience, intermediate Paranormal Practitioner Visiting and Paranormal Ability (n =34). Class 2 =Moderately high Paranormal Experience and Paranormal
Ability, intermediate Paranormal Practitioner Visiting (n =44). Class 3 =Intermediate Paranormal Experience and Paranormal Practitioner Visiting, low Paranormal Ability (n =68). Class
4=Moderately low Paranormal Experience, low Paranormal Practitioner Visiting, and moderately high Paranormal Ability (n =144). Class 5 =Moderately low Paranormal Experience
and Ability, low Paranormal Practitioner Visiting (n =148). Class 6 =Low Paranormal Experience, Paranormal Practitioner Visiting and Paranormal Ability (n =518).
higher Belief in the Paranormal, Proneness to Reality Testing
Deficits and Emotion-Based Reasoning. In contrast, Belief in
Science was less sensitive to experiential variations; only small
differences were observed between profiles.
The effect of profile membership on Proneness to Reality
Testing Deficits was demonstrated further by inter class
comparisons of IPO-RT subscales. These revealed that class 6
(Low Paranormal Experience, Paranormal Practitioner Visiting
and Paranormal Ability) scored significantly lower on Auditory
and Visual Hallucinations, Delusional Thinking, and Social
Deficits than the other classes. Additionally, class 1 (High
Paranormal Experience, intermediate Paranormal Practitioner
Visiting and Paranormal Ability), class 2 (Moderately high
Paranormal Experience and Paranormal Ability, intermediate
Paranormal Practitioner Visiting), and class 4 (Moderately low
Paranormal Experience, low Paranormal Practitioner Visiting,
and moderately high Paranormal Ability) scored relatively
higher than class 3 (Intermediate Paranormal Experience and
Paranormal Practitioner Visiting, low Paranormal Ability) and
class 5 (Moderately low Paranormal Experience and Ability, low
Paranormal Practitioner Visiting). Most differences occurred for
Auditory and Visual Hallucinations and Delusional Thinking.
Overall, IPO-RT subscale comparisons showed that experiential
profiles influenced levels of intrapsychic activity in subtle and
intricate ways, especially Auditory and Visual Hallucinations and
Delusional Thinking.
With reference to previous research, findings accord with
the typically reported positive relationship between experience
in and belief of the paranormal (Glicksohn, 1990; Musch and
Ehrenberg, 2002; Dagnall et al., 2016). Additionally, results
indicated that the two constructs interact in intricate ways.
Particularly, via a combination of constructionist (Irwin et al.,
2013), cultural (Hufford, 1982; McClenon, 1994), and existential
factors (Bennett, 1987) that vary within individuals. Thus,
profile differences were consistent with the interpretation that
experiences stimulate interest and belief in the supernatural (van
Elk, 2017), and that beliefs reciprocally encourage the search for
confirmatory personal paranormal occurrences (Dagnall et al.,
2015a, 2020). Psychologically, this perspective aligns with the
concept of worldview, precisely the notion that experiences
inform and are comprehended within an overarching cognitive
framework that makes the world intellectually coherent and
meaningful (Overton, 1991; Miller and West, 1993; Koltko-
Rivera, 2004; Dagnall et al., 2015b). Although, it is important to
note that experiential factors are not necessarily antecedent to
belief. This dynamic synergy is encapsulated within Van Leeuwen
and van Elk’s (2019) Interactive Religious Experience Model and
generalises well to the paranormal experience-belief relationship.
Frontiers in Psychology | www.frontiersin.org 11 June 2021 | Volume 12 | Article 670959
Drinkwater et al. Latent Profile Analysis
Results were consistent also with preceding work on thinking
styles to the extent that profiles with greater levels of paranormal
experience and ability tended to score higher on Proneness to
Reality Testing Deficits. Explicitly, the notion that attribution of
paranormality is associated with intuitive-experiential processing
(Wolfradt and Watzke, 1999; Wolfradt et al., 1999; Irwin and
Wilson, 2013). Although, this supposition derives from the use of
the IPO-RT, which is an indirect measure, the conclusion appears
sound since the IPO-RT has become an established index of
intuitive-experiential processing/thinking (Dagnall et al., 2017a,
2019; Denovan et al., 2017, 2020).
Rational-analytical processing, in the form of Belief in Science,
did not differ consistently as a function of profile membership.
While, this finding concurred with Irwin and Wilson’s (2013)
conclusion that rational thinking was not a significant predictor
of either proneness to anomalous experiences or paranormal
attributions, it requires qualification. Principally because few
studies to date have employed BISS as an index of rational
thinking. This means that the effectiveness of BISS to assess
critical, evaluative processing remains largely unattested. Indeed,
Dagnall et al. (2019) found that only BISS scores above the
median (second quartile) produced a reduction in experiential-
based thinking. This suggests that only higher levels of belief
in science are associated with rational processing. Explicitly,
that elevated belief in science reflects greater comprehension
of its strengths (and limitations). This tentative conclusion is
supported by the observation that a significant difference was
observed between class 1 (high paranormal experiential factors)
vs. class 6 (low paranormal experiential factors).
Noting this issue, future research should employ a range
of critical thinking measures to determine which best predicts
paranormal experience. Researchers have previously used
this approach to clarify conflicting findings in the domain
of paranormal belief and reasoning (Dagnall et al., 2007).
Additionally, investigators could utilise direct, objective
measures of rational thinking (see Pennycook et al., 2012).
Clearly, additional work is required in this area, especially as
rational processing appears differentially related to paranormal
experience and belief (Irwin and Wilson, 2013), and it remains
unclear whether belief in science is an ineffective index of
critical thinking.
The observation that higher Emotion-Based Reasoning was
concomitant with increased levels of Belief in the Paranormal
and Proneness to Reality Testing Deficits, offers tentative support
for the notion that paranormal attributions represent a form of
non-psychotic delusions within general populations. Explicitly,
that attribution of paranormal experiences and abilities derives
from persistent ideas based on emotional appeal, which persevere
without empirical support and maintain despite the existence
of conflicting evidence (Irwin et al., 2012). However, it is
important to note that there were subtle variations across class
comparisons that suggest that generally emotion-based reasoning
is a contributing, rather than defining factor.
Lastly, to advance understanding about the processes
associated with belief, subsequent studies could focus on
preselecting different types of paranormal experience/belief
groupings to examine how they differ on cognitive-perceptual
factors associated with endorsement of scientifically
unsubstantiated beliefs (e.g., schizotypy, proneness to
hallucinations, and delusional ideation). A further elaboration
related to variations in paranormal experience/belief could
involve investigating how interactions influence health-
related factors such as life satisfaction, mental well-being, and
psychological outlook.
Limitations
Outcomes further illustrated the usefulness of mixture models
(i.e., profile analyses/latent class) in social sciences (McLachlan
and Peel, 2000). Particularly, that they reveal discrete populations
of interest as a function of responses to sets of items (Whittaker
and Miller, 2020). Regarding the combined experiential indices
within this study, the sample comprised several groups.
Nevertheless, it is important to note that classes formed via LPA
possess relative differences that may not necessarily represent
intrinsic meaning or clinical importance (Achterhof et al., 2019).
The authors are aware, despite criticising previous work
for using limited indices of paranormal experience, that the
measures used in the present study were also relatively
restricted in scope and breadth. Explicitly, this paper focused
on psychic phenomena and considers only the influence of
Paranormal Experience, Paranormal Practitioner Visiting and
Perceived Paranormal Ability. Despite being a core element of
parapsychology, psychic phenomena reference only a restricted
range of subject matter. Other commonly studied facets are
religious belief, witchcraft, superstition, psychokinesis, ghosts
and haunting, near death experiences, and out of body
experiences (see Dagnall et al., 2010a).
Though, there is overlap between psychic phenomena, via
the concepts of spiritualism and life after death, other facets
such as haunting/ghosts are associated with relatively commonly
reported experiences and high levels of endorsement. Hence,
future research needs to establish whether similar experiential
profiles exist for other important paranormal phenomena.
Intuition suggests that the different contextual nature of
experiences and their inherent plausibility is likely to produce
divergent classes. For example, given that ghost activity is
typically context-based and draws less on perceptions of ability,
it is likely that emergent classes would differ to those produce for
psychic powers.
Regarding extent, while the present study used a wider range
of experiential indices than typical studies, the phenomena
selected failed to cover the full array of paranormal happenings.
Thus, subsequent studies should examine the contribution of
additional factors to determine how these influence classes and
more generally best predict related factors (i.e., belief in the
paranormal). These could include life history (e.g., indirect
experiences via accounts of others), social identity (e.g., mixing
with experiencers in contexts such as clubs and societies), and
involvement with paranormal media (e.g., watching paranormal
films and programmes). Although, the outlined “experiences” are
indirect as opposed to the direct observations used in this article,
they still represent important experiential factors that influence
an individual’s beliefs and processing style preferences.
Frontiers in Psychology | www.frontiersin.org 12 June 2021 | Volume 12 | Article 670959
Drinkwater et al. Latent Profile Analysis
A concern with LPA is that recoding of continuous data to
produce categorical variables (classes) may result in information
loss (Lanza and Rhoades, 2013). Furthermore, LPA can generate
profiles that are statistically sound but conceptually ill defined.
This was less of a concern in the present paper since the approach
was exploratory. Accordingly, the emergent profiles were of
theoretical significance because they established the multifaceted,
heterogeneric nature of paranormal experience. Explicitly, they
indicated that experiential factors combine to effect beliefs and
thinking style in intricate and subtle ways. The authors were
aware of the descriptive nature of the emergent classes and that
additional work is required. Hence, researchers should focus
on the improvement of extant indices and the addition of
supplementary factors (e.g., family and peer influences). This
iterative process is essential for class development and will
facilitate profile operationalisation.
A higher number of females compared to males were
recruited. Although, this imbalance was commensurate with
related research (e.g., Hergovich and Arendasy, 2005; Drinkwater
et al., 2012) it potentially limits the generalisability of findings
because some studies have previously reported gender differences
in paranormal belief (Dag, 1999,Irwin, 1993). Therefore,
future research should explore with gender influences profile
membership and composition.
Finally, it is important to acknowledge that LPA categories
reflect heterogeneity across model dimensions, not classifications
of individuals present within the population (Lanza and Rhoades,
2013). Thus, misspecification can occur in the form of identifying
too few or too many classes. To limit this, ensuing research
could use cross-validation methods, such as double cross-
validation (Collins et al., 1994) or progressive elaboration (see
Donovan and Chung, 2015). These approaches objectively assess
model fit and help to establish class stability. In the case
of cross-validation, this does however only provide the best
approximation to the true model (Collins et al., 1994).
DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be
made available by the authors, without undue reservation.
ETHICS STATEMENT
The studies involving human participants were reviewed and
approved by Manchester Metropolitan University Faculty of
Health, Psychology and Social Care Ethics Committee (October
2018). Written informed consent for participation was not
required for this study in accordance with the national legislation
and the institutional requirements.
AUTHOR CONTRIBUTIONS
KD focused theoretically, collected the data, analysed the data,
developed the article, and reviewed the draft. ND and AD focused
theoretically, analysed the data, and developed the article. CW
reviewed the draft. All authors contributed to the article and
approved the submitted version.
ACKNOWLEDGMENTS
We would like to thank the BIAL Foundation (Project ID:
082/2018) for their support with this research.
REFERENCES
Achterhof, R., Huntjens, R. J., Meewisse, M. L., and Kiers, H. A. (2019).
Assessing the application of latent class and latent profile analysis
for evaluating the construct validity of complex posttraumatic stress
disorder: cautions and limitations. Eur. J. Psychotraumatol. 10:1698223.
doi: 10.1080/20008198.2019.1698223
Akaike, H. (1987). Factor analysis and AIC. Psychometrika 52, 317–332.
American Psychiatric Association (2013). Diagnostic and Statistical Manual of
Mental Disorders: DSM-5. Arlington, VA: American Psychiatric Association.
Barron, D., Furnham, A., Weis, L., Morgan, K. D., Towell, T., and
Swami, V. (2018). The relationship between schizotypal facets and
conspiracist beliefs via cognitive processes. Psychiatry Res. 259, 15–20.
doi: 10.1016/j.psychres.2017.10.001
Beck, A. T., Rector, N. A., Stolar, N., and Grant, P. (2008). Schizophrenia: Cognitive
Theory, Research, and Therapy. New York, NY: Guilford Press.
Bennett, G. (1987). Traditions of Belief: Women and the Supernatural.
London: Penguin.
Blackmore, S. J. (1984). A postal survey of OBEs and other experiences. J. Soc.
Psychical Res. 52, 225–244.
Byrne, B. M. (2010). Structural Equation Modeling With AMOS: Basic Concepts,
Applications, and Programming. New York, NY: Routledge.
Castro, M., Burrows, R., and Wooffitt, R. (2014). The paranormal is
(still) normal: the sociological implications of a survey of paranormal
experiences in Great Britain. Sociol. Res. Online 19, 30–44. doi: 10.5153/sr
o.3355
Cella, M., Vellante, M., and Preti, A. (2012). How psychotic-like are
paranormal beliefs? J. Behav. Ther. Exp. Psychiatry 43, 897–900.
doi: 10.1016/j.jbtep.2012.01.003
Cohen, J. (1992). A power primer. Psychol. Bull. 112, 155–159.
doi: 10.1037/0033-2909.112.1.155
Collins, L. M., Graham, J. W., Long, J. D., and Hansen, W. B. (1994).
Crossvalidation of latent class models of early substance use onset. Multivariate
Behav. Res. 29, 165–183. doi: 10.1207/s15327906mbr2902_3
Coltheart, M., Menzies, P., and Sutton, J. (2010). Abductive
inference and delusional belief. Cogn. Neuropsychiatry 15, 261–287.
doi: 10.1080/13546800903439120
Dag, I. (1999). The relationships among paranormal beliefs, locus of control and
psychopathology in a Turkish college sample. Pers. Individ. Dif. 26, 723–737.
doi: 10.1016/S0191-8869(98)00184-6
Dagnall, N., Denovan, A., Drinkwater, K., Parker, A., and Clough, P. J. (2017a).
Urban legends and paranormal beliefs: the role of reality testing and schizotypy.
Front. Psychol. 8:942. doi: 10.3389/fpsyg.2017.00942
Dagnall, N., Denovan, A., Drinkwater, K. G., and Parker, A. (2019).
An evaluation of the Belief in Science Scale. Front. Psychol. 10:861.
doi: 10.3389/fpsyg.2019.00861
Dagnall, N., Denovan, A., Parker, A., Drinkwater, K., and Walsh, R. S. (2018).