Validation of the English and French versions of the Community Assessment of Psychic Experiences (CAPE) with a Montreal community sample.

Abstract

The aim of this study was to examine the reliability, validity and factor structure of the Community Assessment of Psychic Experiences (CAPE), a 42-item self-report questionnaire. We analyzed the internal consistency of the CAPE to determine whether the 3-factor structure (positive, negative and depressive symptoms) found by the CAPE authors would also be found in our sample.
A sample of 2275 individuals from the general community in the Montreal area completed the questionnaire in either French or English.
The internal consistencies of the original three subscales were good and the confirmatory factor models had a good fit. The exploratory factor analysis suggested a 3-5-factor solution, without improving the alternative factor structures. The 4-factor solution separated positive symptoms into factors we called 'bizarre positive symptoms' and 'social delusions', and the 5-factor solution separated positive symptoms further and included a 'popular psychic beliefs' factor. Results suggest that the scalability might be improved by shortening the original questionnaire to 23 items with the same 3 original scales.
We support the internal consistency of the CAPE. Although alternative scaling (4 and 5 factors) did not improve the model fit, researchers interested in distinguishing 3 factors of positive symptoms could find utility in these two new scales. Finally, reducing the number of CAPE items could be useful for shorter surveys. Future studies should test the implications of these suggestions.

Full text

Page 1

Validation of the English and French versions of the Community
Assessment of Psychic Experiences (CAPE) with a
Montreal community sample
Karène Brennerb, Norbert Schmitzc, Nicole Pawliukd, Ferid Fathallid, Ridha Jooberc,
Antonio Ciampie, Suzanne Kinga,⁎
aMcGill University Douglas Research Institute Pavilion Perry, Rm. E-3131 6875 LaSalle Blvd., Verdun Montreal, Québec, H4H 1R3 Canada
bDepartment of Psychology, University of Montreal, Canada
cDepartment of Psychiatry, McGill University, Douglas Hospital Research Centre, Canada
dDouglas Hospital Research Centre, Quebec, Canada
eDepartment of Epidemiology and Biostatistics, McGill University, Canada
Received 12 January 2007; received in revised form 15 June 2007; accepted 19 June 2007
Abstract
Objective: The aim of this study was to examine the reliability, validity and factor structure of the Community Assessment of
Psychic Experiences (CAPE), a 42-item self-report questionnaire. We analyzed the internal consistency of the CAPE to determine
whether the 3-factor structure (positive, negative and depressive symptoms) found by the CAPE authors would also be found in our
sample.
Method: A sample of 2275 individuals from the general community in the Montreal area completed the questionnaire in either
French or English.
Results: The internal consistencies of the original three subscales were good and the confirmatory factor models had a good fit. The
exploratory factor analysis suggested a 3–5-factor solution, without improving the alternative factor structures. The 4-factor
solution separated positive symptoms into factors we called ‘bizarre positive symptoms’ and ‘social delusions’, and the 5-factor
solution separated positive symptoms further and included a ‘popular psychic beliefs’ factor. Results suggest that the scalability
might be improved by shortening the original questionnaire to 23 items with the same 3 original scales.
Conclusion: We support the internal consistency of the CAPE. Although alternative scaling (4 and 5 factors) did not improve the
model fit, researchers interested in distinguishing 3 factors of positive symptoms could find utility in these two new scales. Finally,
reducing the number of CAPE items could be useful for shorter surveys. Future studies should test the implications of these
suggestions.
© 2007 Elsevier B.V. All rights reserved.
Keywords: Psychosis; CAPE; General community; Questionnaire; Validity
1. Introduction
Schizophrenia research tends to explore psychotic
symptoms primarily within populations of the severely
mentally ill. Yet, a significant body of research has
Schizophrenia Research 95 (2007) 86–95
www.elsevier.com/locate/schres
⁎Corresponding author. Tel.: +1 514 7616131 X2353; fax: +1 514
762 3049.
E-mail address: suzanne.king@douglas.mcgill.ca (S. King).
0920-9964/$ - see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.schres.2007.06.017

Page 2

demonstrated that the severity of psychotic symptoms
appears to lie on a continuum ranging from a severely
ill population to individuals with schizotypal person-
ality disorder, to those from the general population who
do not have a diagnosis. Schizotypy shares many
elements of schizophrenia (Chapman and Chapman,
1987; Chapman et al., 1976; Claridge, 1972; Claridge,
1987; Jackson and Claridge, 1991). Vollema and
Hoijtink (2000) confirm that schizotypy in a commu-
nity sample includes three components corresponding
to the three-factor model of schizophrenia: positive and
negative schizotypy (or symptoms) and cognitive
disorganization. Using the Composite Schizotypal
Traits Questionnaire (CSTQ), Bentall et al. (1989)
found a similar structure of schizotypy in the general
population with three main factors, and possibly four.
The first reflects active or positive psychotic symptoms,
like hallucinations, perceptual aberrations and magical
ideation, borderline personality, hypomania and delu-
sions. The second factor includes social anxiety, with
characteristics involvingneuroticism,and somecognitive
disorganization due to attentional difficulties and dis-
tractibility. The third factor captures anhedonia, with
underlying negative psychotic symptoms. The fourth
factor reflectsa formof disinhibitedor asocial schizotypy
(including hypomanic traits). In other studies of commu-
nity samples, Mason et al. (1995) and Mason and
Claridge (2006) obtained similar findings with a short
form of the CSTQ called the Oxford-Liverpool Inventory
of Feelings and Experiences (O-LIFE). This question-
naire has 4 scales: unusual experiences (or positive
symptomatology), cognitive disorganization, introvertive
anhedonia (or negative symptomatology), and impulsive
non-conformity. Thus, there is a consensus in the
literature highlighting the striking similarities between
the multidimensionality of schizotypal traits and the
multidimensionality of schizophrenia, sharing three
major factors: positive psychotic symptoms, negative
psychotic symptoms, and thought disorder (Vollema and
van den Bosch, 1995); a fourth factor of more excited,
aggressive traits may also describe both schizophrenic
and schizotypal traits.
There is some suggestion that the positive symptom
dimension may warrant additional distinctions. A re-
cent longitudinal community study conducted with the
Symptom Checklist 90-R (SCL-90-R) and a semi-
structured interview, focused solely on paranoid idea-
tion and psychoticism, and revealed 2 distinct symptom
dimensions of subclinical psychosis (Rossler et al.,
2007). The first dimension reflects attenuated forms of
the nuclear symptoms of schizophrenia, like delusions
of control, auditory hallucinations, thought broadcasting
and thought intrusion, overlapping with the concept of
Schneiderian first rank symptoms. The second dimen-
sion reflects schizotypal signs like paranoid ideation as
well as ‘feeling lonely even when with people’ and
‘never feeling close to another person’.
Given similar factors of symptoms in schizophrenia
and schizotypal personality within community samples,
studies show that the distributions of scores for healthy
and psychotic populations have considerable overlap
(Peters et al., 1999a, 2004, 1999b). Using a self-report
questionnaire, Stefanis et al. (2002) also observed that a
non-negligible fraction of the general population expe-
riences positive symptoms of psychosis, even without
having the clinical disorder. van Os et al. (2000) re-
ported that up to 17% of the general population has
experienced at least one hallucination when interviewed
with the CIDI, and Poulton et al. (2000) reported that up
to 20% of 26 year-olds have experienced at least one
delusion or one hallucination in their lifetime when
interviewed with the DIS.
Stefanis et al. (2002) developed a self-report instru-
ment for assessing the frequency of clinical symptoms,
and severity of symptom-related distress, in the general
population. The Community Assessment of Psychic
Experiences, the CAPE, is a 42-item self-report instru-
ment, which evaluates three major clusters of symp-
toms seen in severe mental illness: positive, negative
and depressive. They found that the factor structure of
the CAPE supported these three dimensions in the
general population. Convergent validity of the CAPE
scales was supported by correlations with other,
established scales of depression (the SCL-90, Donias
et al., 1991), of positive symptoms (the PAS, Chapman
et al., 1978; and the SCL-90 Paranoia subscale) and of
negative symptoms (items from the SPQ, Raine, 1991).
The authors concluded that psychosis can be seen as a
continuum of variation in the three independent symp-
tom dimensions. The CAPE authors have also shown
overlap in scores between schizophrenic and non-
clinical samples: although schizophrenia patients have
a higher mean score on the CAPE psychotic symptom
factor (M=1.78; SD=0.56) than the control group
(M=1.30; SD=0.25), there is considerable overlap in
their ranges (1–3.8 vs. 1–2.8) indicating that some
non-clinical subjects have a greater frequency of psy-
chotic experiences than do some patients with schizo-
phrenia (Hanssen et al., 2003).
The CAPE differs from the schizotypy instruments
mentioned above. First, the CAPE aspires to evaluate
clinical psychotic symptoms, rather than attenuated
forms of symptoms as seen in schizotypy scales. Sec-
ond, the CAPE includes a depression scale, which is
87K. Brenner et al. / Schizophrenia Research 95 (2007) 86–95

Page 3

excluded from schizotypy scales, because the authors
believe that affective symptoms form an integral part
of schizophrenia and other psychotic syndromes.
Third, the CAPE does not assess mania or conceptual
disorganization, since these are less amenable to self-
report (Stefanis et al., 2002).
The CAPE's factor structure was first validated in
Greek with 932 young men undergoing basic training in
the Greek air force (Stefanis et al., 2002). Since then,
there have been three other validation studies. The first,
in France, included 571 women (mean age=19.8; range
18–51) (Verdoux et al., 2003). The second, in the
Netherlands, studied 647 community controls (38.4%
male, mean age of 46.5 years) and 415 psychiatric
patients (36.8% male, mean age of 40.5 years) with the
age range being 18–70 for both groups (Hanssen et al.,
2003). And, finally, a second study in Greece with 3500
19 year-olds (45% male) (Stefanis et al., 2004). The
authors of these validation studies all collaborate within
the research group that created the CAPE. It would be
important, therefore, to validate the instrument outside
of this research group, in its English version, with a large
sample of both men and women.
Thus, the aim of the current study was to examine
the reliability, validity and factor structure of the
English and French versions of the CAPE in a non-
European setting. More specifically, our main objec-
tives were: 1) to reconfirm the fit of the original factor
structure, 2) to determine if an alternative factor
structure would provide a better fit, and 3) to determine
if there is a more parsimonious subset of items in order
to have a shortened, but equally valid version of the
CAPE.
2. Method
2.1. Participants
Seven research assistants approached 3208 indivi-
duals who appeared to be between the ages of 18 and
40 years in public places in Montreal, such as mall food
courts, in parks, on street corners, etc. The true age of
the subjects was determined by their reported date of
birth. The recruitment phase was completed between
June 2003 and May 2004. 80% of the subjects were
recruited in the summertime, when people are easier to
approach in streets and parks. Potential participants
were told that if they completed the 15-minute ques-
tionnaire, and provided valid contact information, they
had 1 chance in 25 to win $50, which was sent to the
winners by mail. They were told that the questionnaire
measured the frequency of ‘psychic experiences’ in the
general population. 2470 individuals responded, for a
participation rate of 77%. Of the 2470 respondents, data
from 12 were dropped because they answered part of the
questionnaire in French and part in English making it
impossible to categorize their data as belonging to either
language version. Data from an additional 118 partici-
pants were dropped because their questionnaires were
incomplete, and from 65 participants who were found to
be younger than 18 years old given that our research
ethics board approval was for 18 years old and older.
This left data from a final sample of 2275 subjects for
analysis.
2.2. The instrument
The CAPE (http://www.cape42.homestead.com/
index.html) was derived from a combination of the
Peters Delusions Inventory (the PDI, Peters et al.,
1999b) with some modifications in the formulation of
certain items, and some items were added from the
SANS (Andreasen, 1989) and the SENS (Selten et al.,
1998). The CAPE has 42 symptom items covering three
symptom dimensions: positive symptoms, depressive
symptoms and negative symptoms. It uses a 4-point
Likert scale (0 to 3) to indicate symptom frequency
(“Never”, “Sometimes”, “Often” and “Nearly always”),
and a 4-point scale to indicate degree of distress expe-
rienced due to the symptom, if present (“Not distressed”,
“A bit distressed”, “Quite distressed”, and “Very
distressed”). We used the French version for Franco-
phone participants and the English version for Anglo-
phones. Several demographic questions were added to
the questionnaire.
2.3. Statistical analyses
2.3.1. Confirming the original factor structure
We calculated Cronbach's alpha to estimate the
reliability of the three composite scores. Alpha coeffi-
cients greater than 0.70 are considered indicative of an
acceptable level of internal consistency.
Based upon the published scoring procedures for the
three scales, we conducted confirmatory factor analy-
ses for the total sample, and for the English and French
questionnaires separately using the CALIS procedure
available in SAS for Windows, release 8. This pro-
cedure uses maximum likelihood estimation to solve a
series of regression equations simultaneously that pro-
duces an estimated covariance matrix. Maximum like-
lihood estimation tends to be robust to departures from
multivariate normality (Huba and Harlow, 1987). We
calculated several practical goodness of fit indices,
88 K. Brenner et al. / Schizophrenia Research 95 (2007) 86–95

Page 4

including the Comparative Fit Index (CFI), the
Goodness of Fit Index (GFI), the Adjusted Goodness
of Fit Index (AGFI), as well as the root mean square
residual (RMR). These indexes compare the observed
sample covariance matrix against the matrix estimated
from the model relative to a null model. Goodness of
fit indices (GFI, AGFI, CFI) of 0.90 or greater, and
RMR of less than 0.05, are considered to indicate good
fit. The hypothesized factor models were based on
three assumptions. First, we assumed that items con-
tributed information to one, and only one, factor in the
model. Second, the factors were allowed to be cor-
related. Finally, the variance of each factor was fixed
at 1.0.
Before conducting any analyses, the full sample was
divided into two random samples of equal size. In the
first step, we tested the original factor structure using the
first random subsample. We fit a three-factor model that
corresponds to the CAPE composites recommended by
the authors.
2.3.2. Exploring alternative factor structures
In the second step, using data from the second
random subsample, exploratory factor analyses were
performed for all of the subjects, then for each language
separately, in order to arrive at a factor model that
optimally accounts for the data.
In the third step, the resulting factor model was tested
by employing confirmatory factor analyses using data
from the same second half of the sample.
We provided goodness of fit indices to determine
whether the original model of the instrument should be
retained as is, or whether it could be improved upon.
2.3.3. Seeking parsimony
Non-parametric item response analysis of the data
structure was performed by using the Mokken analysis
which is less restrictive than the well-known Rasch
model (Meijer et al., 1990). Using the assumptions of
item response theory (Embretson and Reise, 2000), one
seeks to measure a latent or unobserved variable. In the
Table 1
Internal consistency of the three CAPE scales as measured by
Cronbach's coefficient alpha
All subjects
(n=2275)
English
(n=532)
French
(n=1743)
Positive
Negative
Depression
0.822
0.805
0.795
0.828
0.808
0.834
0.822
0.803
0.775
Table 2
Factorloadingsandgoodnessoffitindices obtainedfromconfirmatory
factor analysis for the whole sample
All subjects
(n=2275)
English
(n=532)
French
(n=1743)
Factor 1: Positive symptoms
Q2. Double meaning
Q5. Messages from TV
Q6. False appearance
Q7. Being persecuted
Q10. Conspiracy
Q11. Being important
Q13. Being special
Q15. Telepathy
Q17. Influenced by devices
Q20. Voodoo
Q22. Odd looks
Q24. Thought withdrawal
Q26. Thought insertion
Q28. Thought broadcasting
Q30. Thought echo
Q31. External control
Q33. Verbal hallucinations
Q34. Voices conversing
Q41. Capgras
Q42. Visual hallucinations
0.457
0.405
0.399
0.453
0.489
0.387
0.434
0.481
0.374
0.455
0.465
0.490
0.525
0.521
0.522
0.525
0.486
0.435
0.401
0.458
0.504
0.438
0.390
0.490
0.504
0.357
0.372
0.521
0.332
0.459
0.447
0.499
0.568
0.575
0.569
0.559
0.506
0.433
0.429
0.456
0.438
0.400
0.404
0.441
0.486
0.398
0.454
0.470
0.428
0.459
0.474
0.488
0.514
0.504
0.508
0.514
0.480
0.439
0.394
0.460
Factor 2: Negative symptoms
Q3. Lack of enthusiasm
Q4. Not talkative
Q8. No emotion
Q16. No interest in others
Q18. Influenced by devices
Q21. No energy
Q23. Empty mind
Q25. Lack of activity
Q27. Blunted feelings
Q29. Lack of spontaneity
Q32. Blunted emotions
Q35. Lack of hygiene
Q36. Unable to terminate
Q37. Lack of hobby
0.481
0.438
0.412
0.469
0.564
0.525
0.451
0.503
0.478
0.450
0.454
0.436
0.546
0.470
0.435
0.448
0.414
0.482
0.573
0.572
0.436
0.526
0.480
0.449
0.515
0.419
0.565
0.466
0.494
0.429
0.408
0.464
0.563
0.507
0.454
0.500
0.474
0.449
0.437
0.441
0.538
0.479
Factor 3: Depression
Q1. Sad
Q9. Pessimism
Q12. No future
Q14. Not worth living
Q19. Frequently cry
Q38. Guilty
Q39. Failure
Q40. Feeling tense
0.560
0.519
0.556
0.574
0.455
0.560
0.583
0.518
0.603
0.584
0.587
0.605
0.448
0.553
0.625
0.567
0.540
0.511
0.545
0.559
0.447
0.556
0.560
0.493
Fit index
Goodness of Fit Index
Adjusted Goodness of Fit Index
Comparative Fit Index
Root mean square residual
0.862
0.855
0.739
0.062
0.803
0.793
0.703
0.075
0.860
0.853
0.741
0.062
89K. Brenner et al. / Schizophrenia Research 95 (2007) 86–95

Page 5

present analysis, the three CAPE subscales were re-
garded as underlying, latent constructs.
Due to its probabilistic character, the model allows
deviations to a certain extent. Mokken scale analysis
produces a scale coefficient, Loevinger's H, to test the
model fit. Scalability coefficient H is a global indicator
of the degree to which subjects can be accurately
ordered on the latent trait by means of their sum score.
Higher values for H imply fewer violations, and thus a
better hierarchy, with a value ≥50 suggestive of a strong
hierarchical scale, with values of H between 0.40 and
0.50 reflecting medium scalability, and with values be-
tween 0.30 and 0.40 being weak.
The Mokken scale analysis may be used to test
predefined scales or to search for scales within a set of
items. We used the Mokken scale analysis in two ways.
First, we evaluated the unidimensionality of the three
CAPE scales by calculating the scalability coefficients
H. Second, we evaluated whether the unidimensionality
of the three subscales could be improved by reducing
the number of items. Using a bottom-up procedure, we
attempted to identify the items that maximize the H-
value.
3. Results
The final sample included 2275 participants (49.5%
men). Their ages ranged from 18 to 59 years (M=26.0;
SD=6.4). Regarding education, 3% of the sample had
not completed high school, 8% completed high school,
30% completed some post-high school education or had
a CEGEP (junior college) degree, 20% completed some
university, 22% were university graduates and 17%
completed post-graduate studies. All of the participants
were living in the greater Montreal region. CAPE scores
were highest for depressive symptoms (M=2.0, SD=
0.4), then for negative symptoms (M=1.8, SD=0.4) and
were the lowest for positive symptoms (M=1.6,
SD=0.3). A mean of 1 refers to having had the expe-
riences, on average, ‘sometimes’, and a mean of 2 refers
to having had the experience, on average, ‘often’.
3.1. Confirming original factor structure
3.1.1. Internal consistency
Table 1 presents the Cronbach's alpha statistics
for the French, English, and combined samples. The
internal consistencies of all of the original three sub-
scales were good, and Cronbach's alpha ranged from
0.78 to 0.83. The reliability of the Depression scale
was slightly better for the English than for the French
version.
Table 3
Factor solutions for 3, 4 and 5 factors of the CAPE using the 1st
random half of the sample in the exploratory factor analysis, and
goodness of fit indices using the 2nd random half of the sample in the
confirmatory analysis
Factor Item
3-Factor solution:
1 — Depression Q1. Sad
Q2. Double meaning
Q9. Pessimism
Q12. No future
Q14. Not worth living
Q19. Frequently cry
Q21. No energy
Q22. Odd looks
Q32. Blunted emotions
Q38. Guilty
Q39. Failure
Q40. Feeling tense
2 — Positive
symptoms
Q5. Messages from TVQ24. Thought
withdrawal
Q26. Thought insertion
Q28. Thought
broadcasting
Q30. Thought echo
Q31. External control
Q33. Verbal
hallucinations
Q34. Voices
conversing
Q41. Capgras
Q6. False appearance
Q7. Being persecuted
Q10. Conspiracy
Q11. Being important
Q13. Being special
Q15. Telepathy
Q17. Influenced by
devices
Q20. VoodooQ42. Visual
hallucinations
3 — Negative
symptoms
Q3. Lack of enthusiasm Q29. Lack of
spontaneity
Q35. Lack of
hygiene
Q18. Influenced by
devices
Q36. Unable to
terminate
Q23. Empty mind
Q37. Lack of hobby
Q4. Not talkative
Q8. No emotion
Q16. No interest in
others
Q25. Lack of activity
Q27. Blunted feelings
Fit indices for confirmatory factor analysis (2nd random subsample)
Goodness of Fit Index=0.866
Adjusted Goodness of Fit Index=0.860
Comparative Fit Index=0.734
Root mean square residual=0.065
4-Factor solution:
1 — DepressionQ1. Sad
Q9. Pessimism
Q12. No future
Q14. Not worth living
Q19. Frequently cry
Q5. Messages TV
Q15. Telepathy
Q17. Influence by
devices
Q20. Voodoo
Q21. No energy
Q22. Odd looks
Q38. Guilty
Q39. Failure
Q40. Feeling tense
Q30. Thought echo
Q31. External control
Q33. Verbal
hallucinations
Q34. Voices
conversing
2 — Positive-
bizarre
90K. Brenner et al. / Schizophrenia Research 95 (2007) 86–95

Page 6

3.1.2. Confirmatory factor models
Table 2 presents goodness of fit indices for the whole
sample and for the sample split by language (English or
French). All confirmatory factor models were less than
optimal according to the general rule of thumb for ac-
ceptance of model fit (GFI, AGFI, CFIN0.90 and
RMRb0.05). Given that the adjusted goodness of fit
indices were close to 0.90, however, the fit was within
the acceptable range.
Table 2 also presents the individual factor loadings
obtained from the first confirmatory factor analysis.
There are some minor differences between the item
loadings for the English and the French versions of
the CAPE. For example, the items in the Depression
scale tended to have higher loadings in the English
than in the French version. A similar result was noted
in the Cronbach's alpha coefficients noted above
(Table 1).
3.2. Exploring alternative factor structures
Because the confirmatory factor analysis models
using the original scales did not provide an optimal fit,
the optimal underlying structure of the CAPE was in-
vestigated through exploratory factor analysis. Explor-
atory factor analysis was performed on the second
random subsample (50%), while the follow-up confir-
matory factor analysis was performed using the first
random sample of observations.
Exploratory factor analysis suggested a multiple
factor solution. There were 10 eigenvalues N1. How-
ever, the scree plot suggested a 3–5-factor solution
(Table 3). When the number of factors in the principal
component analysis was fixed at 3, the components
accounted for 31.5% of the total variance (10.3% for
Table 3 (continued)
Factor Item
4-Factor solution (Continued):
2 — Positive-
bizarre
Q24. Thought
withdrawal
Q26. Thought insertion
Q41. Capgras
Q42. Visual
hallucinations
Q28. Thought
broadcasting
3 — Negative
symptoms
Q3. Lack of
enthusiasm
Q4. Not talkative
Q8. No emotion
Q29. Lack of
spontaneity
Q35. Lack of hygiene
Q18. Influenced by
devices
Q36. Unable to
terminate
Q37. Lack of hobby
Q32. Blunted emotions
Q16. No interest in
others
Q23. Empty mind
Q25. Lack of activity
Q27. Blunted feelings
4 — Social
delusions
Q2. Double meaning
Q6. False appearance
Q7. Being persecuted
Q10. Conspiracy
Q11. Being important
Q13. Being special
Fit indices for confirmatory factor analysis (2nd random subsample)
Goodness of Fit Index=0.875
Adjusted Goodness of Fit Index=0.868
Comparative Fit Index=0.755
Root mean square residual=0.063
5-Factor solution:
1 — DepressionQ1. Sad
Q9. Pessimism
Q12. No future
Q14. Not worth living
Q18. Influenced by
devices
Q19. Frequently cry
Q21. No energy
Q22. Odd looks
Q23. Empty mind
Q36. Unable to terminate
Q37. Lack of hobby
Q38. Guilty
Q39. Failure
Q40. Feeling tense
2 — Positive-
bizarre
Q5. Messages from TV
Q17. Influenced by
devices
Q24. Thought
withdrawal
Q26. Thought insertion
Q28. Thought
broadcasting
Q30. Thought echo
Q31. External control
Q33. Verbal
hallucinations
Q34. Voices conversing
Q41. Capgras
Q42. Visual
hallucinations
3 — Negative
symptoms
Q3. Lack of
enthusiasm
Q4. Not talkative
Q8. No emotion
Q25. Lack of
activity
Q27. Blunted feelings
Q29. Lack of
spontaneity
Q35. Lack of
hygiene
Q32. Blunted
emotions
Q16. No interest
in others
Table 3 (continued)
FactorItem
5-Factor solution (Continued):
4 — Social
delusions
Q2. Double meaning
Q6. False appearance
Q7. Being persecuted
Q10. Conspiracy
Q11. Being important
Q13. Being special
5 — Popular
psychic
beliefs
Q15. TelepathyQ20. Voodoo
Fit indices for confirmatory factor analysis (2nd random subsample)
Goodness of Fit Index=0.876
Adjusted Goodness of Fit Index=0.869
Comparative Fit Index=0.760
Root mean square residual=0.063
91 K. Brenner et al. / Schizophrenia Research 95 (2007) 86–95

Page 7

depression, 11.4% for positive symptoms and 9.7% for
negative symptoms). When 4 factors were analyzed,
the components accounted for 35.2% of the total
variance (9.8% for depression, 9.3% for positive-
bizarre symptoms, 9.6% for negative symptoms, and
6.5% for social delusions). Finally, when 5 factors were
analyzed, the components accounted for 38.4% of the
total variance (11.7% for depression, 8.9% for positive-
bizarre symptoms, 7.5% for negative symptoms, 5.8%
for social delusions and 4.6% for popular psychic
beliefs).
3.2.1. Follow-up confirmatory factor analyses
The 3-, 4-, and 5-factor solutions produced with the
first half of the sample were then forced into con-
firmatory factor analyses with the second random half of
the sample. The fit indices (Table 3) are similar to those
obtained for the original CAPE version. Therefore, no
overall improvement was reached by the alternative
factor structures.
3.3. Seeking parsimony
The results of the Mokken scale analyses are pre-
sented in Table 4. Unidimensionality of the positive and
negative symptom scales could not be confirmed since
values of H were below 0.30 which is at the lower end of
the “weak” range. For the depression scale, a Loevinger
coefficient of homogeneity of 0.36 was found, indicating
a medium hierarchical scale.
Because the Loevinger coefficients were below the
thresholds for medium or strong scalability, item re-
duction was performed in the next step. Based on a
new 50% random subsample, a bottom-up procedure
proposed by Hemker et al. (1995) was performed for
each scales to construct shortened scales based on high
values of the H Loevinger coefficients. Results were
cross-validated using the second half of the sample.
As shown in Table 4, the results indicate that scal-
ability might be improved by shortening the origi-
nal scales. Our latent trait approach shows that the
CAPE could be reduced by 19 questions, from 42 to
23 items, with no loss of reliability. The Loevinger
coefficients for the three shortened CAPE scales were
between 0.330 and 0.401, indicating at least weak
scalability.
4. Discussion
This study had 3 objectives: 1) to reconfirm the fit of
the original factor structure, 2) to determine if an alter-
native factor structure is better, and 3) to determine if
there is a more parsimonious subset of items in order to
have a shortened, but equally valid version of the CAPE.
Our study is an important addition to existing CAPE
validation research. Previous validation studies sup-
ported the CAPE's factor structure, but the generali-
zation of their results may be limited because they
included either exclusively young males (Stefanis et al.,
2002), women only (Verdoux et al., 2003), or 19 year-
olds only (Stefanis et al., 2004). There was, however,
one study that included both genders with a larger age
range (Hanssen et al., 2003). None of these validations
studied the English CAPE.
Table 4
Items selected for the three CAPE scales based on Mokken analysis: Loevinger H coefficients for random two split halves and for total sample
Scale 1: Positive symptoms Scale 2: Negative symptomsScale 3: Depression
ItemLoev HLoev H Item Loev HLoev H ItemLoev H Loev H
1st half2nd half 1st half 2nd half 1st half 2nd half
Q5
Q6
Q11
Q13
Q24
Q26
Q28
Q30
Q31
Q33
Q34
Q41
Scale:
0.215
0.350
0.345
0.402
0.318
0.336
0.318
0.327
0.350
0.336
0.386
0.371
0.330
Loev H for total sample:
0.291
0.260
0.320
0.352
0.381
0.354
0.320
0.320
0.313
0.340
0.274
0.325
0.390
0.322
Q8
Q18
Q21
Q27
0.380
0.416
0.346
0.446
0.289
0.324
0.295
0.395
Q1
Q9
Q12
Q14
Q38
Q39
Q40
0.483
0.289
0.351
0.395
0.381
0.380
0.340
0.635
0.337
0.397
0.428
0.422
0.441
0.430
Scale: 0.401
Loev H for total sample:
0.228
0.331 Scale:0.342
Loev H for total sample:
0.360
0.378
92 K. Brenner et al. / Schizophrenia Research 95 (2007) 86–95

Page 8

Our study offers new information resulting from
our relatively large sample which allowed us to cross-
validate the exploratory and confirmatory analyses on
split samples. In addition, our study examined the
English and the French versions of the CAPE in North
America. The fact that our study obtained factor load-
ings for the CAPE similar to those of previous studies in
very diverse populations adds additional credibility to
the three-factor model of the psychotic phenotype in the
general population.
Our results support the internal consistency of both
the English and French versions of the instrument in a
non-clinical sample. As for the factor structure, although
the confirmatory factor models were “non-optimal”, the
GFI, AGFI, and RMR were not far from the optimal
criteria. The CAPE questionnaire is easy to use and
useful for examining 3 to 5 clusters of symptoms found
in the population.
Alternative scaling (4 and 5 factors) did not improve
the model fit, but researchers interested in distinguishing
among types of positive symptoms could find these
additional scales useful. This subdivision of the positive
psychotic-like symptoms is in line with Stefanis et al.
(2004) who emphasised the notion of a multidimen-
sional construct of positive symptoms in schizotypy and
related schizophrenia phenotypes. Peralta and Cuesta
(1998, 1999) found that a better fit of their data was also
achieved when delusions were separated into Schnei-
derian and non-Schneiderian. The ‘positive-bizarre'
symptoms in our 5-factor model are primarily Schnei-
derian, supporting this as a legitimate symptom
dimension even among community controls.
Finally, our results suggest that it is possible to re-
duce the number of items in the three scales and still
obtain reliable estimates of the severity of positive,
negative, and depressive symptoms. This shorter ver-
sion of the CAPE (23 items) could be useful for surveys
where a smaller number of items are desired, although
further validation is warranted.
The recruitment plan aspired to generate a random
sample. Although potential subjects were approached in
different public places, the vast majority were found in
areas of downtown Montreal frequented by individuals
with higher than average education: there are 3 univer-
sities within 2 miles of each other in the downtown
core, as well as many financial institutions. Although
there may be a bias towards higher educational
attainment in the current sample relative to a purely
random sample, this limitation may be offset by the
very high response rate obtained by the face-to-face
approach, the chance to win money, and the speed with
which the questionnaire could be completed.
Finally, there is always a danger of response bias in
any self-report questionnaire, and individuals may
respond in an automatic, unreflective manner. We are
optimistic about the sincerity of our participants' re-
sponses, however, by the fact that we obtained fre-
quencies of psychotic symptoms that are similar to
those obtained through clinical interview, and by the
finding that responses on items were consistent within
scales.
The development of the CAPE provides a new
option for the assessment of positive, negative, and
depressive symptoms in the general population. When
positive symptoms are considered along a continuum,
numerous research questions of fundamental interest
can be tested. For example, one may ask whether those
factors that increase risk for schizophrenia are also
associated with an increased propensity to experience
non-clinical levels of positive symptoms. Physical
markers, such as dermatoglyphic anomalies, that are
suggestive of second-trimester neurodevelopmental
insult, are present in individuals meeting criteria
for diagnoses of either schizophrenia (e.g., Markow
and Wandler, 1986) or schizotopy (e.g., Chok et al.,
2005); if the degree of dermatoglyphic asymmetry
correlated with CAPE scores in a non-clinical sample,
this would support dimensional models linking
putative risk factors to varying degrees of psychotic-
like symptoms. Similar questions could be addressed
for a variety of statistical risk factors for psychosis that
also lie along the continua, such as childhood trauma
(e.g., Shevlin et al., 2007) and cannabis use
(Arseneault et al., 2002; Verdoux et al., 2003). This
dimensional approach to psychosis was supported, for
example, by the results of the Verdoux et al. (2003)
study showing the link between cannabis use and both
positive and negative symptoms in a sample of non-
clinical female students. In a large population sample,
Johns et al. (2004) also found the presence of psy-
chotic symptoms being associated with cannabis
dependence, alcohol dependence, victimisation, recent
stressful life events, lower intellectual ability and
neurotic symptoms, which are risk factors commonly
reported for clinical psychosis. Verdoux and van Os
(2002) suggest that the research literature to date
supports the study of psychosis on a continuum for the
elucidation of etiology in the same way that research
on cardiac outcomes examines both risk factors and
outcome along dimensions.
In conclusion, the current study suggests that the
CAPE has satisfactory psychometric properties and is
ready for use among English- and French-speaking
North American populations.
93K. Brenner et al. / Schizophrenia Research 95 (2007) 86–95

Page 9

Role of funding source
This study was supported by a grant from the Fonds de recherche
en santé du Québec (FRSQ: the Quebec Health Research Fund)
awarded to Suzanne King, Ridha Joober and Antonio Ciampi. The
FRSQ had no further role in the study design, in the collection,
analysis and interpretation of data, in the writing of the report, nor in
the decision to submit the paper for publication.
Contributors
S. King designed the study and wrote the protocol. R. Joober and
A. Ciampi were involved at the conceptualization level of the project.
K. Brenner coordinated the research, managed the literature searches
and wrote the first draft of the manuscript. N. Pawliuk entered the data
and performed demographic statistical analysis; F. Fathalli entered the
data as well. N. Schmitz undertook the final statistical analyses. All
authors contributed to and have approved the final manuscript.
Conflict of interest
All authors declare that they have no conflicts of interest.
Acknowledgements
The authors wish to thank the research assistants who collected
data: Sherezad Abadi, Amélie Barras, Eric Chetrit, Jessica Diamond,
Anna Fukuda, Pascale Le Hir, and Sylvain Lemieux.
References
Andreasen, N.C., 1989. The Scale for the Assessment of Negative
Symptoms (SANS): conceptual and theoretical foundations. Br. J.
Psychiatr., Suppl. 49–58.
Arseneault, L., Cannon, M., Poulton, R., Murray, R., Caspi, A.,
Moffitt, T.E., 2002. Cannabis use in adolescence and risk for adult
psychosis: longitudinal prospective study. Bmj 325, 1212–1213.
Bentall, R.P., Claridge, G.S., Slade, P.D., 1989. The multidimensional
nature of schizotypal traits: a factor analytic study with normal
subjects. Br. J. Clin. Psychol. 28 (Pt 4), 363–375.
Chapman, L.J., Chapman, J.P., 1987. The search for symptoms
predictive of schizophrenia. Schizophr. Bull. 13, 497–503.
Chapman, L.J., Chapman, J.P., Raulin, M.L., 1976. Scales for physical
and social anhedonia. J. Abnorm. Psychology 85, 374–382.
Chapman, L.J., Chapman, J.P., Raulin, M.L., 1978. Body-image
aberration in schizophrenia. J. Abnorm. Psychology 87, 399–407.
Chok, J.T., Kwapil, T.R., Scheuermann, A., 2005. Dermatoglyphic
anomalies in psychometrically identified schizotypic young adults.
Schizophr. Res. 72, 205–214.
Claridge, G., 1972. The schizophrenias as nervous types. Br. J.
Psychiatry 121, 1–17.
Claridge, G., 1987. ‘The schizophrenias as nervous types’ revisited.
Br. J. Psychiatry 151, 735–743.
Donias, S., Karastergiou, A., Manos, N., 1991. Standardization of the
Symptom Checklist-90-R rating scale in a Greek population.
Psychiatriki 2, 42–48.
Embretson, S., Reise, S., 2000. Item Response Theory for Psychol-
ogists. Mahwah, NJ: Erlbaum.
Hanssen, M., Peeters, F., Krabbendam, L., Radstake, S., Verdoux, H.,
vanOs,J.,2003.Howpsychoticareindividualswithnon-psychotic
disorders? Soc. Psychiatry Psychiatr. Epidemiol. 38, 149–154.
Hemker, B.T., Sijtsma, K., Molenaar, I.W., 1995. Selection of
unidimensional scales from a multidimensional item bank in the
polytomous Mokken IRT model. Appl. Psychol. Meas. 19,
337–352.
Huba, G.J., Harlow, L.L., 1987. Robust structural equation models:
implications for developmental psychology. Child Dev. 58 (1),
147–166.
Jackson, M., Claridge, G., 1991. Reliability and validity of a psychotic
traits questionnaire (STQ). Br. J. Clin.Psychol.30 (Pt 4), 311–323.
Johns, L.C., Cannon, M., Singleton, N., Murray, R.M., Farrell, M.,
Brugha, T., et al., 2004. Prevalence and correlates of self-reported
psychotic symptoms in the British population. Br. J. Psychiatry
185, 298–305.
Markow, T.A., Wandler, K., 1986. Fluctuating dermatoglyphic
asymmetryandthegeneticsofliabilitytoschizophrenia.Psychiatry
Res. 19, 323–328.
Mason, O., Claridge, G., Jackson, M., 1995. New scales for the
assessments of schizotypy. Pers. Individ. Differ. 18, 7–13.
Mason, O., Claridge, G., 2006. The Oxford-Liverpool Inventory of
Feelings and Experiences (O-LIFE): further description and
extended norms. Schizophr. Res. 82, 203–211.
Meijer, R.R., Sijtsma, K., Smid, N.G., 1990. Theoretical and
empirical-comparison of the Mokken and the Rasch approach to
IRT. Appl. Psychol. Meas. 14 (3), 283–298.
Peralta, V., Cuesta, M.J., 1998. Factor structure and clinical validity of
competing models of positive symptoms in schizophrenia. Biol.
Psychiatry 44, 107–114.
Peralta, V., Cuesta, M.J., 1999. Dimensional structure of psychotic
symptoms:an item-levelanalysisof SAPSandSANSsymptomsin
psychotic disorders. Schizophr. Res. 38, 13–26.
Peters, E., Day, S., McKenna, J., Orbach, G., 1999a. Delusional
ideationinreligiousandpsychoticpopulations.Br.J.Clin.Psychol.
38 (Pt 1), 83–96.
Peters, E.R., Joseph, S.A., Garety, P.A., 1999b. Measurement of
delusional ideation in the normal population: introducing the
PDI (Peters et al. Delusions Inventory). Schizophr. Bull. 25,
553–576.
Peters, E., Joseph, S., Day, S., Garety, P., 2004. Measuring delusional
ideation: the 21-item Peters et al. Delusions Inventory (PDI).
Schizophr. Bull. 30, 1005–1022.
Poulton, R., Caspi, A., Moffitt, T.E., Cannon, M., Murray, R.,
Harrington, H., 2000. Children's self-reported psychotic symp-
toms and adult schizophreniform disorder: a 15-year longitudinal
study. Arch. Gen. Psychiatry 57, 1053–1058.
Raine, A., 1991. The SPQ: a scale for the assessment of schizotypal
personality based on DSM-III-R criteria. Schizophr. Bull. 17,
555–564.
Rossler, W., Riecher-Rossler, A., Angst, J., Murray, R., Gamma, A.,
Eich, D., et al., 2007. Psychotic experiences in the general
population: a twenty-year prospective community study. Schi-
zophr. Res. 92, 1–14.
Selten, J.P., Gernaat, H.B., Nolen, W.A., Wiersma, D., van den Bosch,
R.J., 1998. Experience of negative symptoms: comparison of
schizophrenic patients to patients with a depressive disorder and to
normal subjects. Am. J. Psychiatry 155, 350–354.
Shevlin,M., Dorahy,M.J.,Adamson,G.,2007.Traumaandpsychosis:
an analysis of the National Comorbidity Survey. Am. J. Psychiatry
164, 166–169.
Stefanis, N.C., Hanssen, M., Smirnis, N.K., Avramopoulos, D.A.,
Evdokimidis, I.K., Stefanis, C.N., et al., 2002. Evidence that three
dimensions of psychosis have a distribution in the general
population. Psychol. Med. 32, 347–358.
94K. Brenner et al. / Schizophrenia Research 95 (2007) 86–95

Page 10

Stefanis, N.C., Delespaul, P., Henquet, C., Bakoula, C., Stefanis, C.N.,
Van Os, J., 2004. Early adolescent cannabis exposure and positive
and negative dimensions of psychosis. Addiction 99, 1333–1341.
van Os, J., Hanssen, M., Bijl, R.V., Ravelli, A., 2000. Strauss (1969)
revisited: a psychosis continuum in the general population?
Schizophr. Res. 45, 11–20.
Verdoux, H., Sorbara, F., Gindre, C., Swendsen, J.D., van Os, J., 2003.
Cannabis use and dimensions of psychosis in a nonclinical
population of female subjects. Schizophr. Res. 59, 77–84.
Verdoux, H., van Os, J., 2002. Psychotic symptoms in non-clinical
populations and the continuum of psychosis. Schizophr. Res. 54,
59–65.
Vollema, M.G., Hoijtink, H., 2000. The multidimensionality of self-
report schizotypy in a psychiatric population: an analysis using
multidimensional Rasch models. Schizophr. Bull. 26, 565–575.
Vollema, M.G., van den Bosch, R.J., 1995. The multidimensionality of
schizotypy. Schizophr. Bull. 21, 19–31.
95 K. Brenner et al. / Schizophrenia Research 95 (2007) 86–95