ARTICLE IN PRESS

Behaviour Research and Therapy 43 (2005) 1527–1542

www.elsevier.com/locate/brat

Psychometric validation of the obsessive belief questionnaire

and interpretation of intrusions inventory—Part 2: Factor
analyses and testing of a brief version
Obsessive Compulsive Cognitions Working Group,1
School of Social Work, Boston University, 264 Bay State Road, Boston, MA 02215, USA
Accepted 7 July 2004

Abstract

The Obsessive Belief Questionnaire (OBQ) and the Interpretation of Intrusions Inventory (III) were
designed to assess beliefs and appraisals considered critical to the pathogenesis of obsessions. In previous
reports we have described the construction and psychometric properties of these measures. In this study a
battery of questionnaires assessing anxiety, depression, and obsessive compulsive symptoms was completed
by 410 outpatients diagnosed with obsessive compulsive disorder, 105 non-obsessional anxious patients, 87
non-clinical adults from the community, and 291 undergraduate students. Items from 6 theoretically
derived subscales of the OBQ were submitted to factor analysis. Three factors emerged reflecting (1)
Responsibility and threat estimation, (2) Perfectionism and intolerance for uncertainty, and (3) Importance
and control of thoughts. A 44-item version (OBQ-44) composed of high-loading items from the 3 factors
showed good internal consistency and criterion-related validity in clinical and non-clinical samples.

Corresponding author: Gail Steketee. Tel.: +1 6173530815; fax: +1 6173535612.

E-mail address: steketee@bu.edu.
1
Co-Chairs: Gail Steketee and Randy Frost. The members of the working group who contributed to this study are (in
alphabetical order): Sunil Bhar, Martine Bouvard, John Calamari, Cheryl Carmin, David A. Clark, Jean Cottraux,
Paul Emmelkamp, Elizabeth Forrester, Mark Freeston, Randy Frost, Celia Hordern, Amy Janeck, Michael Kyrios,
Dean McKay, Fugen Neziroglu, Caterina Novara, Gilbert Pinard, C. Alec Pollard, Christine Purdon, Josee Rheaume,
John Riskind, Paul Salkovskis, Ezio Sanavio, Roz Shafran, Claudio Sica, Gregoris Simos, Ingrid Sochting, Debbie
Sookman, Gail Steketee, Steven Taylor, Dana Thordarson, Patricia van Oppen, Ricks Warren, Maureen Whittal,
Sabine Wilhelm, and Jose Yaryura-Tobias. Special thanks to Michael Kyrios, Steve Taylor, and Dana Thordarson for
conducting data analyses, to Cheryl Carmin for drafting the manuscript, and to Neil Rector for contributing additional
data to the project.

0005-7967/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.brat.2004.07.010
 ARTICLE IN PRESS

1528 Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542

Subscales showed less overlap than original scales. Factor analysis of the III yielded a single factor,
suggesting the total score be used in lieu of the 3 rationally derived subscales. The scales performed well on
tests of convergent validity. Discriminant validity was promising; hierarchical regression analyses indicated
that the OBQ subscales and III generally predicted OC symptoms after controlling for general distress. A
revision of the OBQ, the OBQ-44, is included in the appendix.
r 2005 Elsevier Ltd. All rights reserved.

Keywords: Beliefs; Appraisals; Cognitive assessment; OCD; Responsibility; Perfectionism

Introduction

Recent theoretical and empirical work suggests that several cognitive constructs (e.g., inflated
responsibility, overestimation of threat, thought-action fusion, intolerance of uncertainty) are
relevant to the development and exacerbation of obsessive-compulsive (OC) symptoms (Freeston,
Rheaume, & Ladouceur, 1996; Frost & Steketee, 2002; Purdon, 2001). Cognitive strategies for
treating obsessive-compulsive disorder (OCD) have been based on the modification of key beliefs
and appraisals formulated in these theories (Rachman, 1998; Salkovskis & Warwick, 1985; van
Oppen & Emmelkamp, 2000; Whittal & McLean, 1999). Clearly, it is important to measure
cognitive constructs hypothesized to be relevant to OCD. Many such measures have been
developed in recent years, but with little consensus regarding which domains to assess and very
limited psychometric information regarding these scales (Taylor, Kyrios, Thordarson, Steketee, &
Frost, 2002).
To address these concerns a large group of international researchers, the Obsessive Compulsive
Cognitions Working Group (OCCWG, 1997, 2001), identified domains considered central to
OCD and developed the 87-item Obsessive Beliefs Questionnaire (OBQ) and the 31-item
Interpretation of Intrusions Inventory (III). Six rationally derived subscales of the OBQ-87
(overestimation of threat, intolerance of uncertainty, importance of thoughts, control of thoughts,
responsibility, and perfectionism) and the three subscales of the III-31 (control of thoughts,
importance of thoughts, responsibility) showed good internal consistency but were highly
intercorrelated. Findings from initial testing provided preliminary evidence for the reliability and
validity of these instruments, but further study of reliability and specificity of the measures using
larger samples was needed.
Part 1 of this investigation (OCCWG, 2003) described the reliability and convergent
and discriminant validity of the original, rationally derived subscales of the OBQ and III
using four samples who completed the English language version of the instruments.
These included 248 participants with OCD as their primary (most severe) disorder (OC), 105
anxious controls (AC), 87 community controls (CC) and 291 student controls (SC).
Results indicated that the belief and appraisal dimensions represented by the OBQ and III
subscales showed good internal consistency and good test–retest reliability over a 2–3
month interval. Findings for criterion-related (known-groups) validity based upon comparisons
of four subject samples were generally good. The OCs scored significantly higher than both non-
clinical samples (SC and CC) on every subscale, and scored higher than ACs on 3 (Responsibility,
Control, and Importance of thoughts) of the 6 OBQ subscales and 2 (Responsibility, Control of
 ARTICLE IN PRESS

Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542 1529

thoughts) of the 3 III subscales. Thus, beliefs related to Tolerance for Uncertainty, Over-
estimation of Threat, and Perfectionism and appraisals related to Importance of Thoughts
appeared to be relevant but not specific to OCD.
As evident from their moderately high intercorrelations, the subscales of the OBQ and subscales
of the III overlapped, suggesting that the appraisal and belief domains measured by these scales
may not be well differentiated from one another. Convergent validity was evident in the moderate
correlations of almost all OBQ subscales, except Importance of Thoughts with measures of OC
symptoms, but the III showed a more variable pattern of association. With respect to discriminant
validity, correlations of the cognitive measures with measures of non-OC symptoms (e.g.,
depression and general anxiety) was approximately as high as those with OC symptoms. This lack
of specificity was apparent in both OC and non-OC samples. To further clarify discriminant
validity, a series of analyses were conducted to partial out the influence of other measures. The
OBQ and III showed a specific relationship with OC symptoms even after controlling for worry,
but both instruments appeared to have relevance for other negative affective states, such as worry,
and may not be unique to OCD.
The aim of the present paper is to report on a factor analysis of the OBQ and III using a large
sample of OCD participants to determine whether a reduced item pool would be more efficient in
capturing the beliefs underlying OCD. This study also explores whether a smaller number of
dimensions are relevant for assessing the beliefs, interpretations and other cognitive constructs
associated with OC symptoms. The psychometric properties of empirically based, rather than
theoretically derived, subscales are examined using the multi-site data set described in the Part 1
paper that includes non-OCD clinical and non-clinical samples.

Method

The methodology employed for this paper is summarized below. A complete description is
available elsewhere (OCCWG, 2003).

Participants

Subjects were recruited from OCCWG member sites located in Australia, Canada, France,
Greece, the Netherlands, Italy, and the United States. Methods regarding translation are
described elsewhere (OCCWG, 2001). Because preliminary analyses using Box’s M test indicated
that the variance–covariance matrices for the OBQ and III sufficiently differed across samples
with different languages, it was inappropriate to pool these data. The present paper represents the
results from the English-language version of the two measures and constitutes 56% of the
collected data.
The English-language scales were completed by four participant groups: (1) those with a DSM-
IV diagnosis of OCD as their most severe problem OC (n ¼ 410), (2) AC (n ¼ 105), (3) CC
(n ¼ 87), and (4) SC (n ¼ 291). The OCD sample included 248 participants from Part 1 of this
study and an additional sample of 162 participants from OCCWG and other sites who were
included in factor analyses. Like participants in Part 1, the additional OCD subjects were
diagnosed using standardized interviews and did not differ significantly from the original subjects
 ARTICLE IN PRESS

1530 Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542

on demographic characteristics (mean age ¼ 36.5 years, 42% male, 52% single; all Fs and
w2 o1.0).
ACs had one or more of the following diagnoses and did not meet criteria for OCD: panic
disorder (72%), agoraphobia (58%), posttraumatic stress disorder (25%), generalized anxiety
disorder (17%), specific phobia (16%), social phobia (13%), and hypochondriasis (2%).
(Percentages do not add up to 100 because patients could have more than one disorder.) OCs
and ACs were excluded if they had current psychotic or substance use disorders. DSM-IV
diagnoses (American Psychiatric Association, 2000) for clinical participants were established via
the Structured Clinical Interview for DSM-IV (First, Spitzer, Gibbon, & Williams, 1996) for 44%,
the Anxiety Disorders Interview Schedule for DSM-IV (Brown, DiNardo, & Barlow, 1994) for
47%, and unstructured interview by an experienced clinician for 9%. CCs included friends or
family of OCCWG members as well as teachers and community service organization members.
SCs were primarily first year university or college students most of whom received course credit
for participating in the study.
Demographic information regarding subject groups is available in Part 1 of this study
(OCCWG, 2003). Clinical participants (OC and AC, mean age of 35 years) and CC (mean ¼ 42
years) were older than student participants (mean age ¼ 21 years). The OC group had more men
(44%) than the other samples (32–33% men) and more unmarried participants (52%) than AC or
CC subjects (34% and 25% single, respectively). The CC group was significantly better educated
than other groups (mean ¼ 17 years), and OC and SC samples (mean of 15 years each) were more
educated than ACs (14 years). The two clinical samples contained more unemployed or disability
payment-assisted participants (14–25%) than community or student samples (0%). No differences
were evident for ethnicity; most participants (88–94%) were Caucasian.

Procedure

The OC and AC groups were recruited primarily from specialty clinics for anxiety disorders;
SCs from college classes, and CCs from workplaces, acquaintances, and community service
organizations. Informed consent was obtained for all participants.

Measures

OBQ consists of 87 belief statements considered characteristic of obsessive thinking (OCCWG,

1997, 2001). Scale items represent six rationally determined subscales thought to represent the key
belief domains of OCD. The subscales are (1) control of thoughts—14 items, (2) importance of
thoughts—14 items, (3) responsibility—16 items, (4) intolerance of uncertainty—13 items, (5)
overestimation of threat—14 items, and (6) perfectionism—16 items. Respondents indicate their
general level of agreement with items on a 7-point rating scale that ranges from (3) ‘‘disagree
very much’’ to (0) ‘‘neutral’’ to (+3) ‘‘agree very much’’. Item responses were transformed to a
1–7 scale, and subscale scores were calculated by summing across their respective items.
Interpretation of Intrusions Inventory (III) is a 31-item scale that assesses immediate appraisals
or interpretations of unwanted, distressing intrusive thoughts, images or impulses (OCCWG,
1997, 2001). Instructions provide a definition of unwanted intrusions and illustrative examples.
Participants write down two intrusive thoughts, images or impulses that they experienced recently
 ARTICLE IN PRESS

Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542 1531

and then rate the recency, frequency, and distress associated with these intrusions. Respondents
then rate 31 statements as they pertain to intrusive thoughts like those recorded on the
questionnaire using a scale from 0 (‘‘I did not believe this idea at all’’) to 100 (‘‘I was completely
convinced this idea was true’’). Three subscales were scored: (1) importance of thoughts—10
items, (2) control of thoughts—11 items, and (3) responsibility—10 items. For data analysis, the
100-point scale was transformed to a 10-point scale with scores ranging from 0 to 310. Subscale
scores were calculated by summing across their respective items.
For the purpose of the analyses in the present study, participants at most sites completed
additional scales including the Padua Inventory (Washington State University revision, PI-R;
Burns, Keortge, Formea, & Sternberger, 1996), Beck Anxiety Inventory (BAI, Beck & Steer,
1993a), and Beck Depression Inventory (BDI, Beck & Steer, 1993b). Additional measures of OC
symptoms and worry completed in a subset of sites (see OCCWG, 2001, 2003) were not included
in the present study because they were not administered to a sufficiently large number of research
sites or participants.

Results

Factor analyses

Given the elevated inter-correlations among the rationally derived subscales of the OBQ and
III, we sought to determine whether more distinct empirically derived dimensions might underlie
the OBQ and III. A priori, it was unclear how the scales might be combined into a parsimonious
set of underlying dimensions. The high subscale intercorrelations indicated that it was unlikely
that each subscale would correspond to a distinct factor. In the absence of alternative, previously
defined factor models, we conducted exploratory rather than confirmatory factor analyses. An
exploratory factor analysis was conducted separately on the OBQ and III items. Principal axis
factor analysis with oblique rotation was conducted on OBQ items2 for OCD participants only
(n ¼ 410 after listwise deletion). The Kaiser–Meyer–Olkin index of sampling adequacy was .95,
indicating that the correlation matrix was suitable for factor analysis (Tabachnick & Fidell, 1996).
To determine the optimal number of factors to retain in the analysis, the scree plot, the
interpretability of the factor loadings, and the internal stability of each factor according to
Kaiser’s coefficient alpha of generalizability (Kline & Barrett, 1983) were utilized. These criteria
suggested a 3-factor solution that accounted for 42% of the variance.3 Eigenvalues for the first 5
factors were 26.4, 5.4, 4.3, 3.4 and 2.6. The correlations among factors were as follows: Factors 1
and 2 (.52), Factors 1 and 3 (.49), and Factors 2 and 3 (.42).
To assess the robustness (replicability) of the factor structure and content, we also conducted
factor analyses of two randomly split samples of n ¼ 205 OCs each. This method yielded virtually
identical 3-factor solutions with almost complete overlap in high loading items and similar
intercorrelations among factors. Factor analyses conducted with the student sample (the only
2
Because the wording of item 71 was nearly identical to that for item 23, by the flip of a coin, item 71 was omitted for
factor analyses which were conducted on 86 items.
3
Findings available from the corresponding author.
 ARTICLE IN PRESS

1532 Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542

sample large enough to permit a separate analysis) also produced a 3-factor solution, and again,
items loading high on factors in the student sample overlapped closely with those identified for the
OC sample. This was especially evident for Factors 2 and 3; Factor 1 included a few perfectionism
and certainty items that loaded on Factor 2 in the OCD sample. Overall, the factor structure
appeared to be consistent across samples.
Items with factor loadings of .50 or higher (corresponding to ‘‘large’’ loadings; Cohen, 1988)
were retained to generate subscales. This relatively high criterion was chosen because nearly all
OBQ items loaded above .35 on a factor so that including additional items defeated our aim to
reduce the length of the instrument. In addition, including items with lower loadings (e.g., .40–.49)
did not alter the basic content of the scales.
The first scale was labelled Responsibility/Threat Estimation and included seven threat items,
eight responsibility items and one uncertainty item from the original theoretically derived scales.
The 16 items on this factor dealt with preventing harm from happening to oneself or others, the
consequences of inaction, and responsibility for bad things happening. Examples of high loading
items are: ‘‘Harmful events will happen unless I am very careful’’, ‘‘Avoiding serious problems
(for example, illness or accidents) requires constant effort on my part’’, and ‘‘For me, not
preventing harm is as bad as causing harm’’.
The second scale was labelled Perfectionism/Certainty and included 12 perfectionism and four
uncertainty items from the original theoretical scales. These 16 items reflected high, absolute
standards of completion, rigidity, concern over mistakes and feelings of uncertainty. High loading
items are ‘‘For me, things are not right if they are not perfect’’, and ‘‘If I can’t do something
perfectly, I shouldn’t do it at all’’, and ‘‘I must be certain of my decisions’’.
The third scale labelled Importance/Control of Thoughts included nine importance of thoughts
and three control of thoughts items from the original theoretically derived scales. These 12 items
concerned the consequences of having intrusive and/or distressing thoughts or images,
thought–action fusion, and the need to rid oneself of intrusive thoughts. Sample items included
‘‘Having a bad thought is morally no different than doing a bad deed’’, ‘‘Having bad thoughts
means I am weird or abnormal’’ and ‘‘Having intrusive thoughts means I’m out of control’’.
Factor scores were moderately inter-correlated (r’s ranged from .42–.52).
A Principal Axis Factor Analysis (with oblique rotation) based on the total OC
sample (N ¼ 410) was performed on all 31 items of the III. The Kaiser–Meyer–Olkin index of
sampling adequacy was again high (.96). All items loaded 4.47 on a single factor which
accounted for 44% of the variance. Eigenvalues for the first five factors were 11.5, 2.5, 1.8, 1.2 and
1.1. Because of the consistently high loadings, we decided to retain all 31 items rather than shorten
the scale. As was done with the OBQ, additional analyses on the randomly split sample of OC
participants and of SC also produced a single factor with all items loading 4.45 in both OC
samples, and 4.49 in the student sample except for 2 items with factor loadings of .36 and .39. In
view of the consistency of these findings, retention of all 31 items as a single scale was deemed
appropriate.
Table 1 presents the zero-order correlations among the three OBQ factor analytically derived
subscales and the III for the OC and the combined non-OCD samples. The three OBQ subscales
were moderately correlated in OCD samples (r’s ¼ :422:57), and had higher correlations in the
non-OCD sample (r’s ¼ :642:72). The OBQ subscales correlated moderately highly with the III
total score (r’s ¼ :352:63) in OCD and non-OCD samples.
 ARTICLE IN PRESS

Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542 1533

Table 1
Correlations among the empirically derived OBQ (44-items) subscales and the III (31-items)

Empirically derived OBQ scales OBQ-44 Total RT PC ICT III-Total

OBQ-44 total — .88 .83 .76 .59

Responsibility/threat estimation (RT) .91 — .57 .55 .55
Perfectionism/certainty (PC) .91 .73 — .42 .35
Importance/control of thoughts (ICT) .83 .67 .63 — .60
III total score .66 .63 .54 .60 —

OBQ ¼ Obsessive Beliefs Questionnaire, RT ¼ Responsibility/Threat Estimation, PC ¼ Perfectionism/Certainty,

ICT ¼ Importance/Control of Thoughts, III ¼ Interpretation of Intrusions Inventory. Correlations above the diagonal
are for OC participants (n ¼ 244 after listwise deletion). Correlations below the diagonal are for the combined anxious,
community and student control samples (n ¼ 469 after listwise deletion). All correlations are significant at po:001.

Internal consistency

The internal consistency coefficients (Cronbach a) for all three OBQ subscales were high and
comparable to those reported for the rationally derived subscales (OCCWG, 2003). Figures for
the OCD sample were .93 for Responsibility/Threat Estimation and Perfectionism/Certainty; .89
for Importance/Control of Thoughts; .95 for OBQ Total score. Internal consistency for the III
total score was also high (.94). Coefficients a for these scales were comparable for men and women
and for the OCD (n ¼ 248) and non-OCD (n ¼ 483) samples. According to 2 (gender)  2(OCD/
non-OCD) ANOVAs, no effect of gender was observed (all p’s4.30) nor interaction of gender by
OCD status (p’s4.13) for OBQ total and subscale scores and for the III. In summary, the results
support the internal consistence form of reliability of the OBQ and III scales.

Group comparisons

Table 2 presents the means, standard deviations, F values and the results of post hoc
comparisons for the four samples on the three OBQ subscales. A one-way MANOVA performed
on the three OBQ subscales was highly significant; Pillai F ð3332Þ ¼ 588:91, po.001). One-way
ANOVAs conducted for each subscale revealed significant group differences on all three OBQ
scales. Post hoc comparisons (Table 2) indicated that the OC group scored significantly higher
than the non-OCD anxious patients on OBQ Responsibility/Threat Estimation and on
Importance/Control of Thoughts, but not on the Perfectionism/Certainty subscale. On all
comparisons, anxious patients scored significantly higher than students who scored higher than
CC. Overall, group differences on the empirically derived scales were very similar to results
obtained with the rationally determined subscales in which Perfectionism and Tolerance for
Uncertainty did not significantly distinguish OC samples from AC. Findings from a one-way
ANOVA comparing groups on the III total score were significant, F ð3638Þ ¼ 97:28, po.001. Post
hoc comparisons indicated that the OC group scored significantly higher than all other groups. To
summarize, the results generally support the criterion-related (known groups) validity of the OBQ
and III subscales.
 1534
Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542
Table 2
Criterion-related (known-groups) validity: Comparisons of group means on the OBQ-44 total score and subscales

ARTICLE IN PRESS
Obsessive-compulsive Anxious Student Community F df Significant SNK
disorder (OC) n ¼ 244 controls (AC) controls (SC) controls (CC) post-hoc comparisons
n ¼ 103 n ¼ 284 n ¼ 86 (po .05)

M SD M SD M SD M SD

OBQ-44 total 174.3 50.2 159.3 53.0 131.3 44.3 96.0 35.1 74.72*** 3713 OC4AC4SC4CC
Responsibility/threat 64.5 22.4 59.8 22.8 48.4 18.7 34.2 13.0 60.36*** 3713 OC4AC4SC4CC
Perfectionism/certain 69.9 22.1 65.7 21.7 55.5 20.1 41.4 18.1 48.26*** 3713 OC, AC4SC4CC
Import./control of Thgt 39.8 16.3 33.9 15.8 27.1 11.6 20.5 9.3 59.13*** 3713 OC4AC4SC4CC
III-31 total score 1599.3 670.0 1365.1 789.8 720.6 569.5 514.6 454.3 100.79*** 3628 OC4AC4SC4CC

OBQ ¼ Obsessive Beliefs Questionnaire, ***po.001.

 ARTICLE IN PRESS

Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542 1535

Correlations with measures of OC symptoms

Convergent validity was tested by computing correlations between the OBQ and III with the
measures of OC symptoms for the OC sample (n ¼ 186 after listwise deletion). Table 3 shows the
correlations, most of which were significant. Correlations were based on the OC sample because
this was the largest of the samples completing the measures in the table. Inclusion of the other,
smaller samples induces a confound between sample size and statistical significance; a small
correlation based on a large sample may be statistically significant because of high statistical
power, whereas a large correlation based on a smaller sample may not be significant because of
lower power. The results in Table 3 support the convergent validity of the empirically derived
OBQ and III.

Regression analyses

Hierarchical regression analyses, again using the OC sample, provided a stringent test of the
cognitive measures’ capacity to predict specific types of OC symptoms after controlling for general
distress (as indexed by the BDI and BAI). Accordingly, these analyses were used as tests of
discriminant validity. PI-R subscales served as the dependent variables because these represent
several different manifestations of OC symptoms (harming, checking, contamination, grooming).
The BDI and BAI were simultaneously entered in step 1 as a block, followed in step 2 by the 3
OBQ subscales simultaneously entered as a block (n ¼ 179) or by the III total score (n ¼ 167).
(The OBQ and III were examined in separate regressions.) Tolerance values were well within
acceptable limits, indicating that multicollinearity was not a problem. Beta weights and
significance levels are presented in Table 4. The results indicate that all but one of the cognitive
measures predicted at least some types of OC symptoms represented on the Padua Inventory.
After controlling for general distress, OBQ Responsibility/Threat Estimation predicted Harming

Table 3
Convergent validity: Correlations with scales from the Revised Padua Inventory (PI-R)

PI-R harm PI-R harm PI-R PI-R PI-R

impulses thoughts grooming checking contamination

OBQ-44 total .27**** .59**** .31**** .37**** .29****

Responsibility/threat .22*** .62**** .19** .27**** .31****
Perfectionism/certainty .19** .39**** .43**** .45**** .27****
Importance/control of .27**** .42**** .10 .14 .10
thoughts
III-31 total score .20** .43**** .09 .19** .11

*po.05, **po.01, ***po.005, ****po.001.

OBQ ¼ Obsessive Beliefs Questionnaire, RT ¼ Responsibility/Threat Estimation, PC ¼ Perfectionism/Certainty,
ICT ¼ Importance/Control of Thoughts, III ¼ Interpretation of Intrusions Inventory, PI-R ¼ padua inventory-
revised, Harm impulses ¼ impulses to harm self or others, Harm thoughts ¼ thoughts of harm to self or others,
Groom ¼ dressing and grooming. Based on OC sample (n ¼ 186 after listwise deletion).
 ARTICLE IN PRESS

1536 Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542

Table 4
Discriminant validity: Beta weights from hierarchical regressions examining the association of cognitive measures with
OC symptoms, controlling for general distress (general anxiety and depression)

Predictor PI-R harm impulses PI-R harm thoughts PI-R grooming PI-R checking PI-R contamination

Regressions for OBQ subscales:

BDI .08 .05 .04 .07 .12
BAI .35*** .20** .10 .11 .04
OBQ–RT .02 .46**** .04 .06 .23*
OBQ–PC .06 .01 .48**** .45**** .12
OBQ–ICT .11 .09 .07 .09 .13
Regressions for III total score:
BDI .05 .14 .15 .11 .22*
BAI .37**** .28**** .12 .12 .07
III .09 .30**** .00 .12 .01

*po.05, **po.01, ***po.005, ****po.001.

OBQ ¼ Obsessive Beliefs Questionnaire, RT ¼ Responsibility/Threat Estimation, PC ¼ Perfectionism/Certainty,
ICT ¼ Importance/Control of Thoughts, III ¼ Interpretation of Intrusions Inventory, PI-R ¼ padua inventory-
revised, Harm impulses ¼ impulses to harm self or others, Harm thoughts ¼ thoughts of harm to self or others,
Groom ¼ dressing and grooming, BDI ¼ Beck Depression Inventory, BAI ¼ Beck Anxiety Inventory. Based on the
OC sample (n ¼ 179 for OBQ analyses and 167 for III analyses, after listwise deletion).

thoughts and Contamination subscales of the PI-R, whereas the Perfectionism/Certainty subscale
predicted the PI-R Grooming and Checking scales. The OBQ Importance/Control of Thoughts
subscale did not predict any PI-R scale, but in contrast, the III total score predicted Harming
thoughts. Comparable regression analyses using the original subscales of the Padua Inventory
(Sanavio, 1988) as the dependent variable provided very similar results; again, the Importance/
Control of Thoughts scale did not predict any Padua scale. In sum, the results of Table 4 provide
some support for the discriminant validity of the OBQ and III.

Discussion

The present study sought to determine whether factor analyses of the 87-item OBQ and 31-item
III would permit reduction of the number of items, particularly for the OBQ which is quite
lengthy for use in clinical settings. We also sought to reduce overlap among factors evident in high
correlations among subscales in our previous study (OCCWG, 2003). Three factors emerged for
the OBQ, and inclusion of high loading items yielded a 44-item total scale with three subscales of
approximately similar length. Subscales were labelled Responsibility/Threat Estimation,
Perfectionism/Certainty, and Importance/Control of Thoughts. Subscales had high internal
consistency. The intercorrelations among subscales in the OCD sample (r’s ¼ :422:57) appeared
to improve somewhat on those from the original theoretically derived subscales (r’s ¼ :502:79).
The combined factors derived from the factor analysis are intuitively understandable. That
responsibility and danger items loaded on the same factor is consistent with the assumption in
 ARTICLE IN PRESS

Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542 1537

cognitive theories about OCD that excessive concerns about responsibility lead to obsessions
about causing or preventing harm (Rachman, 1997; Salkovskis, 1985, 1989). The overlap of
perfectionism and the need for certainty is consistent with the clinical observation that OCD
patients typically try to minimize mistakes or errors in their efforts to dispel doubts provoked by
intrusive thoughts. In our earlier studies of the OBQ in OC samples (OCCWG, 2001, 2003),
Perfectionism and Certainty subscales were highly correlated, although need for certainty was also
closely related to threat estimation as might be expected. It also makes sense that beliefs about
overimportance and control of thoughts loaded on the same factor. Overimportance given to
intrusive thoughts commonly leads OC patients to try to control these thoughts via suppression,
rethinking and a variety of other strategies designed to remove the thoughts (Freeston &
Ladouceur, 1997).
In factor analytic findings for the III a single factor emerged, indicting that its theoretically
derived subscales are not meaningfully differentiated empirically. That is, items written to capture
hypothesized domains of importance and control of thoughts and responsibility loaded together
on a single factor that can be described as negative interpretation of intrusive thoughts. As in Part
1 of this study, the III total score correlated moderately highly with OBQ subscales (r’s ¼ :352:60
for OCs and .54–.63 for non-OCs), suggesting that the distinction between appraisals/
interpretations of thoughts and beliefs may also be difficult for many people to make. Perhaps
these psychological distinctions postulated by theorists in cognitive models of OCD (Freeston et
al., 1996; Rachman, 1997; Salkovskis, 1985, 1989) may not be detectable with self-report
questionnaire methods. Because the rational evaluation of interpretations and of beliefs is a
backbone of cognitive therapy for OCD (and for other disorders), this issue merits further study.
Although the empirically derived subscales of the OBQ do not substantially improve upon
convergent or discriminant validity compared to the theoretically based scales, they are at least
comparable and all means are in the predicted direction for the shorter instrument. All three
subscales displayed good known-groups validity in distinguishing OC patients from non-clinical
controls. OC patients’ scores on OBQ subscales were somewhat higher than those of anxiety
disordered control patients. This difference was significant for all but the Perfectionism/
Uncertainty scale, a finding also reported for the original perfectionism and uncertainty subscales
of the OBQ-87. This finding is not surprising in view of strong perfectionistic tendencies identified
in a number of other disorders (e.g., body dysmorphic disorder or eating disorders; Shafran, 2002;
Wilhelm & Neziroglu, 2002). Findings indicated good convergent validity and promising, but
somewhat weaker discriminant validity. For the latter, regression analyses controlling for general
distress revealed that the OBQ and III generally predicted OC symptom subtypes in patterns that
might be expected clinically. This finding is consistent with theoretical postulations (e.g.,
Rachman, 2002) that certain types of negative interpretations and beliefs associated with intrusive
thoughts foster particular OC symptoms, although considerable additional evidence will be
required for conclusions about causality.
That Responsibility/Threat Estimation was most strongly linked to harming thoughts and to
contamination/washing symptoms accords with theoretical thinking about these obsessive
symptoms (Rachman, 1997, 1998; Salkovskis, 1985) and with our previous findings of a strong
relationship of threat estimation to contamination obsessions and washing compulsions
(OCCWG, 2001). Likewise, the association of Perfectionism/Certainty with grooming/dressing
symptoms that typically include symmetry concerns and with checking aimed at achieving
 ARTICLE IN PRESS

1538 Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542

certainty makes theoretical and clinical sense. Surprisingly, although the III, which identifies
cognitive reactions to intrusive thoughts, predicted harming thoughts, this was not true for the
OBQ beliefs about Importance and Control of Thoughts, as we expected. Reasons for this
discrepancy are not apparent since research findings suggest that over-estimation of the
importance and need to control thoughts often occurs in response to ideas of harming (Purdon &
Clark, 2002; Thordarson & Shafran, 2002). In any case, although the regression analyses indicate
that OBQ-44 subscales cannot be considered to assess beliefs that pertain only to obsessive
thoughts, 2 of the 3 scales predicted obsessive symptoms independent of general distress. This is
impressive because the general distress is likely caused, at least in part, by the OC symptoms
themselves. Thus, these results, especially for the OBQ responsibility and perfectionism scales,
provide fairly strong evidence for a specific link between beliefs and symptoms (Appendix A).
There is probably little advantage in using the 6 OBQ-87 subscales or the 3 III subscales,
particularly given their multicollinearity which will make interpretation of findings unreliable.
Further, the use of a greater number of subscales may give the misleading impression of a
multidimensional construct when in reality several belief domains appear to be more similar than
different. Thus, the 44-item OBQ with its 3 subscales has advantages over the 87-item version.
However, retention of more subscales may be useful when investigators wish to examine the
specific content of one or more of the hypothesized domains contained in the OBQ (e.g.,
importance of thoughts), comparing it to related constructs (e.g., controlling thoughts). Further,
hypothesized domains may yet prove relevant to specific manifestations of OC symptoms (e.g.,
washing, checking) or to development of cognitive interventions for OCD.
The domains included in OBQ were not intended to include all content areas that are relevant
or specific to OCD. Nor does the present study address whether the belief domains assessed in the
OBQ provide greater explanatory power for OC symptoms compared to more general measures
of beliefs (e.g., Dysfunctional Attitude Scale; Weissman, 1980) intended to assess depressive
cognitions. We encourage further research into other areas of cognitive content (e.g., fear of
positive experiences, sensitivity to discomfort, causal inference) that may elucidate further the
nature of obsessional thinking and ritualistic acts.
Appendix A. Obsessional beliefs questionnaire (OBQ-44)

Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542
This inventory lists different attitudes or beliefs that people sometimes hold. Read each statement carefully and decide
how much you agree or disagree with it.
For each of the statements, choose the number matching the answer that best describes how you think. Because people
are different, there are no right or wrong answers.
To decide whether a given statement is typical of your way of looking at things, simply keep in mind what you are like
most of the time.
Use the following scale:

1 2 3 4 5 6 7

Disagree Disagree moderately Agree a Neither agree Agree a Agree Agree very much

ARTICLE IN PRESS
very much little nor disagree little moderately
In making your ratings, try to avoid using the middle point of the scale (4), but rather indicate whether you usually disagree or agree
with the statements about your own beliefs and attitudes.
RT 1. I often think things around me are unsafe. 1 2 3 4 5 6 7
PC 2. If I’m not absolutely sure of something, I’m bound to make a mistake. 1 2 3 4 5 6 7
PC 3. Things should be perfect according to my own standards. 1 2 3 4 5 6 7
PC 4. In order to be a worthwhile person, I must be perfect at everything I do. 1 2 3 4 5 6 7
RT 5. When I see any opportunity to do so, I must act to prevent bad things from
happening.
RT 6. Even if harm is very unlikely, I should try to prevent it at any cost. 1 2 3 4 5 6 7
ICT 7. For me, having bad urges is as bad as actually carrying them out. 1 2 3 4 5 6 7
RT 8. If I don’t act when I foresee danger, then I am to blame for any consequences 1 2 3 4 5 6 7
PC 9. If I can’t do something perfectly, I shouldn’t do it at all. 1 2 3 4 5 6 7
PC 10. I must work to my full potential at all times. 1 2 3 4 5 6 7
PC 11. It is essential for me to consider all possible outcomes of a situation. 1 2 3 4 5 6 7
PC 12. Even minor mistakes mean a job is not complete. 1 2 3 4 5 6 7
ICT 13. If I have aggressive thoughts or impulses about my loved ones, this means I may 1 2 3 4 5 6 7
secretly want to hurt them.
PC 14. I must be certain of my decisions. 1 2 3 4 5 6 7
RT 15. In all kinds of daily situations, failing to prevent harm is just as bad as deliberately 1 2 3 4 5 6 7
causing harm

1539
 1540
RT 16. Avoiding serious problems (for example, illness or accidents) requires constant 1 2 3 4 5 6 7
effort on my part.
RT 17. For me, not preventing harm is as bad as causing harm. 1 2 3 4 5 6 7

Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542
PC 18. I should be upset if I make a mistake. 1 2 3 4 5 6 7
RT 19. I should make sure others are protected from any negative consequences of my 1 2 3 4 5 6 7
decisions or actions
PC 20. For me, things are not right if they are not perfect. 1 2 3 4 5 6 7
ICT 21. Having nasty thoughts means I am a terrible person. 1 2 3 4 5 6 7
RT 22. If I do not take extra precautions, I am more likely than others to have or cause a 1 2 3 4 5 6 7
serious disaster.
RT 23. In order to feel safe, I have to be as prepared as possible for anything that could go 1 2 3 4 5 6 7
wrong.
ICT 24. I should not have bizarre or disgusting thoughts. 1 2 3 4 5 6 7
PC 25. For me, making a mistake is as bad as failing completely. 1 2 3 4 5 6 7

ARTICLE IN PRESS
PC 26. It is essential for everything to be clear cut, even in minor matters. 1 2 3 4 5 6 7
ICT 27. Having a blasphemous thought is as sinful as committing a sacrilegious act. 1 2 3 4 5 6 7
ICT 28. I should be able to rid my mind of unwanted thoughts. 1 2 3 4 5 6 7
RT 29. I am more likely than other people to accidentally cause harm to myself or to others. 1 2 3 4 5 6 7
ICT 30. Having bad thoughts means I am weird or abnormal. 1 2 3 4 5 6 7
PC 31. I must be the best at things that are important to me. 1 2 3 4 5 6 7
ICT 32. Having an unwanted sexual thought or image means I really want to do it. 1 2 3 4 5 6 7
RT 33. If my actions could have even a small effect on a potential misfortune, I am 1 2 3 4 5 6 7
responsible for the outcome.
RT 34. Even when I am careful, I often think that bad things will happen. 1 2 3 4 5 6 7
ICT 35. Having intrusive thoughts means I’m out of control. 1 2 3 4 5 6 7
RT 36. Harmful events will happen unless I am very careful. 1 2 3 4 5 6 7
PC 37. I must keep working at something until it’s done exactly right. 1 2 3 4 5 6 7
ICT 38. Having violent thoughts means I will lose control and become violent. 1 2 3 4 5 6 7
RT 39. To me, failing to prevent a disaster is as bad as causing it. 1 2 3 4 5 6 7
PC 40. If I don’t do a job perfectly, people won’t respect me. 1 2 3 4 5 6 7
RT 41. Even ordinary experiences in my life are full of risk. 1 2 3 4 5 6 7
ICT 42. Having a bad thought is morally no different than doing a bad deed. 1 2 3 4 5 6 7
PC 43. No matter what I do, it won’t be good enough. 1 2 3 4 5 6 7
ICT 44. If I don’t control my thoughts, I’ll be punished. 1 2 3 4 5 6 7
RT ¼ Responsibility/Threat estimation.
PC ¼ Perfectionism/Certainty.
ICT ¼ Importance/Control of Thoughts.
 ARTICLE IN PRESS

Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542 1541

References

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., text revision).
Washington, DC: Author.
Beck, A. T., & Steer, R. A. (1993a). Manual for the Beck Anxiety Inventory. San Antonio, TX: Psychological
Corporation.
Beck, A. T., & Steer, R. A. (1993b). Manual for the Beck Depression Inventory. San Antonio, TX: Psychological
Corporation.
Brown, T. A., DiNardo, P. A., & Barlow, D. H. (1994). Anxiety Disorders Interview Schedule for DSM-IV (ADIS-IV).
San Antonio, TX: Psychological Corporation.
Burns, G. L., Keortge, S. G., Formea, G. M., & Sternberger, L. G. (1996). Revision of the Padua Inventory for
obsessive compulsive disorder symptoms: Distinctions between worry, obsessions, and compulsions. Behaviour
Research and Therapy, 34, 163–173.
Cohen, J. (1988). Statistical Power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.
First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B. W. (1996). Structured Clinical Interview for DSM-IV—
Patient Edition (SCID-I/P). New York: Biometrics Research Department. New York State Psychiatric Institute.
Freeston, M. H., & Ladouceur, R. (1997). What do patients do with their obsessional thougths? Behaviour Research and
Therapy, 35, 335–348.
Freeston, M. H., Rheaume, J., & Ladouceur, R. (1996). Correcting faulty appraisals of obsessional thoughts. Behaviour
Research and Therapy, 34, 433–446.
Frost, R. O., & Steketee, G. (2002). Cognitive approaches to obsessions and compulsions: Theory, assessment and
treatment. Oxford: Elsevier.
Kline, P., & Barrett, P. (1983). The factors in personality questionnaires among normal subjects. Advances in Behaviour
Research and Therapy, 5, 141–202.
Obsessive Compulsive Cognitions Working Group. (1997). Cognitive assessment of obsessive-compulsive disorder.
Behaviour Research and Therapy, 35, 667–681.
Obsessive Compulsive Cognitions Working Group. (2001). Development and initial validation of the obsessive beliefs
questionnaire and the interpretation of intrusions inventory. Behaviour Research and Therapy, 39, 987–1006.
Obsessive Compulsive Cognitions Working Group. (2003). Psychometric validation of the obsessive belief
questionnaire and the interpretation of intrusion inventory: Part 1. Behaviour Research and Therapy, 41, 863–878.
Purdon, C. (2001). Appraisal of obsessional thought recurrences: Impact on anxiety and moodstate. Behavior Therapy,
32, 47–64.
Purdon, C., & Clark, D. A. (2002). The need to control thoughts. In R. O. Frost, & G. Steketee (Eds.), Cognitive
approaches to obsessions and compulsions: Theory, research and treatment (pp. 29–44). Oxford: Elsevier.
Rachman, S. (1997). A cognitive theory of obsessions. Behaviour Research and Therapy, 35, 793–802.
Rachman, S. (1998). Cognitive theory of obsessions: Elaborations. Behaviour Research and Therapy, 36, 385–402.
Rachman, S. (2002). A cognitive theory of compulsive checking. Behaviour Research and Therapy, 40, 625–639.
Salkovskis, P. M. (1985). Obsessional-compulsive problems: A cognitive-behavioural analysis. Behaviour Research and
Therapy, 23, 571–583.
Salkovskis, P. M. (1989). Cognitive behavioural factors and the persistence of intrusive thoughts in obsessional
problems. Behaviour Research and Therapy, 27, 677–682.
Salkovskis, P. M., & Warwick, H. (1985). Cognitive therapy of obsessive-compulsive disorder: Treating treatment
failures. Behavoural Psychotherapy, 13, 243–255.
Sanavio, E. (1988). Obsessions and compulsions: The Padua Inventory. Behaviour Research and Therapy, 26, 169–177.
Shafran, R. (2002). Eating disorders and obsessive compulsive disorder. In R. O. Frost, & G. Steketee (Eds.), Cognitive
approaches to obsessions and compulsions: Theory, assessment and treatment (pp. 215–231). Oxford: Elsevier.
Tabachnick, B. G., & Fidell, L. S. (1996). Using multivariate statistics (3rd ed.). New York: HarperCollins.
Taylor, S., Kyrios, M., Thordarson, D. S., Steketee, G., & Frost, R. O. (2002). Development and validation of
instruments for measuring intrusions and beliefs in obsessive-compulsive disorder. In R. O. Frost, & G. Steketee
(Eds.), Cognitive approaches to obsessions and compulsions: Theory, research and treatment (pp. 117–137). Oxford:
Elsevier.
 ARTICLE IN PRESS

1542 Obsessive Compulsive Cognitions Working Group / Behaviour Research and Therapy 43 (2005) 1527–1542

Thordarson, D. S., & Shafran, R. (2002). Importance of thoughts. In R. O. Frost, & G. Steketee (Eds.), Cognitive
approaches to obsessions and compulsions: Theory, research and treatment (pp. 15–28). Oxford: Elsevier.
van Oppen, P., & Emmelkamp, P. M. G. (2000). Issues in cognitive treatment of obsessive compulsive disorder. In
W. K. Goodman, M. V. Rudorfer, & J. D. Maser (Eds.), Obsessive compulsive disorder: Contemporary issues in
treatment (pp. 117–132). Mahwah, NJ: Lawrence Erlbaum.
Weissman, A. N. (1980). Assessing depresogenic attitudes: A validation study. Paper presented at the annual meeting of
the Eastern Psychological Association, Hartford, CT.
Whittal, M. L., & McLean, P. D. (1999). Cognitive behavior therapy for OCD: The rationale, protocol, and challenges.
Cognitive and Behavioral Practice, 6, 383–396.
Wilhelm, S., & Neziroglu, F. (2002). Cognitive theory of body dysmorphic disorder. In R. O. Frost, & G. Steketee
(Eds.), Cognitive approaches to obsessions and compulsions: Theory, assessment and treatment (pp. 203–214). Oxford:
Elsevier.