Responsibility, metacognition and unrealistic pessimism in obsessive–compulsive disorder

Responsibility, metacognition and unrealistic pessimism

in obsessive–compulsive disorder

Helen Niemeyer a,n,1, Steffen Moritz b,1, Reinhard Pietrowsky a

a Heinrich-Heine-University, Institute for Experimental Psychology, Department of Clinical Psychology, Universitatsstraße 1, 40225 Dusseldorf, Germanyb University Medical Center Hamburg-Eppendorf, Department of Psychiatry and Psychotherapy, Martinistraße 52, 20246 Hamburg, Germany

a r t i c l e i n f o

Article history:

Received 8 June 2012

Received in revised form

15 December 2012

Accepted 2 January 2013

Available online 11 January 2013

Keywords:

Obsessive–compulsive disorder

Responsibility

Metacognition

Overestimation of threat

Unrealistic optimism

a b s t r a c t

Cognitive models stress the importance of cognitive belief domains (CBD) for the pathogenesis of

obsessive–compulsive disorder (OCD). However, the relative contribution of responsibility and

metacognition – core aspects of CBD – to OC symptoms is not fully understood yet. Furthermore,

two subcomponents of overestimation of threat (OET), overestimation of the personal probability

(unrealistic pessimism) and overestimation of the general risk of negative events, require clarification.

First, we investigated the relative contribution of responsibility and metacognition to OC symptoms.

Second, we hypothesized that OCD patients overestimate the personal risk and display unrealistic

pessimism. Thirty-four OCD patients and 34 healthy controls completed the Obsessive Beliefs Ques-

tionnaire (OBQ) and the Unrealistic Optimism Questionnaire (UO). Responsibility significantly predicted

obsessive symptoms after controlling for metacognition. In contrast to previous findings, responsibility

is not fully explained by metacognition. Finally, our results confirm unrealistic pessimism in OCD, even

after controlling for depression.

& 2013 Elsevier Inc. All rights reserved.

1. Introduction

Cognitive theories of obsessive–compulsive disorder (OCD)

stress the importance of cognitive and metacognitive belief

domains (Rachman, 1997; Salkovskis, 1985, 1989; Wells, 2009).

A large body of literature have demonstrated the importance of an

inflated sense of responsibility (Salkovskis et al., 2000; Shafran,

Thordarson, & Rachman, 1996; Smari & Holmsteinsson, 2001;

Steketee, Frost, & Cohen, 1998) as well as the relevance of

metacognition (Amir, Cashman, & Foa, 1997; Emmelkamp &

Aardema, 1999; Solem, Myers, Fisher, Vogel, & Wells, 2010;

Wells & Papageorgiou, 1998). In recent studies metacognition

significantly predicted inflated responsibility (Gwilliam, Wells, &

Cartwright-Hatton, 2004; Myers & Wells, 2005) but not vice

versa. It is currently unresolved whether responsibility is merely

a by-product of metacognition with no additional contribution to

OC symptoms (Wells, 1997, 2009), or has predictive value in its

own right.

Overestimation of threat (OET) is another prominent cognitive

bias discussed as being relevant for OCD. It is closely related to

responsibility: both load on the same factor in the Obsessive

Beliefs Questionnaire (OBQ-44; Obsessive–Compulsive Cognitions

Working Group (OCCWG, 2005). OET is a complex construct

comprising a number of independent components, which require

systematic study of their individual contributions to obsessive

symptoms. Recently, our group (Moritz & Jelinek, 2009; Moritz &

Pohl, 2006, 2009) applied the unrealistic optimism paradigm

(Weinstein, 1982) to OCD to shed light on the relevance of two

subcomponents for OCD, namely the overestimation of the

personal and the objective incidence probability of negative

events. This paradigm is well suited to help to clarify whether

OCD patients are especially prone to a biased perception of their

personal incidence probability. It can be used to investigate

whether subjects display a bias towards enhanced subjective

vulnerability, whether they have indeed experienced negative

events in the past more often, or whether they merely or

additionally overestimate the severity of harm (Moritz & Jelinek,

2009; Moritz & Pohl, 2009). Unrealistic optimism arises from the

overestimation of the subjective likelihood for positive events to

happen to oneself, while at the same time the personal risk for

negative events is underestimated. The reverse response pattern,

labeled as unrealistic pessimism, might be an important compo-

nent of OET. The UO paradigm goes beyond present question-

naires for OET which do not differentiate between these

important components. Our group (Moritz & Jelinek, 2009;

Moritz & Pohl, 2006, 2009) found that patients with OCD over-

estimated their personal risk for negative events compared to

healthy controls, but not the objective probability of threat. Since

results are not fully consistent across studies, further clarification

Contents lists available at SciVerse ScienceDirect

journal homepage: www.elsevier.com/locate/jocrd

Journal of Obsessive-Compulsive and Related Disorders

2211-3649/$ - see front matter & 2013 Elsevier Inc. All rights reserved.

http://dx.doi.org/10.1016/j.jocrd.2013.01.001

n Corresponding author. Tel.: þ49 211 81 12272; fax: þ49 211 81 14261.

E-mail address: [email protected] (H. Niemeyer).1 HN and SM have equally contributed and split first authorship.

Journal of Obsessive-Compulsive and Related Disorders 2 (2013) 119–129

Author's Personal Copy

www.elsevier.com/locate/jocrd

www.elsevier.com/locate/jocrd




mailto:[email protected]


of unrealistic pessimism in OCD is necessary. Furthermore, its

relationship with responsibility and metacognition awaits

examination.

1.1. Inflated responsibility and metacognition

Responsibility is considered to be a core cognitive belief domain

for the development and maintenance of OCD (Salkovskis &

Forrester, 2002). It has been defined by Salkovskis, Forrester, and

Richards (1998, p. 285) as ‘‘the belief that one has power which is

pivotal to bring about or prevent subjectively crucial negative

outcomes. These outcomes are perceived as essential to prevent.

They may be actual, that is having consequences in the real world,

and/or at a more moral level’’. Cognitive models, predicated on

Beck’s (1976) cognitive theory, posit that obsessions and compul-

sions arise from specific dysfunctional beliefs. Individuals with OCD

interpret intrusions, which are common phenomena (Rachman &

De Silva, 1978; Salkovskis & Harrison, 1984), misleadingly as

indicating an inflated responsibility for preventing the harm the

intrusive thought implied. This responsibility appraisal leads to

intense fear and tension and urges the individual to conduct

neutralizing behavior with the aim of averting the negative con-

sequences. A reduction of discomfort after neutralizing and the

ascribed importance lead to the maintenance of compulsions and

avoidance, thus consolidating the dysfunctional belief of an inflated

responsibility (Salkovskis, 1991, 1996; Salkovskis & Forrester,

2002).

Beliefs about thoughts and thinking processes are referred to

as metacognitive beliefs (Purdon & Clark, 2002). Metacognition is

defined as any knowledge or cognitive process that is involved in

the appraisal, monitoring or control of cognition (Flavell, 1979). A

closely related concept that can also be defined as a subset of the

general concept of metacognition is thought-action fusion (TAF;

Rachman, 1997; Shafran, Thordarson, & Rachman, 1996). It

denotes that thoughts can cause events or are equivalent to the

actions they concern. In patients with OCD, intrusions activate

certain dysfunctional metacognitions, for example ‘‘If I think

about something, bad it will happen’’ (thought-action fusion).

Subjects with OCD interpret intrusions as overly important and

dangerous or threatening. Wells explicit metacognitive theory of

OCD (Wells & Matthews, 1994) was built on this earlier meta-

cognitive perspective by Rachman and Shafran and is part of his

generic metacognitive Self-Regulatory Executive Function Model

(S-REF; Wells, 2009; Wells & Matthews, 1994). According to Wells

(2009), metacognition can be subdivided into three domains:

beliefs about the uncontrollability of thoughts, beliefs about their

meaning and importance, and beliefs about the need to perform

rituals. This metacognitive appraisal activates metacognitive

beliefs about rituals, and as a consequence compulsions and

neutralizing behavior are implemented as attempts to control

one�s thoughts. The negative appraisal of the intrusions and the

dysfunctional control strategies invoked in turn enhance their re-

occurrence. Thus, the appraisal of one�s own thoughts determines

the focus of attention as well as the personal reaction to the

thoughts (Wells, 2009).

The exact relationship between responsibility and metacogni-

tion is controversial and unclear. In Wells�S-REF-model metacog-

nition is explicitly elaborated and distinguished from

responsibility. According to this model, metacognition causes

inflated responsibility (Wells, 2009; Wells & Matthews, 1994).

However, in Salkovskis� cognitive model of OCD, metacognitions

are not explicitly mentioned, but are rather defined as subtypes of

responsibility (Salkovskis, 1989, 1996; Salkovskis & Forrester,

2002; Salkovskis, Shafran, Rachman, & Freeston, 1999). It is

important to note that responsibility within the context of the

appraisal model shows much resemblance to the construct of

metacognition, as the interpretation of intrusive cognitions is a

central part of the appraisal model of OCD. Due to this concep-

tualization the Responsibility Interpretations Questionnaire (RIQ;

Salkovskis et al., 2000), for example, assesses interpretations of

inflated responsibility regarding intrusive thoughts through items

about thought control (e.g., ‘‘I must regain control of my

thoughts’’) and items about responsibility.

Whether inflated responsibility is a contributing or causal factor

in dysfunctional metacognitions, or whether it is itself a conse-

quence or by-product of metacognition is an unresolved question.

Experimental designs used the manipulation of metacognitive

beliefs (thought-action fusion; Rassin, Merckelbach, Muris, &

Spaan, 1999) and cognitive beliefs about responsibility and danger

(Arntz, Voncken, & Goosen, 2007), all of which evoked intrusions

and neutralizing behavior. However, Coles, Pietrefesa, Schofield,

and Cook (2008) found only a modest prediction of metacognition

for distress caused by obsessive symptoms. Myers, Fisher, and

Wells (2009a, 2009b) could show in two studies that only TAF

predicted OC symptoms significantly, but not the responsibility/

overestimation of threat and the importance of thoughts/control of

thoughts subscales (Myers et al., 2009b) of the Obsessive Beliefs

Questionnaire (OBQ-44; OCCWG, 2003). Gwilliam et al. (2004),

Myers and Wells (2005) and Myers, Fisher, and Wells (2008) found

that metacognition significantly predicted OCD when responsibil-

ity was controlled, whereas the association between responsibility

and obsessive symptoms was not significant after controlling for

metacognition. However, some of the questionnaires aimed to

measure responsibility like the Responsibility Appraisal Question-

naire (RAQ2; Rachman, Thordarson, Shafran, & Woody, 1995) or

the Responsibility Attitude Scale (RAS; Salkovskis et al., 2000)

include items measuring TAF and thus might overlap with

metacognition.

1.2. Unrealistic pessimism as component of OET

A related cognitive belief domain to responsibility and meta-

cognition is OET. Responsibility and OET both loaded on the same

factor in a factor analytic study (OCCWG, 2005). Furthermore,

Rachman (2002) argued that high perceived responsibility in

combination with high perceived probability of harm and per-

ceived seriousness of harm might determine the intensity and

duration of compulsive checking. Moreover, he states that ele-

vated responsibility leads to increased perception of danger, and

that intense checking will enhance responsibility in turn. Also,

metacognitive dimensions are likely to be implicated in threat

monitoring, as they may be responsible for predisposing indivi-

duals to engage in monitoring for negative stimuli (Sica, Steketee,

Ghisi, Chiri, & Franceschini, 2007). Responsibility concerns and

metacognitions might be particularly related to personal vulner-

ability and overestimation of the severity of the consequences of

events. Both might increase subjective vulnerability, and vice

versa. However, no studies about possible associations to respon-

sibility and metacognition have been conducted yet, and further

research is necessary to shed light on this.

OET is a multidimensional construct, according to Salkovskis�

cognitive model for OCD (Salkovskis & Wahl, 2003; Sookman &

Pinard, 2002). It consists of several components, including over-

estimation of the objective risk, a perceived enhanced personal

vulnerability, and dysfunctional coping strategies. Most of the

questionnaires tapping OET do not contain subscales for its

different components. As our group (Moritz & Jelinek, 2009)

pointed out, the overestimation of threat subscale of the OBQ-44

(OCCWG, 2003) covers a variety of different components: general

overestimation of threat, overestimation due to prior experiences and

overestimation of the personal vulnerability. A more detailed

H. Niemeyer et al. / Journal of Obsessive-Compulsive and Related Disorders 2 (2013) 119–129120


investigation of the relevance of each of these subfacettes is

required in order to elucidate the core distortion involved in OCD.

Moritz and Pohl (2006) began to analyze the aforementioned

components separately and found that OCD patients did not differ

from healthy controls in their estimation of the incidence fre-

quency of both OCD-related and OCD-unrelated negative events in

general. In two recent studies the paradigm of unrealistic optimism

was applied to investigate whether subjects with OCD show a

biased perception of their personal incidence probability of the

occurrence of negative and positive events (Moritz & Jelinek, 2009;

Moritz & Pohl, 2009). As mentioned, unrealistic optimism means

that people overestimate the probability of positive events hap-

pening to them, whereas they think that negative events happen

more likely to other people. This bias is almost universal and has

been found in both healthy and high risk samples for severe bodily

diseases (Weinstein, 1982; Weinstein & Klein, 1996; Weinstein,

Marcus, & Moser, 2005). As opposed to unrealistic optimism,

unrealistic pessimism refers to overestimating the probability of

negative events happening to oneself, while assuming that positive

events happen more likely to other people. Moritz and Jelinek

(2009) and Moritz and Pohl (2009) administered the Unrealistic

Optimism Questionnaire (UO), which is described in detail in the

methods section, to OCD patients and healthy controls.

Unrealistic pessimism concerning negative events is connected

to vulnerable self-themes and sensitive self-domains in OCD, as

the results of Moritz and Pohl (2009) and Moritz and Jelinek

(2009) suggest. Moritz and Jelinek (2009) found that in a

comparison of self versus others the healthy sample displayed

an unrealistic optimism bias, whereas the OCD sample, as

expected, showed the opposite bias, unrealistic pessimism, espe-

cially for negative events that were related to washing and

checking. The bias is investigated with paradigms tapping implicit

and explicit comparisons. The explicit comparison entails that the

person directly compares oneself to others regarding probability

estimates for positive and negative events, whereas the implicit

comparison contrasts two separate estimates, one for oneself and

one for others (first versus second block). The results of the two

blocks are compared post-hoc (Moritz & Jelinek, 2009; Moritz &

Pohl, 2009). Moritz and Jelinek (2009) found evidence for an

implicit but not an explicit bias in an OCD sample. This pattern of

results is consistent with former studies on smokers which found

that implicit measures are more sensitive to capture unrealistic

optimism (Weinstein et al., 2005). Moritz and Pohl (2009) did not

apply an implicit, but only an explicit comparison, to investigate

this component in detail. They found that OCD patients felt less

safe to experience negative and especially OCD-related events,

but could not confirm an unrealistic pessimism bias, as partici-

pants with OCD rather showed an attenuated unrealistic opti-

mism regarding negative events in an explicit comparison. In

contrast to unrealistic optimism, which means that persons think

that positive events are more likely to happen to themselves than

to others, whereas negative events are less likely, the term

‘‘attenuated unrealistic optimism’’ is used to refer to the pattern

that the usual unrealistic optimism bias is present (and not

reversed) but less strong.

Thus, in their OCD sample Moritz and Pohl (2009) did not find

unrealistic pessimism in the explicit comparison, but rather less

unrealistic optimism, whereas Moritz and Jelinek (2009) found no

bias in the explicit comparison, but an unrealistic pessimism (i.e.,

the unrealistic optimism bias was reversed) in the implicit

comparison. To summarize, the results of the two studies differ

with respect to the explicit comparison, whereas an implicit

comparison was not implemented in the study by Moritz and

Pohl (2009). Thus, it is not yet resolved whether participants with

OCD are more prone to unrealistic pessimism or instead tend to

show attenuated unrealistic optimism.

OC symptoms correlated with the personal risk for washing-

and checking-related events, also pointing to the relevance of the

subjective personal vulnerability in OCD (Moritz & Jelinek, 2009).

In addition, the UO captures the valence of negative and positive

events, and the OCD sample saw graver consequences of negative

events than did the control sample, indicating an overestimation

of harm severity (Moritz & Jelinek, 2009). According to self-report,

OCD patients had not actually experienced the relevant events

more often in the past than the healthy controls, thus making it

unlikely that the pessimism arises from prior experience.

1.3. Present study

Three aims were pursued in the present investigation: First,

we wanted to clarify the relevance of responsibility and meta-

cognition for obsessive symptoms by examining the relative

contribution of each belief domain to OC symptoms using

hierarchical regression. While some previous research obtained

mixed results, the studies by Wells and colleagues showed a

greater influence of metacognition on OC symptoms compared to

responsibility. The interpretation of these findings is complicated

by the fact that there is an overlap in the operationalization of

metacognition and responsibility in some of the studies. For the

present study, responsibility was operationalized with as little

overlap with metacognition as possible, thus leading us to expect

that both metacognition and responsibility would predict OC

symptoms. To our knowledge, it is the first study investigating

the relative contribution of responsibility and metacognition to

OC symptoms in a German sample. We think that our results

can be generalized to other patient samples in industrialized

countries.

The second aim was to investigate the relevance of the

components overestimation of the personal vulnerability and over-

estimation of the objective likelihood of the incidence of events (for

other persons) in both an implicit and an explicit comparison, and

in addition to investigate the overvaluation of the valence of

negative and positive events. We hypothesized that participants

with OCD will be prone to a subjective personal vulnerability, but

not to a general overestimation, thus showing an unrealistic

pessimism. Furthermore, we expected that they appraise the

consequences of negative events as more grave. To determine

whether unrealistic pessimism occurs depending on OCD we

controlled for depression in subsidiary analyses, as depression

has been shown to correlate with OC symptoms (Moritz, Meier,

Hand, Schick, & Jahn, 2004) and to induce a negative view of

events (Clark, 2001). Finally, exploratory correlational analyses

were conducted to examine the interrelationships between the

components of an unrealistic pessimism to inflated responsibility

and metacognition.

2. Methods

2.1. Participants

Sixty-eight participants, all at least 18 years old, took part in the study. The 34

participants in the OCD group were recruited in the psychotherapeutic out-patient

clinic of the Psychology Department at the University Duesseldorf and from

therapists in private practice. This group had (1) a total score in Obsessive–

Compulsive Inventory-Revised (OCI-R; Foa et al., 2002, German Version: Goenner,

Ecker, & Leonhart, 2009) of Z18, constituting the German cut-off, as well as (2) a

self-report Yale-Brown Obsessive–Compulsive Scale (Y-BOCS-SR; Goodman et al.,

1989, German Version: Baer, 1993) total score ofZ10 in case of either mental or

behavioral obsessions, orZ16 if it both occurred; and (3) the OCD diagnosis was

verified with the Structured Clinical Interview for DSM-IV (Wittchen, Zaudig, &

Fydrich, 1997) or confirmed by the clinicians who were treating the patients for OCD.

In the latter case, clinicians applied DSM-IV criteria for diagnosis, but without

applying a structured interview. Exclusion criteria for the OCD sample were current

H. Niemeyer et al. / Journal of Obsessive-Compulsive and Related Disorders 2 (2013) 119–129 121


or previous diagnoses of (1) psychosis or schizophrenia, (2) bipolar disorder,

(3) substance dependence, or (4) neurological disorders.

The healthy participants were recruited via an established subject pool and

word-of-mouth advertisement and consisted of students enrolled at the university

as well as non-student participants recruited from the community, who were

mainly prior study participants. In a questionnaire, they were asked about current

or former psychological disorders and whether they had ever undergone psy-

chotherapeutic or pharmacological treatment. Furthermore, they had to (1) score

below both the OCI-R cut-off, (2) also the Y-BOCS-SR cut-off, (3) score low on

depressive symptoms (score lower than 14 in the BDI-II; Beck, Steer, & Brown,

1996, German Version: Hautzinger, Keller, & Kuehner, 2009), and (4) as described

above, have no current or previous psychiatric diagnoses.

2.2. Measures

OCD symptom severity was assessed using the OCI-R and the Y-BOCS-SR. The

psychometric properties of both the original and the German versions are good

(Goodman et al., 1989; Schaible, Armbrust, & Nutzinger, 2001; Steketee, Frost, &

Bogart, 1996; Goenner et al., 2009). The OCI-R consists of 18 items inquiring the

extent of distress caused by OC symptoms during the last month. The Y-BOCS-SR

comprises 10 items and demonstrated a good convergent validity with the Y-BOCS

clinical interview (Schaible et al., 2001; Steketee et al., 1996).

Participants also completed the Beck Depression Inventory-II to assess the

comorbid depression symptom severity (BDI-II; Beck et al., 1996, German Version:

Hautzinger et al., 2009), which also has good psychometric properties.

2.2.1. Obsessive Beliefs Questionnaire (OBQ)

We administered three self-report inventories to assess the cognitive variables

and considered the particular subscales that covered the belief domains of interest

for our statistical analysis. To assess metacognition beliefs, the German version of

the OBQ (Ertle et al., 2008) was applied, which contains 24 items from the English

version that loaded on the designated factor with a minimum alpha of a ¼ .30. Due

to this reduction of the number of items the subscales of the German OBQ cover

mainly the domains importance of thoughts, perfectionism and OET. All subscales

share good reliability and validity. For the present study the subscale importance of

thoughts was used.

2.2.2. Responsibility and Interpersonal Behaviors and Attitudes Questionnaire

(RIBAQ)

The RIBAQ is an inventory designed to assess pro-social (one subscale) and

anti-social (two subscales) attitudes associated to responsibility, aimed to capture

interpersonal ambivalence. An initial long form of 60 items developed by clinical

experts on OCD was shortened to 32 items following a factor analysis (Moritz

et al., 2009). The questionnaire consists of three subscales (excessive worry and

responsibility, latent aggression, and suspiciousness/distrust); only those items

were included that had loadings greater than .4 and differed from the loadings on

the two other factors by at least .2 (absolute value) in order to ensure

independence of subscales. Only the responsibility subscale was used for our

statistical analyses. This scale of 17 items shows good reliability (Cronbach’s

alpha: a¼ .87) and convergent validity with the respective subscale of the OBQ

(r¼ .64, p o .001; Moritz et al., 2009). Furthermore, the scale correlated with the

Y-BOCS (r¼ .17, p¼ .02) and the MOCI total score (r¼ .29, p o .001; Moritz et al.,

2009).

Participants were asked to rate how well the following behaviors, attitudes

and habits apply to them using a 4-point scale with response categories ranging

from ‘‘does not apply’’ (¼0) to ‘‘applies only slightly’’ (¼1), ‘‘mostly applies’’ (¼2)

and ‘‘applies completely’’ (¼4). Example items of the subscale ‘‘excessive worry

and responsibility’’ employed in the present study are ‘‘To save the life of a

relative, I would donate an organ’’, ‘‘I suffer from a strict conscience concerning my

relatives’’, or ‘‘I fear that people at work do not think that I am 100% reliable’’.

2.2.3. Unrealistic Optimism Questionnaire (UO)

To assess unrealistic pessimism we administered a self-report questionnaire

that allows contrasting an implicit and an explicit comparison (Moritz & Jelinek,

2009). It consists of five blocks in which the same negative and positive events are

presented. In block 1, participants estimated the likelihood that the depicted event

would happen to themselves in the future (personal vulnerability), without direct

comparison with other persons, while block 2 in turn requested an estimate of the

likelihood that an average person of the same sex and age would experience such

an event in the future (general vulnerability). For both blocks a 7-point Likert scale

with response categories from ‘‘very unlikely’’ (1) until ‘‘highly probable’’ (7) with

‘‘medium’’ (4) as midpoint was applied. Contrasting block 1 and 2 allows for an

implicit comparison, resulting in four dependent measures (self-positive, self-

negative, others positive and others negative).

Block 3 required participants to compare the relative probability that the

event would happen to them in the future versus an average person of the same

age and sex, using a 7-point scale ranging from ‘‘much less compared to others’’

(1) to ‘‘much higher compared to others’’ (7), with ‘‘equally high’’ (4) representing

the midpoint. Thus, block 3 asks in direct comparison how likely negative and

positive events might happen to oneself versus to another person (comparative

vulnerability).

Block 4 requested participants to appraise the consequences of each event

(appraisal or valence): Each person appraised the subjective negativity of the

negative and the subjective positivity of the positive events for him or herself.

Endpoints were ‘‘extremely negative’’ (1) and ‘‘extremely positive’’ (7), the

midpoint was ‘‘neutral’’ (4).

In block 5, participants indicated how often the respective events had occurred

to either themselves or to a friend, relative or acquaintance (prior experience) using

the three rating options ‘‘never’’ (1), ‘‘once’’ (2), and ‘‘several times’’ (3).

Except for the different instructions, the events to be rated were the same for

all five blocks. Examples are ‘‘y to become 90 years old‘‘ for positive events,

and’’y to transmit a disease to someone because of inattentiveness or insufficient

hygiene’’ for the OCD-related events or ‘‘y to directly or indirectly cause a fire

(e.g. oven or candle left burning)’’ for negative events. For the present study blocks

1, 2, 3 and 4 were statistically analyzed.

All questionnaires were delivered in paper–pencil format. The healthy parti-

cipants received them in personal contact or by mail. The participants in the OCD

group received the questionnaire from their therapists and returned the anon-

ymized questionnaires to the out-patient department via mail. The therapists in

turn were given the questionnaire in person. Participants were informed about the

study and gave written consent to participate. They were offered the option of

generating an anonymous code, which allowed them to access their test results

anonymously at the end of the study.

2.3. Overview of statistical analyses

T-tests for independent samples were conducted comparing the samples with

respect to the clinical measures. Univariate analyses of variance (ANOVAs) with

group as factor (OCD and control sample) were conducted to test for significant

differences between the dependent variables responsibility, metacognition and

the two investigated subcomponents of OET. Two hierarchical regression analyses

were conducted to investigate how much variance in OC symptoms was explained

by either responsibility or metacognition. In the first regression responsibility was

entered on step 1 and metacognition on step 2, whereas in the second regression

analysis we reversed the sequence. Three-way mixed ANOVAs and post-hoc

univariate ANOVAs were calculated to investigate unrealistic pessimism in the

implicit comparison. To control for the influence of depression on unrealistic

pessimism we applied a median split: We divided the OCD sample into two groups

with BDI larger or smaller than the median. We then conducted subsidiary

analyses with the subsample of low level depression patients (n¼17). This

approach has been adopted before (Moritz, Kloss, Jahn, Schick, & Hand, 2003).

In addition, we conducted a second, more stringent subsidiary analysis with all

patients below the BDI-cut-off of 19 for moderate depression (Hautzinger et al.,

2009; Kuehner, Buerger, Keller, & Hautzinger, 2007). One-sample t-tests were

used to test the explicit comparison of unrealistic pessimism against the neutral

midpoint (4) of the UO scale. Pearson correlations (r) or, if the variables were not

normally distributed, non-parametric Spearman’s rho coefficients (rs) were calcu-

lated to investigate the relationships between all variables. To compensate for an

inflation of the alpha error the p-value was adjusted, when necessary, that is in

calculations with the OCI-R and the Y-BOCS-SR, which both measure OC symp-

toms. All correlations and the regression analyses were calculated using the

complete sample (n¼68), assuming the quasi-dimensionality of non-clinical and

clinical obsessions and compulsions (Rachman, 1997; Salkovskis, 1991). Subclini-

cal symptoms of OCD are assumed to differ only gradually from clinical obses-

sions, but not qualitatively (Gibbs, 1996). Missing data occurred only for the use of

psychotropic drugs, but not in the questionnaires.

3. Results

3.1. Sociodemographic variables

The group of OCD patients did not differ from the healthy

control subjects with respect to age and gender on a minimum

significance level of p 4 .9 (see Table 1). The mean age of patients

was 36.94 years (SD¼12.25, range 18–65), that of the control

group was 37.12 years (SD¼12.53, range 18–63). Both groups

consisted of 20 women and 14 men. The highest level of educa-

tion was also identical (w2 (4)¼ .00, p 4 .9). 97.1% of OCD subjects

were currently in treatment due to OCD, and 61.8% had been in

treatment in the past. The mean age of patients in the subsample

with low depression was 37.5 years (SD¼12.93, range 21–65),

and the group consisted of eight women and nine men. The

subsample of patients below the cut-off for moderate depression



(five men and five women) had a mean age of 42.3 (SD¼12.32,

range 27–65).

3.2. Clinical measures

As expected, OCD subjects scored higher on OCI-R than the

controls, t (66)¼14.57, p o .001. Likewise, the mean Y-BOCS-SR

score was higher in the OCD group, t (66)¼20.03, p o .001. The

mean BDI-score was M¼25.59 in the OCD group, and M¼5.03 in

the control group; t (66)¼9.08, p o .001. Twenty-nine patients

met the criteria of a comorbid major depression episode. In the

OCD group 19 subjects (55.90%) were unmedicated, nine (26.50%)

were medicated, and six (17.60%) did not state their medication

status. These omissions were treated as missing data. All 34

subjects in the control group were unmedicated (100%). Detailed

information about the sociodemographic, clinical and cognitive

variables is given in Table 1.

3.3. Metacognition and responsibility

3.3.1. Group differences

Subjects with OCD exceeded controls significantly on respon-

sibility, F (1, 66)¼49.29, p o .001, and on metacognition, F (1,

66)¼21.62, p o .001 (see Table 1).

3.3.2. Hierarchical regression analyses

Two hierarchical regression analyses with responsibility and

metacognition as predictor variables for obsessive symptoms

were conducted. Because the OCI-R and the Y-BOCS-SR were

highly correlated, while the Y-BOCS-SR scores were not normally

distributed, we restricted these analyses to the OCI-R. Due to the

high intercorrelations between responsibility and metacognition

we tested for multicollinearity, but the test showed an acceptable

tolerance of .69 (see Tabachnik & Fidell, 2007). The normality of

the scores was assessed and verified for both variables. Homo-

scedasticity of the standardized residuals, assessed by examining

the scatterplots of all variables, was also confirmed.

The results of the regression are shown in Table 2. In a first

step, responsibility was entered into the analysis. It explained a

significant amount of variance (42%). On step 2 metacognition

was included, giving an additional 3.9% of explained variance;

DF (1, 65)¼4.67, p o .05; b¼ .24, p o .05; B¼1.84. With both

variables as significant predictors, we obtained an R2 of .46 and an

adjusted R2 (adj.) of .45 for the final model.

In the second analysis, metacognition was entered on step

1 and responsibility on step 2. Metacognition entered on step

1 explained 28% of variance in obsessive symptoms. Additional

19% of the variance was significantly explained by responsibility

on step 2 (DF (1, 65)¼22.58, p o .01; b¼ .52, p o .01; B¼15.07).

In the final equation the R2 was .46 and the adjusted R2 (adj.)¼ .45.

3.4. Unrealistic optimism

3.4.1. Implicit comparison

In order to assess an unrealistic pessimism in the implicit

comparison, a three-way mixed ANOVA was conducted. Locus

(self versus comparable other) and event type (positive versus

negative) were entered as within-subject factors and group as

between-subject factor. The event probability, on a rating scale

from 1 (‘‘very unlikely’’) to 7 (‘‘highly probable’’), was defined as

dependent variable. The main effect of locus was significant, F

(1,66)¼23.47, p o .001, and so was the main effect of event type F

(1,66)¼33.55, p o .001, as was the interaction between group

and event, F (1,66)¼15.26, p o .001. The main effects of locus and

event type arise from significant differences in pair wise

comparisons between all levels, except for one non-significant

difference between the levels self-positive and others negative.

The interaction between group and locus was not significant,

F (1,66)¼ .87, p 4 .05, but the three-way interaction was,

F (1,66)¼8.71, p o .01. The three-way interaction effect indicates

that the incidence probability of positive and negative events for

oneself versus others differed significantly between the two

groups. Univariate ANOVAs, conducted post-hoc with group as

factor and locus/event type as dependent variable, confirmed that

the estimate of the OCD sample of the likelihood of positive

events happening to them was lower than that of the healthy

control sample, F (1,66)¼15.45, p o .001, while they thought that

their personal likelihood to experience negative events was

higher than did the controls, F (1,66)¼8.01, p o .01. There was

no difference in the estimate of the probability for other persons,

neither for positive events, F (1,66)¼2.21, p 4 .05, nor for

negative events, F (1,66)¼ .84, p 4 .05. The results of this implicit

comparison are displayed in Table 1.

Furthermore, we controlled for the influence of depression by

conducting two subsidiary analyses. First, we included all OCD

patients with a BDI-score below the median of the patient group

(n¼17, cut-off¼24.5) and repeated the above three-way mixed

Table 1

Sociodemographic, psychopathological und cognitive variables and between-

group comparisons. Means and standard deviations (in parentheses).

Variable OCD (n¼34) Healthy

(n¼34)

Statistics

Sociodemographic variables:

Age 36.94

(12.25)

37.12 (12.53) t(66)¼ .00, p 4 .9

Sex (male/female) 14/20 14/20 w2(1)¼ .00, p 4 .9

Psychopathology:

OCI-R 30.71 (7.93) 7.53 (4.82) t(66)¼14.57, p o .001

Y-BOCS-SR 23.03 (5.95) 1.12 (2.31) t(66)¼20.03, p o .001

BDI-II 25.59

(12.45)

5.03 (4.4) t(66)¼9.08, p o .001

Cognitive variables:

Responsibility 2.54 (.37) 1.94 (.34) F(1,66)¼49.29, p

o .001

Metacognition 3.96 (1.73) 2.27 (1.23) F(1,66)¼21.62, p

o .001


Implicit, self (þ) 2.90 (.97) 3.81 (.94) F(1,66)¼15.45, p

o .001

Implicit, self (�) 2.93 (.78) 2.43 (.67) F(1,66)¼8.01, p o .01

Implicit, others (þ) 3.69 (.84) 3.97 (.74) F(1,66)¼2.21, p 4 .05

Implicit, others (�) 3.20 (.95) 2.99 (.97) F(1,66)¼ .84, p 4 .05

Explicit (þ) 3.24 (.76) 3.95 (.59) F(1,66)¼18.65, p

o .001

Explicit (�) 3.71 (.64) 3.38 (.58) F(1,66)¼4.99, p o .05

Valence (þ) 5.07 (1.82) 5.97 (.79) F(1,66)¼7.04, p o .05

Valence (�) 1.86 (.84) 1.62 (.66) F(1,66)¼1.70, p 4 .05

Note: OCD¼Obsessive–compulsive disorder; OCI- R¼Obsessive–Compulsive

Inventory-Revised; Y-BOCS-SR¼Yale-Brown Obsessive–Compulsive Scale self-

report; BDI-II¼Beck Depression Inventory-II; (þ)¼ likelihood of positive events;

(�)¼ likelihood of negative events.

Table 2

Hierarchical regression.

Variable df1 df2 DF p b p B R2 R2adj.

Regression 1

Step 1 Responsibility 1 66 48.54 o .001 .65 o .001 18.87 .42 .42

Step 2 Metacognition 1 65 4.67 o .05 .24 o .05 1.84 .46 .45

Regression 2

Step 1 Metacognition 1 66 25.11 o .001 .53 o .001 4.09 .28 .27

Step 2 Responsibility 1 65 22.58 o .01 .52 o .01 15.07 .46 .45

Note: R2adj.¼R2 adjusted.



ANOVA. Results remained essentially unchanged. Again, the main

effect of locus was significant, F (1,49)¼15.70, p o .001, as was

the main effect of event type F (1,49)¼28.56, p o .001. Compar-

able to the analysis with the full sample, the interaction between

group and locus, F (1,49)¼ .47, p 4 .05, was non-significant,

whereas the interaction between group and event was significant,

F (1,49)¼6.33, p o .05. The three-way interaction was also

significant, F (1,49)¼6.17, p o .05. Univariate ANOVAs confirmed

that the estimate of the OCD sample of the likelihood of positive

events happening to them was lower than that of the healthy

control sample, F (1,49)¼6.70, p o .05. However, there was no

difference between the groups regarding their personal likelihood

to experience negative events, F (1,49)¼3.90, p 4 .05. As in the

full sample, there was also no difference in the estimate of the

probability for other persons, neither for positive events, F

(1,49)¼ .88, p 4 .05, nor for negative events, F (1,49)¼ .41, p

4 .05. The results of this subsidiary analysis to control for

depression are displayed in Table 3.

Second, we conducted a more stringent subsidiary analysis

considering all OCD patients with a BDI-score below the cut-off

for moderate depression (BDI score of 19, n¼10). The three-way

mixed ANOVA was repeated and yielded much the same results as

before. The main effect of locus was significant, F (1,42)¼7.79, p

o .01, as was the main effect of event type, F (1,42)¼17.79, p

o .001. The interaction between group and locus, F (1,42)¼ .002, p

4 .05, was non-significant, whereas the interaction between

group and event was significant, F (1,42)¼6.04, p o .05. One

result did change, however: the three-way interaction was no

longer significant, F (1,42)¼2.67, p o .05, perhaps owing to the

smaller sample size. For the estimated likelihood of positive

events happening to themselves the univariate ANOVA resulted

in a non-significant difference, F (1,42)¼3.65, p 4 .05, which

deviates from the results in the full sample and in the subsample

below the median for depression. However, we found a difference

between the groups regarding the subjective likelihood to experi-

ence negative events, F (1,42)¼4.44, p o .05. This result was

found for the full sample, as well. Finally, no difference occurred

for the judgments for other persons, neither for positive events, F

(1,42)¼1.85, p 4 .05, nor for negative events, F (1,42)¼ .42, p

4 .05 (see Table 4).

We also investigated the unrealistic biases separately for the

OCD group and the healthy controls. Comparing for each sample

separately the estimation of the personal probability to the

estimation of the probability for others (implicit self versus

implicit others), we found that the OCD sample estimated that

positive events were less likely to happen to themselves than to

another comparable person (p o .001, see Table 5). Thus, patients

are unrealistically pessimistic regarding positive events. The

estimation of the incidence of negative events did not differ

significantly. The healthy sample showed the opposite pattern:

the estimation of positive events did not differ significantly, but

they estimated the incidence of negative events for themselves as

significantly lower (p o .001). To summarize, the OCD sample

demonstrates halfway unrealistic pessimism, while the control

subjects are unrealistically optimistic (see Table 5; Fig. 1).

3.4.2. Explicit comparison

Univariate ANOVAs showed that for the explicit comparison

(block 3) groups differed in the estimated incidence probability of

positive events, F (1,66)¼18.65, p o .001. The OCD sample rated

the incidence probability for positive events as significantly

lower. The incidence probability rating of negative events like-

wise differed, with the OCD sample judging personal probability

as significantly higher, F (1,66)¼4.99, p o .05. To conclude, in

comparison to healthy controls the OCD group displays an

unrealistic pessimism (see Table 1). In the subsidiary analysis

with low level depressed patients, the groups also differed in the

estimated probability of positive events, F (1,49)¼9.76, p o .01,

but not of negative events, F (1,49)¼1.03, p 4 .05 (see Table 3).

However, results changed in the second subsidiary analysis with

OCD patients below the cut-off for moderate depression: there

was a significant difference for negative events, F (1,42)¼5.72, p

o .05, but none for positive events, F (1,42)¼3.26, p 4 .05 (see

Table 4).

To compare the rating of each full sample against the neutral

score of the scale (4), we conducted one-sample t-tests for each

group. The OCD sample differed significantly from the neutral

score for both positive, t (33)¼�5.84, p o .001, and negative

event probabilities, t (33)¼�2.68, p o .05, but the healthy

control sample differed merely for the rating of negative events,

t (33)¼�6.28, p o .001. All groups scored lower than the neutral

center point of the scale, which means that they all under-

estimated the incidence probability for both positive and negative

events (see Table 6).

Participants with OCD and healthy controls rated the personal

valence of negative events (block 4) equally (see Table 1). The

valence of positive events was rated significantly lower by the

OCD sample than the control group, F (1,66)¼7.04, p o .05. In a

subsidiary analysis with all OCD patients below the median of the

depression score of the experimental group, there was no longer a

difference between the two groups in the valence of positive

events, F (1,49)¼ .24, p 4 .05. The rating of the valence of negative

events was equal across samples, F (1,49)¼ .43, p 4 .05 (see

Table 3

Subsidiary analyses for unrealistic optimism with low level depression patients.

Means and standard deviations (in parentheses).

Variable OCD (n¼17) Healthy (n¼34) Statistics


Implicit, self (þ) 3.08 (.98) 3.81 (.94) F(1,49)¼6.70, p o .05

Implicit, self (�) 2.83 (.72) 2.43 (.67) F(1,49)¼3.90, p 4 .05

Implicit, others (þ) 3.76 (.74) 3.97 (.74) F(1,49)¼ .88, p 4 .05


Explicit (þ) 3.35 (.74) 3.95 (.59) F(1,49)¼9.76, p o .01

Explicit (�) 3.56 (.71) 3.38 (.58) F(1,49)¼1.03, p 4 .05

Valence (þ) 5.82 (1.37) 5.97 (.79) F(1,49)¼ .24, p 4 .05

Valence (�) 1.5 (.53) 1.62 (.66) F(1,49)¼ .96, p 4 .05

Note: OCD¼Obsessive–compulsive disorder; OCI-R¼Obsessive–Compulsive




Table 4

Subsidiary analyses for unrealistic optimism with patients below the cut-off for

moderate depression (BDI scoreo19). Means and standard deviations (in

parentheses).

Variable OCD (n¼10) Healthy (n¼34) Statistics


Implicit, self (þ) 3.17 (.93) 3.81 (.94) F(1,42)¼3.65, p 4 .05

Implicit, self (�) 2.98 (.87) 2.43 (.67) F(1,42)¼4.44, p o .05

Implicit, others (þ) 3.63 (.48) 3.97 (.74) F(1,42)¼1.85, p 4 .05


Explicit (þ) 3.57 (.59) 3.95 (.59) F(1,42)¼3.26, p 4 .01

Explicit (�) 3.85 (.43) 3.38 (.58) F(1,42)¼5.72, p o .05

Valence (þ) 6.33 (.57) 5.97 (.79) F(1,42)¼1.81, p 4 .05

Valence (�) 1.41 (.30) 1.62 (.66) F(1,42)¼ .96, p 4 .05

Note: OCD¼Obsessive–compulsive disorder; OCI-R¼Obsessive–Compulsive






Table 3). The same significances were found for the subsample

below the cut-off for moderate depression (see Table 4).

3.5. Correlational analyses

3.5.1. Associations between cognitive variables and obsessive

symptoms

The two questionnaires for symptom assessment, OCI-R and

Y-BOCS-SR were significantly correlated. Moreover, responsibility

and metacognition were both significantly associated with the

OCI-R and the Y-BOCS-SR scores. All intercorrelations are dis-

played in Table 7.

3.5.2. Intercorrelations between the belief domains

Responsibility and metacognition were significantly associated

(r¼ .56, p o .001). For all correlations of the components of the

implicit and explicit comparison with each other and with the

valences of positive and negative events, see Table 7.

Responsibility was positively associated with the subjective

likelihood of negative events for self and others in the implicit

comparison and in the explicit comparison, and negatively with

the personal incidence likelihood of positive events. Metacogni-

tion correlated with the personal incidence probability rating of

negative events in the implicit comparison, and with negative

events in the explicit comparison, but not with any other

components. There was no correlation between the evaluation

of positive and negative events (valence) and either metacogni-

tion or responsibility.

3.5.3. Intercorrelations between depression and unrealistic

pessimism

The BDI score in the OCD sample correlated negatively with

the personal valence of positive (rs¼� .49, p o .01) and positively

with the valence of negative (rs¼ .40, p o .05) events, but not with

any of the components of implicit and explicit comparison for an

unrealistic pessimism. In the control group, the BDI score corre-

lated negatively with the self positive component of the implicit

comparison (r¼� .34, p o .05), and positively with the personal

valence of positive events (rs¼ .48, p o .01).

4. Discussion

Our study first aimed to investigate the relative contribution of

responsibility and metacognition to the severity of OC symptoms,

and second to assess a possible overestimation of personal risk

and thus an unrealistic pessimism in OCD patients. We turn first

to the results pertaining to metacognition and responsibility and

then proceed with the data on unrealistic optimism.

Previous studies (Gwilliam et al., 2004; Myers & Wells, 2005)

found that responsibility did not significantly contribute to OC

symptoms when metacognition was controlled for, whereas

metacognition made a significant contribution after controlling

for responsibility. To clarify the relevance of each of these

constructs for OCD, our study aimed to reinvestigate this con-

troversial issue. To this end, we took care to operationalize

responsibility with as little overlap to metacognition as possible,

and used a scale that operationalizes a dimension of social

responsibility, namely inflated worry/responsibility. As expected,

both metacognition and responsibility where highly correlated

with OCD. Conducting two hierarchical regression analyses,

entering the two constructs in reverse order, we found that

metacognition indeed contributed significantly but modestly to

obsessive–compulsive symptoms after responsibility was con-

trolled for, and conversely responsibility made a significant

contribution after controlling for metacognition. The amount of

variance explained by responsibility on step two was even about

four times that explained by metacognition on step two in the

corresponding regression analysis.

There are some important differences between our study and

previous studies conducted by other groups. First, Myers, Wells

and colleagues used non-clinical samples in all their studies,

whereas our sample consisted of 34 clinical and 34 healthy

participants. Second, we chose to operationalize a different

dimension of responsibility: whereas some of the forerunner

studies used questionnaires such as the RAS (Myers & Wells,

2005; Salkovskis et al., 2000) or the RAQ (Gwilliam et al., 2004;

Rachman et al., 1995), we employed the inflated worry/responsi-

bility scale of the RIBAQ which measures social responsibility.

Responsibility should be defined and operationalized as a

Table 5

Implicit comparison of unrealistic optimism. Means and standard deviations (in

parentheses).

Group and valence Implicit, self Implicit, others F (df¼33)

OCD

(þ) 2.90 (.97) 3.69 (.84) 15.23, p o .001

(�) 2.93 (.78) 3.20 (.95) 2.66, p 4 .05

Healthy controls

(þ) 3.81 (.94) 3.97 (.74) .91, p 4 .05

(�) 2.43 (.67) 2.99 (.97) 15.66, p o .001

Note: OCD¼Obsessive–compulsive disorder; (þ)¼ likelihood of positive events;


Fig. 1. Unrealistic optimism in the implicit comparison. Probability that a positive

or negative event will happen to oneself (block 1) and to a person with the same

age and sex (block 2), respectively. Scores ranged from ‘‘highly unlikely’’ (¼2) over

‘‘unlikely’’ (¼3) to ‘‘fifty–fifty’’ (¼4).

Table 6

Comparison of explicit ratings in OCD and healthy control group against neutral

score of the unrealistic optimism scale. Means and standard deviations (in

parentheses).

Group and valence Explicit Neutral score t (df¼33)

OCD

(þ) 3.24 (.76) 4 �5.84, p o .001

(�) 3.71 (.64) 4 �2.68, p o .05

Healthy controls

(þ) 3.95 (.59) 4 � .48, p 4 .05

(�) 3.38 (.58) 4 �6.28, p o .001

Note: OCD¼Obsessive–compulsive disorder; (þ)¼ likelihood of positive events;




multidimensional construct, according to Rachman et al. (1995),

who distinguishes between responsibility in social contexts,

responsibility for harm/ property damage, positive responsibility

and TAF as separate dimensions. The RAQ is likewise based on

these subscales including thought-action fusion (TAF), but its

total score was used in the study by Gwilliam et al. (2004).

Similarly, one item included in the RAS is ‘‘If I think bad things,

this is as bad as doing bad things’’. There are not many items

regarding TAF in the RAS, but still this overlap might contribute to

the finding that responsibility did not significantly explain var-

iance of OC symptoms after controlling for metacognition. Cer-

tainly the amount of variance that is explained by both

metacognition and an operationalization of responsibility that

includes TAF will be higher than the conjoint variance with a

dimension of responsibility like social responsibility. However, in

three other studies (Myers et al., 2008, 2009a, 2009b) the subscale

responsibility of the OBQ-44 (OCCWG, 2005; modified 4-factor

version of Myers et al., 2008) was used and did not explain

significant variance, either. Therefore we conclude that the sub-

component of inflated worry/responsibility for other persons, as

measured by the RIBAQ, might be especially important in OCD.

Recent studies using this scale of the RIBAQ have also shown that

the scores for the inflated worry/responsibility scale were higher

in OCD patients than in healthy controls (Moritz et al., 2009;

Moritz, Kempke, Luyten, Randjbar, & Jelinek, 2011).

However, as the items for responsibility did not constitute a

factor on their own in the factor analysis of the constitution of

the RIBAQ (Moritz et al., 2009), the operationalisation of respon-

sibility is somewhat mixed with that of worry. This finding is in

line with factor analyses for other questionnaires tapping

responsibility, none of which obtained a pure scale. Most

notably, the OBQ (OCCWG, 2003) mixes items for responsibility

with those for threat according to factor analytic results. Pre-

sumably, the fact that responsibility is not a unidimensional

construct (Rachman et al., 1995) vitiates the extraction of the

core of responsibility using construction measurement tools.

Nevertheless, it is desirable to extract an item set that only taps

responsibility. The linkage of worry and responsibility in the

RIBAQ can be viewed as a problem, as worry has been shown to

correlate with OC symptoms (Myers, Fisher, & Wells, 2008;

Solem et al., 2010). Further research is necessary to define and

investigate different dimensions of responsibility, and to study

their relationships to OC symptoms and metacognition, because

we can conclude from the present results that metacognition

according to Wells (2009) and inflated responsibility are non-

redundant.

Regarding the two subcomponents of OET, overestimation of

the personal risk versus overestimation of the objective risk, we

found an unrealistic pessimism: in the implicit comparison

participants with OCD underestimated merely the personal like-

lihood of experiencing positive events and thus showed a relative

pessimism, whereas the full pessimistic bias occurred in the

explicit comparison. Interestingly, the OCD sample did not over-

estimate the incidence likelihood of negative events, but rather

underestimated the likelihood of positive events for themselves in

the implicit comparison. OCD patients did not overestimate the

subjective or objective threat, but they expected to experience

positive events less often. The focus on the underestimation of

positive events in the implicit comparison awaits replication and

clarification in future studies. In the explicit comparison OCD

patients also overestimated their personal risk to endure negative

events, compared to the healthy controls. Furthermore, the

personal valence of negative events did not differ between OCD

patients and the healthy controls, in contrast to the results of our

earlier study (Moritz & Jelinek, 2009). This implies that an overly

severe appraisal of the consequences of threatening events as a

component of OET in OCD was not a major factor in this sample.

However, they judged positive events as less favorable than

healthy participants do.

Differences of our findings compared to the forerunner studies

might stem from sample differences. Whereas the two preceding

studies (Moritz & Jelinek, 2009; Moritz & Pohl, 2009) were

conducted via the internet, this time we recruited a clinical group,

and used a paper- and pencil version of the questionnaires. The

participants in the present study had higher OC symptom scores

than before: the Y-BOCS-SR mean score was 23.03 compared to

17.93 in the sample of Moritz and Pohl (2009). In Moritz and

Jelineks (2009) study the MOCI was used, which cannot be

directly compared to Y-BOCS-SR and OCI-R scores.

In order to test if relationships between unrealistic pessimism

and obsessive–compulsive symptoms are specific, we controlled

for depression by conducting two subsidiary analyses, one con-

sidering only patients with low levels of depression below the

median of the depression score of the OCD sample, and one

considering only patients below the BDI-cut-off of 19 for moder-

ate depression. Depression had no impact on the results of the

three-way ANOVA for the low-level depression group; the level of

significance remained unchanged. However, some results in the

univariate ANOVAs for the implicit and explicit comparisons and

the rating of the valence slightly changed when controlling for

depression: group differences in the ratings of negative events for

oneself (implicit comparison and explicit comparison) and the

Table 7

Intercorrelations of the psychopathological and cognitive variables.

Variable 2 3 4 5 6 7 8 9 10 11 12 13

1 OCI-R .84snnn .76nnn .65nnn .53nnn

� .37nn .40nn� .09 .16 � .40nn .31n

� .23s .19s2 Y-BOCS-SR � .78s

nnn .57snnn .46s

nnn� .34s

nn .26sn

� .18s .10s � .43snnn .35s

nn� .18s .10s

3 BDI-II � .71nnn .57nnn� .45nnn .34nn

� .12 .15 � .50nnn .28n� .20s .15s

4 Responsibility � .56nnn� .26n .57nnn

� .02 .29n� .22 .32nn

� .06s .09s5 Metacognition � � .19 .37nn

� .12 .10 � .16 .28n� .11s .12s


6 Implicit, self (þ) � � .16 .31n� .11 .65nnn

� .09 .33snn

� .30sn

7 Implicit, self (�) � � .09 .48nnn� .17 .39nn

� .21s � .03s8 Implicit, others (þ) � .09 .09 .18 .39s

nn� .25s

n

9 Implicit, others (�) � � .06 .02 � .10s .02s10 Explicit (þ) � � .10 .28s

n� .13s

11 Explicit (�) � .18s � .09s12 Valence (þ) � � .48s

nnn

13 Valence (�) �

Note: Reported values¼Pearson correlation coefficients, if not otherwise marked; s¼Spearman Correlation Coefficient; OCI-R¼Obsessive–Compulsive Inventory-Revised;

Y-BOCS-SR¼Yale-Brown Obsessive–Compulsive Scale self-report; BDI-II¼Beck Depression Inventory-II; (þ)¼ likelihood of positive events; (�)¼ likelihood of negative

events. np o .05; nnp o .01; nnnp o .001.



evaluation of positive events disappeared. Thus, the estimation of

the incidence rates for negative events is affected by depression,

as is the personal valence of events. Nevertheless, even OCD

patients with low levels of depression display elevated levels of

unrealistic pessimism for positive events. In addition, when

analyses were confined to patients below the cut-off for moderate

depression most of the results remained unchanged. However,

the three-way interaction was no longer significant, and in the

post-hoc univariate ANOVA the difference between the groups for

the estimated likelihood of positive events happening to oneself

disappeared. While results in the explicit comparison differed, the

personal valence of events remained the same as in the low-level

depression group. We attribute the changes to the relatively small

sample size in the subgroup (n¼10). To conclude, we found that

controlling for depression somewhat attenuated but did not

abolish unrealistic pessimism in OCD.

Furthermore, in separately calculating correlations between

depression and unrealistic pessimism, we found that both in the

healthy sample and in the OCD sample the two variables were

hardly associated with each other. In the OCD sample, depression

correlated only with the personal valence of positive and negative

events, but not with implicit and explicit comparison for unrea-

listic pessimism. In the healthy sample, depression correlated

only minimally with unrealistic pessimism: only with the self

positive component of implicit comparison. Furthermore, as for

the OCD sample, it was associated with the personal valence of

positive events. We can conclude from these results that depres-

sion and unrealistic pessimism are not closely intertwined.

In summary, similar to previous studies OCD patients showed

rather pessimistic evaluations (Moritz & Jelinek, 2009; Moritz &

Pohl, 2009). Moritz and Jelinek (2009) found an unrealistic

pessimism in the implicit, but not the explicit comparison, and

Moritz and Pohl (2009) found an attenuated unrealistic optimism

in the explicit comparison. The present study is the first in which

an unrealistic pessimism was found in both implicit and explicit

comparisons. The healthy controls in the present study displayed

a relative unrealistic optimism in the implicit comparison only

regarding the negative events, contrary to previous findings

where an optimistic bias was found in both cases (Moritz &

Jelinek, 2009; Weinstein et al., 2005).

It is important to note that in the explicit comparison the

estimated incidence rates were lower than the neutral baseline

(score¼4) in both the OCD as well as the healthy samples. This

indicates that both groups underestimated the incidence rates for

both positive and negative events for themselves compared to

others. It is difficult to explain the tendency that both groups

underestimated incidence probabilities. As not that many studies

on OCD samples and with the UO questionnaire have been

conducted yet, and as this finding has occurred for the first time,

it awaits replication by future studies.

Some components of the UO questionnaire show associations

with OC symptoms. We were able to replicate the relationship

between OC symptoms and personal vulnerability found earlier

(Moritz & Jelinek, 2009), which points to the importance of this

subcomponent of OET for OCD. Unlike in another foregoing study

(Moritz & Pohl, 2009) we found that the Y-BOCS-SR scores were

not inversely, but positively related to the subjective vulnerability

for negative events, in both explicit and implicit comparisons.

However, Moritz and Pohl (2009) found an attenuated optimism,

whereas we found an unrealistic pessimism in the explicit

comparison. Furthermore, we found inverse correlations between

OC symptoms and the personal incidence rating for positive

events in implicit and explicit comparisons, indicating that higher

OCD scores go along with more pronounced pessimistic attitudes.

The relationship between OCD and subjective vulnerability might

differ between subgroups of OCD, and probably certain subtypes

of OCD display differences in showing an attenuated optimism or

an unrealistic pessimism. Further studies differentiating between

the subtypes of OCD could shed light on different causal mechan-

isms of pessimistic biases.

This is the first study to investigate interrelationships between

unrealistic pessimism and responsibility and metacognition.

Responsibility is closely related to the personal vulnerability for

negative events judged in both implicit and explicit comparison.

Presumably, people who feel responsible for negative events will

experience more discomfort (Rassin et al., 1999). Furthermore,

responsibility is negatively correlated with implicit self positive

and positively correlated with implicit others negative. Like

responsibility, metacognition is closely associated with the per-

sonal vulnerability for negative events (implicit comparison).

However, the only other correlation besides this relationship is

with the estimation of negative events in explicit comparison.

None of the other components of unrealistic pessimism correlate

with metacognition. Surprisingly, neither construct was related to

the valence of positive and negative events.

The associations of metacognition with the vulnerability for

negative events are congruent with theoretical assumptions, as

people who believe that thinking about a threatening situation

makes it more likely that the situation actually occurs ought to be

more afraid. Accordingly, OET is regarded a consequence of

metacognition in the framework of Wells’ model (Myers &

Wells, 2005). Of interest, higher responsibility goes along with

less expectation to experience positive events, and is connected to

more pessimistic expectations for other persons to experience

negative events as well. However, more research is necessary

here. In particular, experimental study would be desirable, since

from the present results we cannot conclude whether responsi-

bility/worry or metacognition are causal for unrealistic pessimism

or vice versa.

4.1. Limitations of the present study

The present study has certain limitations. A limitation of the

screening procedure of the control group is that we relied only on

self-report to exclude the presence of psychological disorders.

Participants were not tested in a structured interview or accord-

ing to DSM-IV criteria by a clinician. Moreover, a limitation of our

study is the small sample size, with respect to the analyses we

conducted. Therefore, the results need to be interpreted with

caution. Confirmation of the present findings using larger samples

is desirable. For reasons of statistical power, the correlation and

regression analyses were calculated using both the clinical and

non-clinical sample. Some researchers agree in using non-clinical

samples because OC symptoms are quasi-dimensional in non-

clinical and clinical samples (Gibbs, 1996). Nevertheless, the

occurrence of cognitive and metacognitive biases might differ in

clinical and non-clinical samples, and further studies should use

entirely clinical samples to confirm the present findings regarding

the relations between the three constructs. However, in most of

the previous studies regarding responsibility and metacognition

non-clinical samples were used.

Furthermore, most of the patients who had undergone pre-

vious treatment had received CBT. Therefore, it cannot be fully

excluded that this affected their knowledge and judgment of the

cognitive variables and metacognitions addressed in our study.

However, as all of them still suffered from clinically relevant OC

symptoms and metacognition is not usually integrated into

standard CBT protocols. We thus consider it unlikely that a long

lasting modification of their cognitions had occurred that had a

major impact on our results.

To operationalize responsibility with as little overlap with

metacognition as possible we chose the Responsibility and



Interpersonal Behaviors and Attitudes Questionnaire (RIBAQ,

Moritz et al., 2009) scale inflated worry/responsibility that does

not include items about TAF or thought control, but focuses on

social responsibility. However, as the appraisal of responsibility is

per se conceptually similar to the construct of metacognition, as

we have described above, any operationalization of responsibility

could also be considered an operationalization of metacognition

to some extent. From this point of view the present operationa-

lization of responsibility would still share some elements with

metacognition.

Moreover, we did not control for worry in the hierarchical

regression analyses, unlike the study by Myers et al. (2008), for

example. Additionally, we operationalized responsibility with

regard to its subcomponent inflated worry for other persons.

Worry as a general thinking style and inflated worry regarding

other persons might be highly correlated and future studies

should address it by controlling for worry.

Finally, one flaw of the UO is that its content is rather narrow,

as it captures 15 events only. Not all themes that are relevant for

OCD are covered, as for example the Y-BOCS checklist consists of

72 items. A further development of the instrument is in the

planning stage, including items about hitherto not covered

domains, such as compulsive orderliness and hoarding. In addi-

tion, items concerning the salience of the events asked for would

be advantageous, as well as items concerning further positive

events, in order to have a broader base and to be able to

generalize the results regarding an over- or underestimation of

positive events.

4.2. Conclusions

In conclusion, several implications can be drawn from our

results. First, it could be helpful to alter the underestimation of

positive events and the overestimation of negative events in

cognitive therapy with OCD patients, since obviously they reflect

a cognitive distortion. Not only the pessimistic bias, but also the

lack of the opposite bias of unrealistic optimism, which might

have a protective function in healthy persons, should be given

attention by therapists. The overestimation of subjective threat,

but not of objective risk can be well addressed by cognitive

restructuring. Second, responsibility and metacognition should

both be considered in OCD. Our findings did not completely

support the model of Wells, which proposes that responsibility

is merely a by-product of metacognition that does itself not

contribute to OCD after metacognition is taken into account.

The present findings support the role of both responsibility,

especially its social component, and metacognition in OCD. The

treatment of OCD should include interventions of both cognitive

therapy to address the cognitive biases responsibility and unrea-

listic pessimism as well as metacognitive therapy to address the

distortions regarding importance and control of thoughts.

Acknowledgment

The authors would like to thank Helen-Rose Cleveland, Daniela

Bertram and Annett Schmitz for their helpful technical assistance.

References

Amir, N., Cashman, L., & Foa, E. B. (1997). Strategies of thought control inobsessive–compulsive disorder. Behavior Research and Therapy, 33, 775–777.

Arntz, A., Voncken, M., & Goosen, A. C. A. (2007). Responsibility and obsessive–compulsive disorder: An experimental test. Behavior Research and Therapy, 45,425–435.

Beck, A. T. (1976). Cognitive therapy and the emotional disorders. New York:International Universities Press.

Beck, A. T., Steer, R. A., & Brown, G. (1996). Manual for the Beck Depression

Inventory-II. San Antonio. TX: Psychology Corporation (German Version:Hautzinger, M., Keller, F. & Kuehner, Ch. (2009). Beck depressions-Inventar

Revision (BDI-II). Manual. (2. Auflage). Frankfurt am Main: Pearson Assessment& Information).

Clark, D. A. (2001). The persistent problem of negative cognition in anxiety anddepression: New perspectives and old controversies. Behavior Therapy, 32,3–12.

Coles, M. E., Pietrefesa, A. S., Schofield, C. A., & Cook, L. M. (2008). Predictingchanges in obsessive–compulsive symptoms over a six-month follow-up: Aprospective test of cognitive models of obsessive–compulsive disorder. Cog-nitive Therapy and Research, 32, 657–675.

Emmelkamp, P. M. G., & Aardema, A. (1999). Metacognition, specific obsessive–compulsive beliefs and obsessive–compulsive behavior. Clinical Psychology &

Psychotherapy, 6, 139–145.Ertle, A., Wahl, K., Bohne, A., Moritz, S., Kordon, A., & Schulte, D. (2008).

Dimensionen zwangsspezifischer Einstellungen. Der Obsessive-Beliefs Ques-tionnaire (OBQ) fur den deutschen Sprachraum analysiert [The dimensions ofobsessive beliefs. An analysis of the Obsessive-Beliefs Questionnaire (OBQ) forthe German language area]. Zeitschrift fur Klinische Psychologie und Psychother-

apie, 37, 263–271.Flavell, J. H. (1979). Meta-cognition and meta-cognitive monitoring: A new area of

cognitive developmental inquiry. American Psychologist, 34, 906–911.Foa, E. B., Huppert, J. D., Leiberg, S., Langner, R., Kichic, R., Hajcak, G., et al. (2002).

The Obsessive-Compulsive Inventory Development and validation of a shortversion. Psychological Assessment, 14, 485–496 (German Version: Goenner, S.,Ecker, W., & Leonhart, R. (Eds.) (2009). Obsessive Compulsive InventoryRevised (OCI-R). Deutsche Adaptation. Manual. Frankfurt am Main: PearsonAssessment & Information.).

Gibbs, N. A. (1996). Nonclinical populations in research on obsessive–compulsivedisorder: A critical review. Clinical Psychology Review, 16, 729–773.

Goenner, S., Ecker, W., & Leonhart, R. (Eds.). (2009). Obsessive–Compulsive Inven-

tory- Revised (OCI-R). Deutsche Adaptation. Manual. Frankfurt am Main: PearsonAssessment & Information.

Goodman, W. K., Price, L. H., Rasmussen, S. A., Mazure, C., Fleischman, R. L., & Hill,C. L. (1989). The Yale-Brown Obsessive Compulsive Scale: Development, use,and reliability. Archives of General Psychiatry, 46, 1006–1016 (German Version:Baer, L. (1993). Alles unter Kontrolle: Zwangsgedanken und Zwangshandlungen

uberwinden (2., uberarbeitete und erganzte Auflage). Bern: Huber).Gwilliam, P., Wells, A., & Cartwright-Hatton, S. (2004). Does meta-cognition or

responsibility predict obsessive–compulsive symptoms: A test of the meta-cognitive model. Clinical Psychology & Psychotherapy, 11, 137–144.

Hautzinger, M., Keller, F., & Kuehner, Ch. (2009). Beck Depressions-Inventar Revision

(BDI-II). Manual. (2. Auflage). Frankfurt am Main: Pearson Assessment &Information.

Kuehner, C., Buerger, C., Keller, F., & Hautzinger, M. (2007). Reliabilitaet undValiditaet des revidierten Beck- Depressionsinventars (BDI-II): Befunde ausdeutschsprachigen Stichproben [Reliability and validity of the revised BeckDepression Inventory (BDI-II): Results form German Samples]. Nervenarzt, 78,651–656.

Moritz, S., & Jelinek, L. (2009). Inversion of the ‘‘Unrealistic Optimism’’ biascontributes to overestimation of threat in obsessive–compulsive disorder.Behavioural and Cognitive Psychotherapy, 37, 179–193.

Moritz, S., Kempke, S., Luyten, P., Randjbar, S., & Jelinek, L. (2011). Was Freudpartly right on obsessive–compulsive disorder (OCD)? Investigation of latentaggression in OCD. Psychiatry Research, 187, 180–184.

Moritz, S., Kloss, M., Jahn, H., Schick, M., & Hand, I. (2003). Impact of comorbiddepressive symptoms on nonverbal memory and visuospatial performance inobsessive–compulsive disorder. Cognitive Neuropsychiatry, 8, 261–272.

Moritz, S., Meier, B., Hand, I., Schick, M., & Jahn, H. (2004). Dimensional structureof the Hamilton Depression Rating Scale in patients with obsessive–compulsive disorder. Psychiatry Research, 125, 171–180.

Moritz, S., & Pohl, R. (2006). False beliefs maintenance for fear-related informationin obsessive–compulsive disorder: An investigation. Neuropsychology, 20,737–742.

Moritz, S., & Pohl, R. (2009). Biased processing of threat-related information ratherthan knowledge deficits contributes to overestimation of threat in obsessive–compulsive disorder. Behavior Modification, 33, 763–777.

Moritz, S., Wahl, K., Ertle, A., Jelinek, L., Hauschildt, M., Klinge, R., et al. (2009).Neither saints nor wolves in disguise. Ambivalent interpersonal attitudes andbehaviors in obsessive–compulsive disorder. Behavior Modification, 33,274–292.

Myers, S. G., Fisher, P. L., & Wells, A. (2008). Belief domains of the Obsessive BeliefsQuestionnaire-44 (OBQ-44) and their specific relationship with obsessive–compulsive symptoms. Journal of Anxiety Disorders, 22, 475–484.

Myers, S. G., Fisher, P. L., & Wells, A. (2009a). An empirical test of themetacognitive model of obsessive–compulsive symptoms: Fusion beliefs,beliefs about rituals, and stop signals. Journal of Anxiety Disorders, 19, 806–817.

Myers, S. G., Fisher, P. L., & Wells, A. (2009b). Metacognition and cognition aspredictors of obsessive–compulsive symptoms: A prospective study. Interna-tional Journal of Cognitive Therapy, 2, 132–142.

Myers, S. G., & Wells, A. (2005). Obsessive–compulsive symptoms: The contribu-tion of metacognitions and responsibility. Anxiety Disorders, 19, 806–817.

Obsessive Compulsive Cognitions Working Group (OCCWG) (2003). Psychometricvalidation of the Obsessive Beliefs Questionnaire and the Interpretation ofIntrusions Inventory: Part I. Behavior Research and Therapy, 41, 863–878.



OCCWG (2005). Psychometric validation of the Obsessive Beliefs Questionnaire

and Interpretation of Intrusions Inventory: Part 2: Factor analyses and testing

of a brief version. Behavior Research and Therapy, 43, 1527–1542.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,

assessment, and treatment (pp. 29–43). Boston: Pergamon.Rachman, S. (1997). A cognitive theory of obsessions. Behavior Research and

Therapy, 35, 793–802.Rachman, S. (2002). A cognitive theory of checking. Behavior Research and Therapy,

40, 625–639.Rachman, S. J., & De Silva, P. (1978). Abnormal and normal obsessions. Behavior

Research and Therapy, 16, 233–238.Rachman, S. J., Thordarson, D. S., Shafran, R., & Woody, S. R. (1995). Perceived

responsibility: Structure and significance. Behavior Research and Therapy, 33,

779–784.Rassin, E., Merckelbach, H., Muris, P., & Spaan, V. (1999). Thought-action fusion as

a causal factor in the development of intrusions. Behavior Research and

Therapy, 37, 231–237.Salkovskis, P. M. (1985). Obsessive–compulsive problems: a cognitive-behavioural

analysis. Behavior, Research and Therapy, 25, 571–583.Salkovskis, P. M. (1989). Cognitive behavioural factors and the persistence of

intrusive thoughts in obsessional problems. Behavior Research and Therapy, 27,

677–682.Salkovskis, P. M. (1991). The importance of behavior in the maintenance of anxiety

and panic: A cognitive account. Behavioural Psychotherapy, 79, 6–19.Salkovskis, P. M. (1996). Avoidance behavior is motivated by threat beliefs: A

possible resolution of the cognition-behavior debate. In: P. M. Salkovskis (Ed.),

Trends in Cognitive and Behavioural Therapies (pp. 25–41). Chichester: Wiley.Salkovskis, P. M., & Forrester, E. (2002). Responsibility. In: R. O. Frost, & G. Steketee

(Eds.), Cognitive approaches to obsessions and compulsions – Theory, assessment,

and treatment (pp. 45–61). Boston: Pergamon.Salkovskis, P. M., Forrester, E., & Richards, C. (1998). Cognitive-behavioral approach to

understanding obsessional thinking. British Journal of Psychiatry, 173, 53–63.Salkovskis, P. M., & Harrison, J. (1984). Abnormal and normal obsessions: A

replication. Behavior Research and Therapy, 22, 549–552.Salkovskis, P. M., Rachman, S., Ladouceur, R., Freeston, M., Taylor, S., Kyrios, M.,

et al. (1996). Defining responsibility in obsessional problems. In OC beliefs working

group. Northampton: Smith College.Salkovskis, P. M., Shafran, R., Rachman, S., & Freeston, M. H. (1999). Multiple

pathways to inflated responsibility beliefs in obsessional problems: possible

origins and implications for therapy and research. Behavior Research and

Therapy, 37, 1055–1072.Salkovskis, P. M., & Wahl, K. (2003). Treating obsessional problems using

cognitive-behavioural therapy. In: M. Reinecke, & D. A. Clark (Eds.), Cognitive

therapy across the lifespan: Theory, research and practice (pp. 138–171). Cam-

bridge, UK: Cambridge University Press.

Salkovskis, R. M., Wroe, A. L., Gledhill, A., Morrison, N., Forrester, E., Richards, C.,et al. (2000). Responsibility attitudes and interpretations are characteristic ofobsessive–compulsive disorder. Behavior Research and Therapy, 38, 347–372.

Schaible, R., Armbrust, M., & Nutzinger, D. O. (2001). Yale–Brown ObsessiveCompulsive Scale: Sind Selbst- und Fremdrating aequivalent? [The Yale–Brown Obsessive Compulsive Scale: are self-rating and interview equivalentmeasures?]. Verhaltenstherapie, 11, 298–303.

Shafran, R., Thordarson, D., & Rachman, S. (1996). Thought-action fusion inobsessive–compulsive disorder. Journal of Anxiety Disorders, 10, 379–392.

Sica, C., Steketee, G., Ghisi, M., Chiri, L. R., & Franceschini, S. (2007). Metacognitivebeliefs and strategies predict worry, obsessive–compulsive symptoms andcoping styles: a preliminary prospective study on an Italian non-clinicalsample. Clinical Psychology and Psychotherapy, 14, 258–268.

Smari, J., & Holmsteinsson, H. E. (2001). Intrusive thoughts responsibility attitudes,thought-action fusion, and chronic thought suppression in relation toobsessive–compulsive symptoms. Behavioural and Cognitive Psychotherapy,29, 13–20.

Solem, S., Myers, S. G., Fisher, P. L., Vogel, P. A., & Wells, A. (2010). An empirical testof the metacognitive model of obsessive–compulsive symptoms: Replicationand extension. Journal of Anxiety Disorders, 24, 79–86.

Sookman, D., & Pinard, G. (2002). Overestimation of threat and intolerance ofuncertainty in obsessive–compulsive disorder. In: R. O. Frost, & G. Steketee(Eds.), Cognitive approaches to obsessions and compulsions – Theory, assessment,

and treatment (pp. 63–89). Boston: Pergamon.Steketee, G., Frost, R., & Bogart, K. (1996). The Yale–Brown Obsessive Compulsive

Scale: Interview versus self-report. Behavior Research and Therapy, 34,675–684.

Steketee, G., Frost, R. O., & Cohen, I. (1998). Beliefs in obsessive–compulsivedisorder. Journal of Anxiety Disorders, 12, 525–537.

Tabachnik, B. G., & Fidell, L. S. (2007). Using multivariate statistics. New York:Pearson.

Weinstein, N. D. (1982). Unrealistic optimism about susceptibility to healthproblems. Journal of Behavioral Medicine, 5, 441–460.

Weinstein, N. D., & Klein, W. M. (1996). Unrealistic optimism: Present and future.Journal of Social and Clinical Psychology, 15, 1–8.

Weinstein, N. D., Marcus, S. E., & Moser, R. P. (2005). Smokers’ unrealistic optimismabout their risk. Tobacco Control, 14, 55–59.

Wells, A. (2009). Metacognitive therapy for anxiety and depression. New York:Guilford Press.

Wells, A., & Matthews, G. (1994). Attention and emotion: A clinical perspective. Hove,UK: Erlbaum.

Wells, A., & Papageorgiou, C. (1998). Relationships between worry, obsessive–compulsive symptoms and meta-cognitive beliefs. Behavior Research and

Therapy, 39, 899–913.Wittchen, H. U., Zaudig, M., & Fydrich, T. (1997). Strukturiertes Klinisches Interview

fuer DSM-IV [User’s guide for the Structured Clinical Interview for DSM–IV].Gottingen: Hogrefe.



Responsibility, metacognition and unrealistic pessimism in obsessive–compulsive disorder

Documents