carleton mulvogue thibodeau mccabe antony asmundson 2012

(This is a sample cover image for this issue. The actual cover is not yet available at this time.)

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

http://www.elsevier.com/copyright

Author's personal copy

Journal of Anxiety Disorders 26 (2012) 468– 479

Contents lists available at SciVerse ScienceDirect

Journal of Anxiety Disorders

Increasingly certain about uncertainty: Intolerance of uncertainty acrossanxiety and depression

R. Nicholas Carletona,∗, Myriah K. Mulvoguea, Michel A. Thibodeaua, Randi E. McCabeb,Martin M. Antonyc, Gordon J.G. Asmundsona

a Department of Psychology, University of Regina, Regina, SK, Canadab Department of Psychiatry and Behavioural Neurosciences, McMaster University and Anxiety Treatment and Research Centre, St. Joseph’s Healthcare Hamilton, ON, Canadac Department of Psychology, Ryerson University, Canada

a r t i c l e i n f o

Article history:Received 17 September 2011Received in revised form 23 January 2012Accepted 27 January 2012

Keywords:Intolerance of uncertaintyIUS-12Anxiety disordersDepressionDiagnostic differentiationNormativePsychometric properties

a b s t r a c t

Intolerance of uncertainty (IU) – a dispositional characteristic resulting from negative beliefs about uncer-tainty and its implications – may be an important construct in anxiety disorders and depression. Despitethe potential importance of IU, clinical data on the construct remains relatively scant and focused ongeneralized anxiety disorder and obsessive-compulsive disorder. The present study systematically inves-tigated IU, as measured by the Intolerance of Uncertainty Scale-12 (IUS-12), across groups diagnosed withanxiety disorders (i.e., social anxiety disorder, panic disorder, generalized anxiety disorder, obsessive-compulsive disorder) or depression (clinical sample: n = 376; 61% women), as well as undergraduate(n = 428; 76% women) and community samples (n = 571; 67% women). Analysis of variance revealed onlyone statistically significant difference in IUS-12 scores across diagnostic groups in the clinical sample;specifically, people with social anxiety disorder reported higher scores (p < .01; �2 = .03) than people withpanic disorder. People diagnosed with an anxiety disorder or depression reported significantly and sub-stantially higher IUS-12 scores relative to community and undergraduate samples. Furthermore, IUS-12score distributions were similar across diagnostic groups as demonstrated by Kernel density estimations,with the exception of panic disorder, which may have a relatively flat distribution of IU. Response patternswere invariant across diagnostic groups as demonstrated by multi-group confirmatory factor analyses,but varied between clinical and nonclinical samples. Overall, the findings suggest IU may serve as animportant transdiagnostic feature across anxiety disorders and depression. In addition, robust supportwas found for the proposed 2-factor model of the IUS-12. Comprehensive findings, implications, andfuture research directions are discussed.

© 2012 Published by Elsevier Ltd.

Current perspectives for psychological treatment (Brown &Barlow, 2009; Norton & Philipp, 2008) posit that the etiologicaland neurobiological parallels, as well as considerable diagnosticcomorbidity, render anxiety and mood disorders more alike thanunalike (Wilamowska et al., 2010). Such disorders have been his-torically categorized and conceptualized as discrete pathologies;however, increasing evidence implicates common underlying con-structs (Carleton, Abrams, Asmundson, Antony, & McCabe, 2009;Confer et al., 2010; McEvoy & Mahoney, 2011; Norton & Mehta,2007; Starcevic & Berle, 2006; Taylor, 1993). Identifying and under-standing constructs common to anxiety disorders and depressionmay provide directions for investigating potential transdiagnostic

∗ Corresponding author at: Department of Psychology, University of Regina,Regina, Saskatchewan, S4S 0A2, Canada. Tel.: +1 306 337 2473;fax: +1 306 585 4854.

E-mail addresses: [email protected], [email protected] (R.N. Carleton).

vulnerabilities to fear-related distress, and provide further ratio-nale for developing transdiagnostic treatment protocols.

Anxiety is believed to require a “sense of uncontrollabilityfocused on the possibility of future threat, danger, or other poten-tially negative events” (Suárez, Bennett, Goldstein, & Barlow, 2009,p. 153). Accordingly, there appears to be implicit theoretical sup-port for the notion that negative reactions to uncertainty, knownas intolerance of uncertainty (IU), may be an intrinsic constructfor all anxiety disorders (Asmundson & Carleton, 2005; Carleton,Norton, & Asmundson, 2007; Carleton, Sharpe, & Asmundson, 2007;Holaway, Heimberg, & Coles, 2006). As such, examining IU asa potentially transdiagnostic construct appears well-warranted.There has been considerable discussion regarding the definitionof IU (Starcevic & Berle, 2006). The construct has been broadlyconceptualized as a negative response to ambiguity (Freeston,Rhéaume, Letarte, Dugas, & Ladouceur, 1994) and narrowly as thetendency for an individual to consider the possibility of a negativeevent occurring as unacceptable and threatening irrespective of the

0887-6185/$ – see front matter © 2012 Published by Elsevier Ltd.doi:10.1016/j.janxdis.2012.01.011


R.N. Carleton et al. / Journal of Anxiety Disorders 26 (2012) 468– 479 469

probability of its occurrence (Carleton, Sharpe, et al., 2007; Dugas,Gosselin, & Landouceur, 2001). IU may involve a negative reactionto uncertainty as well as beliefs about the inability to cope withambiguity and change (Holaway et al., 2006; Obsessive CompulsiveCognitions Working Group, 1997); moreover, uncertainty itself canbe considered threatening (Epstein, 1972), promoting or main-taining anxiety, and exacerbating the perception of threat (Dugas,Hedayati, et al., 2005; Dugas, Marchand, & Ladouceur, 2005; Hock& Krohne, 2004).

The most recent definition describes IU as a dispositional char-acteristic resulting from negative beliefs about uncertainty and itsimplications (Dugas & Robichaud, 2007). Available research sug-gests that IU is future-oriented, therein differing from the morepresent-oriented construct described as intolerance of ambigu-ity (Grenier, Barrette, & Ladouceur, 2005). There is increasinglyrobust evidence that IU can be represented as having two dimen-sions – prospective IU (i.e., the cognitively focused dimension ofIU; e.g., Unforeseen events upset me greatly) and inhibitory IU (i.e.,the behaviourally focused dimension of IU; e.g., The smallest doubtcan stop me from acting). Each of the dimensions as representedby the respective factors has been associated with different anx-iety disorder symptoms (McEvoy & Mahoney, 2011). Specifically,the prospective IU subscale was expected to be more related toworry and obsessive compulsive symptoms, whereas the inhibitoryIU subscale was expected to be more related to social anxiety, panic,agoraphobia, and depression.

There is growing evidence that IU is ubiquitous. Researchershave explored IU in large undergraduate (Berenbaum, Bredemeier,& Thompson, 2008; Norton, 2005), community (Sexton & Dugas,2009), and clinical populations (Dugas et al., 2007; McEvoy &Mahoney, 2011; Tolin, Abramowitz, Brigidi, & Foa, 2003), demon-strating a range of scores, substantial construct variability, anda generally normal distribution. In clinical populations, IU hasbeen most thoroughly investigated in generalized anxiety disorder(GAD). Relative to people with a range of other anxiety disor-ders, those with GAD have historically reported higher scores onIU (Ladouceur et al., 1999). An initial comparison was made usingsmall clinical samples comparing persons with a principal diagnosisof GAD (n = 24), an additional diagnosis of GAD (n = 24), other anx-iety disorders (n = 38), and a nonclinical group (n = 20). The resultsindicated significantly higher levels of IU as measured by the Intol-erance of Uncertainty Scale (IUS; Freeston et al., 1994) for thosewith a principal or additional diagnosis of GAD relative to indi-viduals with other anxiety disorders, and all clinical participantsreported higher IU than did nonclinical controls (Ladouceur et al.,1999). In a similar study comparing IUS scores across small samplesof individuals with GAD (n = 17) and panic disorder (n = 28), the GADgroup reported higher scores (Dugas, Hedayati, et al., 2005; Dugas,Marchand, et al., 2005). Similarly, theorists have suggested IU asa broad specifier for worry, the hallmark symptom of GAD, whichis a cognitive strategy used in attempts to control the unknown(Dugas, Buhr, & Ladouceur, 2004; Ladouceur, Gosselin, & Dugas,2000). Indeed, IU is a robust predictor of worry (Buhr & Dugas,2006) and people with high IU have been shown to worry morewhen anxious than not (Buhr & Dugas, 2009).

Contrasting earlier notions that heightened IU distinguishesGAD from other anxiety disorders (Dugas, Marchand, et al., 2005;Ladouceur et al., 1999; Sexton, Norton, Walker, & Norton, 2003),researchers have identified important relationships between IUand other anxiety disorders using regression-based analyses inundergraduate and clinical samples (Gentes & Ruscio, 2011; Norton& Mehta, 2007; Norton, Sexton, Walker, & Norton, 2005). A recentmeta-analysis found no support for the hypothesis that IU may bespecific only to GAD (Gentes & Ruscio, 2011); instead, the authorssuggested that findings indicating IU differentiates GAD from otherdisorders may have been influenced by the GAD-specific content

assessed by the original IUS. Indeed, elevated IU has been associ-ated with OCD at levels comparable to GAD (Holaway et al., 2006),particularly for those people focused on checking (Lind & Boschen,2009; Tolin et al., 2003). IU has also accounted for variance insocial anxiety in comparable degree to the hallmark fear of neg-ative evaluation typically associated with SAD (Boelen & Reijntjes,2009; Carleton, Collimore, & Asmundson, 2010; Norton et al., 2005;van der Heiden et al., 2010). A similar relationship has been foundbetween IU and panic disorder, being comparable in degree to therelationship between panic disorder and anxiety sensitivity (Buhr& Dugas, 2009; Carleton, Sharpe, et al., 2007). There is also substan-tial evidence of an association between IU and depression (Boelen,Vrinssen, & van Tulder, 2010; Butzer & Kuiper, 2006; Miranda,Fontes, & Marroquin, 2008; Norton & Mehta, 2007; Norton et al.,2005; van der Heiden et al., 2010; Yook, Kim, Suh, & Lee, 2010), withsome researchers finding evidence that IU may be more stronglyassociated with depression symptoms than with GAD symptoms(Miranda et al., 2008).

There have been indirect comparisons of IU endorsement rates(i.e., the aggregated Likert-scale item scores within a self-reportmeasure of IU) across disorders made by assessing relative contri-butions of IU to different diagnostic symptoms (Gentes & Ruscio,2011; Norton & Mehta, 2007; Norton et al., 2005; Sexton et al.,2003); moreover, levels of IU have been compared in nonclinicalsamples (Carleton, Sharpe, et al., 2007), between two disorders(Dugas, Marchand, et al., 2005), and between GAD and a range ofother disorders (McEvoy & Mahoney, 2011). Other studies have pro-vided further evidence that IU predicts several types of symptomsbeyond neuroticism (Boelen & Reijntjes, 2009), anxiety sensitiv-ity (Boelen & Reijntjes, 2009; Dugas et al., 2001), fear of anxiety(Buhr & Dugas, 2009), metabeliefs (de Bruin, Rassin, & Muris, 2007;Dugas et al., 2007), and positive and negative affectivity (Carleton,Collimore, et al., 2010), providing increasingly robust evidence forthe broad and important influence of IU in psychopathology. Theaforementioned evidence suggests that IU is important for anxietyand depression symptoms. Nevertheless, the current study repre-sents the first attempt at systematically examining IU endorsementrates and response patterns (i.e., the relative endorsement ratesof different Likert-scale items within a self-report measure of IU)across undergraduate, community, and clinical samples with a vari-ety of relevant anxiety disorders and depression. Such a study isnecessary to consolidate IU as ubiquitous, demonstrate the relativeincrease in persons with psychopathology, and support investi-gating it as a potentially important transdiagnostic construct forthe development and maintenance of anxiety and mood disorders(Carleton, Sharpe, et al., 2007; McEvoy & Mahoney, 2011).

The primary purpose of the present study was to investigatethe endorsement rates and response patterns of IU – as well asdifferences in these rates and patterns – across individuals withvarious principal anxiety disorder diagnoses or depression relativeto undergraduate and community samples. Establishing the pres-ence or absence of differential endorsement patterns representsan important step towards evaluating IU as a transdiagnostic cog-nitive vulnerability factor (Garber & Hollon, 1991; Ingram, 2003).Based on the previous research, IU endorsement was expected tobe higher in the clinical sample relative to both the undergradu-ate and community samples; however, endorsement was expectedto be even higher for persons with a principal diagnosis of GADor OCD relative to persons with other principal diagnoses (Dugas,Marchand, et al., 2005; Ladouceur et al., 1999; Sexton et al., 2003). Inline with previous research (Carleton, Sharpe, et al., 2007; Khawaja& Yu, 2010; Sexton & Dugas, 2009), a 2-factor solution was expectedfor the IUS-12, comprising prospective IU and inhibitory IU. Simi-larly, and based on previous research (McEvoy & Mahoney, 2011),the prospective IU subscale was expected to be more related toworry and obsessive compulsive symptoms, whereas the inhibitory


470 R.N. Carleton et al. / Journal of Anxiety Disorders 26 (2012) 468– 479

Table 1Rates of comorbid diagnoses.

Principal diagnosis Additional diagnosis

None MDD PDA GAD SAD OCD Other Axis I

MDD 8% – 23% 19% 38% 4% 8%PDA 22% 29% – 12% 15% 4% 8%GAD 13% 24% 14% – 27% 10% 13%SAD 17% 32% 13% 21% – 6% 13%OCD 18% 32% 12% 12% 13% – 13%

Notes: MDD, major depressive disorder; PDA, panic disorder with or without agoraphobia; GAD, generalized anxiety disorder; SAD, social anxiety disorder; OCD, obsessive-compulsive disorder.

IU subscale was expected to be more related to social anxiety,panic, agoraphobia, and depression. Given such precedent differ-ences associated with the dimensions of IU, assessing for suchdifferences in the current study appears well warranted.

1. Method

1.1. Participants

The primary participants for this study included patients(n = 376; 146 men [Mage = 36.55; SD = 13.58] and 230 women[Mage = 35.09; SD = 11.81]) from an established outpatient anxietytreatment and research center. Participants received a principalAxis I diagnosis based upon the disorder that was found to be mostdisabling at the time of the assessment, including SAD (n = 120;32%), panic disorder with or without agoraphobia (PDA; n = 89;24%), GAD (n = 63; 17%), OCD (n = 60; 16%), or major depressive dis-order (MDD; n = 26; 7%). The most common additional diagnosesincluded MDD and SAD (Table 1). There were insufficient num-bers of people reporting specific phobia (n = 11) or anxiety disordernot otherwise specified (ADNOS; n = 7) as their principal diagno-sis to include those two groups. In addition, posttraumatic stressdisorder (PTSD) was not included as it is not a disorder treatedat the treatment center. Diagnostic criteria were based on theDiagnostic and Statistical Manual of Mental Disorders (4th ed., textrevision; DSM-IV-TR; American Psychiatric Association, 2000), anddiagnoses were assigned using the Structured Clinical Interview forDSM-IV (SCID-I; First, Spitzer, Gibbon, & Williams, 1996). Most par-ticipants completed at least some postsecondary education (69%),graduated from high school (17%), or completed some high schoolbut did not graduate (12%). The majority described themselves asCaucasian (94%), Asian (3%), or Aboriginal (2%), and as either single(45%), married/cohabitating (46%), or divorced (8%).

Additional participants were included from undergraduate andcommunity samples who completed the demographic question-naire as well as a measure of IU as part of larger, ongoinginvestigations. The two convenience samples were included to con-textualize IU responding by the clinical participants relative to whatcan be presumed to be generally nonclinical samples. In addition,the data from these two samples further expand the normative IUdata available in the literature.

The first normative sample included undergraduates (n = 428;103 men [Mage = 20.58; SD = 3.04] and 325 women [Mage = 20.47;SD = 3.86]) who completed measures as part of other investigationsapproved by the University of Regina Research Ethics Board. Partici-pants identified their ethnicity as Caucasian (87%), Aboriginal (2%),Asian (7%), or other (4%). Most reported being single (82%), mar-ried or cohabitating (12%), separated or divorced (1%), or chose notto answer (5%). Undergraduates were recruited via campus adver-tisements directing them to a secure website for completion of anonline questionnaire package.

The second normative sample included community members(n = 571) from across Canada (187 men [Mage = 27.86; SD = 10.37]

and 384 women [Mage = 28.72; SD = 10.81]) who completed themeasures as part of other web-based investigations approved bythe University of Regina Research Ethics Board. Participants weresolicited with web advertising requests to participate in an onlinestudy of variables associated with anxiety. Most (67%) reportedhaving at least some postsecondary education or completing Grade12 (21%). The sample identified as being single (55%), married orcohabitating (34%), separated or divorced (9%), or chose not toanswer (2%). Most participants identified their ethnicity as Cau-casian (84%), Aboriginal (3%), Asian (5%), or other (8%).

1.2. Measures

Intolerance of uncertainty scale, short form (IUS-12; Carleton,Norton, et al., 2007; Carleton, Sharpe, et al., 2007). The IUS-12is a 12-item short-form of the original 27-item Intolerance ofUncertainty Scale (Freeston et al., 1994) that measures reactionsto uncertainty, ambiguous situations, and the future (e.g., “Unfore-seen events upset me greatly”). Items are scored on a 5-point Likertscale ranging from 1 (not at all characteristic of me) to 5 (entirelycharacteristic of me). The IUS-12 has a strong correlation withthe original scale, rs = .94 to .96 (Carleton, Norton, et al., 2007;Khawaja & Yu, 2010), and has been shown to have two factors,prospective anxiety (7 items; e.g., “I can’t stand being taken by sur-prise”) and inhibitory anxiety (5 items; e.g., “When it’s time to act,uncertainty paralyses me”), both with identically high internal con-sistencies, ̨ = .85 (Carleton, Norton, et al., 2007). The IUS-12 hasexcellent internal consistency and convergent validity with theoriginal (Carleton, Norton, et al., 2007; Carleton, Sharpe, et al.,2007). The psychometric properties of the IUS-12 have all beenreplicated and reified in clinical and nonclinical samples (Carleton,Sharpe, et al., 2007; Khawaja & Yu, 2010; McEvoy & Mahoney,2011).

The IUS-12 was selected to measure IU because it is psychome-trically comparable to, but briefer than, the original IUS (Khawaja& Yu, 2010) and the new symptom-focused Intolerance of Uncer-tainty Index (IUI; Carleton, Gosselin, & Asmundson, 2010; Gosselinet al., 2008). In addition, while the IUI was developed largely as aclinical and outcome measure, the IUS-12 has been designed specif-ically to research the construct in a number of different populations,across a number of different disorders.

McEvoy and Mahoney (2011) have provided a compelling argu-ment that prospective IU and inhibitory IU are more appropriatelabels for the IUS-12 subscales, reflecting the fact that responseshave not been anxiety-specific; accordingly, the revised subscalenames have been adopted herein. In addition, the subscales will beassessed separately, as well as the total score. Evidence to date hasindicated differential discriminant validity associated with eachsubscale, such that prospective IU appears more strongly associ-ated with GAD and OCD (i.e., anticipation of uncertainty), whereasinhibitory IU appears more strongly associated with panic dis-order, SAD, and depression (i.e., uncertainty produces inhibition;Carleton, Collimore, et al., 2010; McEvoy & Mahoney, 2011).



In the clinical sample, the internal consistency for the totalscore ( ̨ = .91), the prospective IU subscale score ( ̨ = .87), and theinhibitory IU subscale score ( ̨ = .86) ranged from good to excellent,and the average inter-item correlation was .47. In the undergrad-uate sample, the internal consistency for the total score ( ̨ = .91),the prospective IU subscale score ( ̨ = .85), and the inhibitory IUsubscale score ( ̨ = .86) ranged from good to excellent, and theaverage inter-item correlation was .45. In the community sample,the internal consistency for the total score ( ̨ = .92), the prospec-tive IU subscale score ( ̨ = .86), and the inhibitory IU subscale score( ̨ = .89) ranged from good to excellent, and the average inter-itemcorrelation was .48.

Structured Clinical Interview for DSM-IV (SCID-I; First et al., 1996).The SCID-I is an established semi-structured clinical interviewused to facilitate diagnosis of Axis I disorders based on DSM-IV criteria. The SCID was administered only to participants fromthe established outpatient anxiety treatment and research center.Interviewers in the present study included psychologists, post-doctoral fellows, and predoctoral graduate students, all of whomhad received extensive training and supervision in conducting thisinterview. All SCID-I interviews for this study were presented at aweekly team meeting chaired by a psychologist with more than 5years of experience in training others to administer the measure.At each meeting, diagnostic questions were reviewed, and a con-sensus diagnosis was reached. Earlier versions of the SCID-I havebeen found to have adequate inter-rater reliability for all disorders(overall reliabilities range from .69 to 1.0; Zanarini & Frankenburg,2001).

1.3. Analyses

Descriptive statistics were calculated for each of the individualitems and summed total and subscale scores for each of the diagnos-tic groups, as well as the undergraduate and community samples.Across the diagnostic groups, global assessment of function (GAF)was also assessed (American Psychiatric Association, 2000). The100-point GAF was used to determine disorder severity ratherthan the 3-point SCID-I severity index because, with differentia-tion between ten rather than three symptom severity/impairmentsubcategories, and the freedom to use intermediate level ratings(e.g., 42, 75) when appropriate, the GAF offers a richer, and arguablymore accurate, indication of disorder severity.

Comparative analyses were performed across men and womento assess for sex differences. Empirical distributions of IUS-12scores across diagnostic and community and undergraduate groupswere studied using Kernel density estimation curves. A Kernel den-sity estimation is a data smoothing algorithm wherein populationinferences are made based on an empirical sample to draw sampledistribution characteristics (Salgado-Ugarte & Perez-Hernandez,2003). Frequency histograms effectively display relatively discretedata of one variable; however, examining and comparing multiplehistograms (e.g., between groups) is difficult, and these typicallydo not serve as good representations of relatively continuous data(Scott, 1979). Compiled univariate Kernel density estimation curvesallow a parsimonious examination of distributions across numer-ous groups on one variable and in one plot. A Gaussian function wasused with a bandwidth of 1 to compile the curves. Visual inspec-tion of the plot revealed how distribution features (e.g., variance,skew, kurtosis) and modality (e.g., relative normality, bimodal-ity) differed across diagnostic groups (Salgado-Ugarte, Shimizu, &Taniuchi, 1994).

There were two analyses of variance (ANOVA) conducted tocompare total and subscale score means of the IUS-12 across diag-nostic and undergraduate and community groups. Bootstrappingwas performed to ensure the robust nature of statistically sig-nificant results (Byrne, 2001; Davison & Hinkley, 2006; Nevitt

& Hancock, 2001). The first ANOVA replicated and extended theANOVA conducted by Ladouceur et al. (1999) by comparing par-ticipants with a principal diagnosis of GAD to those with (1)an additional diagnosis of GAD, (2) a principal diagnosis of anyother anxiety disorder, (3) a principal diagnosis of MDD, (4) theundergraduate sample, and (5) the community sample. Tukey posthoc comparisons were conducted to assess for individual differ-ences between group means (p < .05). Given the large number ofcomparisons, the alpha level was adjusted to p = .01. Additionaladjustments to control for Experiment-Wise Type I error were notincluded to avoid the risk of a Type II error (Tabachnick & Fidell,2007); accordingly, effect sizes were also reported and interpreted(Osborne, 2008). The second ANOVA was used to compare the totaland subscale scores between each of the principal diagnostic groups(rather than collapsing diagnoses as in the first ANOVA) as well asthe undergraduate and community samples. Between-group dif-ferences were again delineated with post hoc Tukey tests.

Confirmatory factor analyses (CFA) were used for three pur-poses: first, to assess the fit of the presumed IUS-12 2-factorstructure for each of the clinical, undergraduate, and communitysamples; second, to assess if the IUS-12 factor structure, mea-surements weights (i.e., the relationship between the measuredvariables and their latent variables), and structural covariances (i.e.,the covariances among the latent variables) differed between menand women; third, to assess if the IUS-12 factor structure, measure-ments weights, and structural covariances differed between eachsample (i.e., clinical, undergraduate, and community) and betweeneach diagnostic, and undergraduate, and community groups. Amultiple-group CFA procedure in AMOS as described by Byrne(Byrne, 2001, 2004) was utilized to reach these purposes. Multiplegroup analysis in structural equation modeling allows comparisonsof the same construct across samples for any identified structuralequation model. AMOS allows testing of whether the groups meetan assumption of equality by examining whether different sets ofpath coefficients are invariant. Statistically significant differencesin measurement weights would suggest that subsequent analy-ses of structural covariances may not be robust; therefore, if themeasurement weights differ, but the structural covariances do not,then the response patterns are similar but cannot be assumed tobe entirely comparable across groups. This procedure serves as astringent test of invariance across men and women for each of thethree samples and across each of the diagnostic groups (i.e., SAD,PDA, GAD, OCD, MDD) to test for invariance based on diagnosis.

The unitary model was assessed with confirmatory factor anal-yses using the following fit indices and 90 percent confidenceintervals (where applicable) as representative of excellent fit andvalues approaching these cut off scores as indicating an increas-ingly good fit (Hu & Bentler, 1999; Tabachnick & Fidell, 2007): (1)chi-square (values should not be significant); (2) chi-square/df ratio(values should be less than 2.0); (3) Comparative Fit Index (CFI;values must be greater than .90, and ideal fits approach or aregreater than .95); (4) the Standardized Root Mean Square Resid-ual (SRMR; values must be less than .10 and ideal fits approachor are less than .05); (5) Root Mean Square Error of Approxima-tion (RMSEA; values must be less than .08 and ideal fits approachor are less than .05, with 90% confidence interval values below.10); and (6) Expected Cross-Validation Index (ECVI); when com-paring these scores across different models, lower values indicatea closer fit (Browne & Cudeck, 1989, 1993). Goodness of fit eval-uations should emphasize the latter four fit indices because ofpotential chi-square inflation (Hu & Bentler, 1999). Multivariatenormality was assessed using Mardia’s coefficient of multivari-ate kurtosis (Byrne, 2001) for all models, with results suggestingnonnormal data; however, parameter estimates and most modelfit indices are robust to nonnormality given maximum-likelihoodestimation and a sample size of 100 or more participants (Lei &



Lomax, 2005). Nonetheless, we used the Bollen–Stine bootstrapchi-square and computed bootstrapped parameter estimates withestimates from a maximum-likelihood procedure, which has beendemonstrated as an adequate method for resolving non-normality(Byrne, 2001; Davison & Hinkley, 2006; Nevitt & Hancock, 2001).In all cases, the statistical significance value for the Bollen–Stinebootstrap chi-square produced results comparable with those fromthe maximum-likelihood procedure for the CFA, suggesting thatnon-normality did not substantially impact the overall results.

2. Results

2.1. Descriptive statistics and sex comparisons

Descriptive statistics are presented in Table 2. There was ahigher proportion, �2(2) = 21.56, p < .01, V = .13, of women in theundergraduate sample (76%) relative to the clinical sample (61%)and the community sample (67%). There were no differences in agebetween the diagnostic groups, F(4, 351) = 2.24, p > .05, eta2 = .03;however, the undergraduate sample was significantly younger thanthe community sample (mean difference = 7.14; p < .01), whichwas significantly younger than the clinical sample (mean differ-ence = 7.94; p < .01), F(2, 1346) = 241.51, p < .01, eta2 = .26. Despitethe statistically significant differences in age, in the undergrad-uate and community samples the correlations between age andeach of the IUS total and subscale scores were not statisticallysignificant (all ps > .05, all rs < .04). In the clinical sample, the corre-lations between age and each of the IUS total score, r(355) = −.11,p = .03, prospective IU score, r(355) = −.11, p = .04, and inhibitory IUscore, r(355) = −.10, p = .06, while at or approaching statistical sig-nificance, were all very small (Cohen, 1988). Accordingly, age wasnot considered further in the analyses. GAF scores from the SCID-I – used as a measure of diagnostic severity – ranged from 40 to85 across all groups (i.e., PDA 40–85; GAD 48–78; OCD 50–82; SAD40–75; MDD 50–80). There were statistically significant differencesin GAF scores across diagnostic groups, F(4, 347) = 3.73, p < .01,eta2 = .04; however, the effect size was relatively small. Following aTukey correction, persons in the SAD group reported slightly lowerscores than persons in the GAD (mean difference = 3.55; p < .05) andOCD (mean difference = 3.65; p < .05) groups. All other group com-parisons, including across men and women, t(367) = 1.43, p > .05,r2 < .01, were non-significant. Given the small differences and effectsizes, as well as the relatively few between group differences, sever-ity was not considered further in the analyses and participantswere not further subdivided or analysed based on a sex by GAFinteraction.

Regarding the IUS-12, there were no items or summed scalealternatives that demonstrated unacceptable levels of skew or kur-tosis (i.e., none had positive standardized skewness values thatexceeded 2 or positive standardized kurtosis values that exceeded7; see Curran, West, & Finch, 1996; Tabachnick & Fidell, 2007). Therewere also almost no statistically significant differences betweenmen and women (i.e., all ps > .10; r2s < .01) on the total or subscalescores in any sample. The exception was that men scored slightlyhigher than women on the inhibitory IU subscale in the communitysample, t(569) = 2.25, p < .05, r2 < .01; however, the effect size wasextremely small.

2.2. Distribution estimations and ANOVA results

The Kernel distribution estimations for all groups are reportedin Fig. 1. Three general patterns emerged. First, SAD, GAD, OCD, andMDD groups demonstrated very similar distributions falling withinthe higher range of IUS-12 scores, and these distributions appearedrelatively normal to negatively skewed. Second, the undergraduate

Fig. 1. Kernel density estimation of IUS-12 scores across groups, with densityreflecting frequency of cases along IUS-12 scores.

and community samples demonstrated very similar distributions toeach other that were disparate from the diagnostic groups, and thedistributions were positively skewed, centering around lower IUS-12 scores. Third, individuals with panic disorder reported a widerrange of IUS-12 scores compared to those with other mental dis-orders and those without. These findings suggest that variance inthe undergraduate and community samples may be attributableto infrequent extremely high IUS-12 scores, while variance in thediagnostic groups is a result of more frequent modest deviationsfrom the mean and some infrequent low IUS-12 scores.

The substantially different sample sizes within each of the diag-nostic groups in the clinical sample, while not prohibitive forANOVA, does make meeting the assumption of homogeneity ofvariance more important (Tabachnick & Fidell, 2007); however,even if violated, a Welch correction can be employed along withdiscriminating post hoc tests to ensure any statistically significantdifferences are likely to be robust (Judd, McClelland, & Culhane,1995; Tabachnick & Fidell, 2007). In both ANOVAs the assump-tion of homogeneity was violated (p < .05) for the total score andboth subscales of the IUS-12. Accordingly, a Welch correction wasapplied to all F-values.

The first ANOVA (replicating Ladouceur et al., 1999) collapseddiagnostic groups creating groups with a principal diagnosis of GAD(n = 63; i.e., with or without an additional diagnosis of another anx-iety disorder), an additional diagnosis of GAD (n = 49; i.e., anotherprincipal diagnosis, but with a concurrent additional diagnosis ofGAD), a principal diagnosis of any other anxiety disorder (n = 225;i.e., no principal or additional diagnosis of GAD), a principal diag-nosis of MDD (n = 21), the undergraduate sample (n = 428), and thecommunity sample (n = 571). Statistically significant differenceswere identified for the total score, F(5, 134.32) = 70.14, p < .001,eta2 = .20, the prospective IU subscale, F(5, 134.61) = 46.05, p < .001,eta2 = .14, and the inhibitory IU subscale, F(5, 134.32) = 83.02,p < .001, eta2 = .23. Tukey post hoc comparisons were conducted toassess for individual differences between group means. The resultsof the Tukey post hoc comparisons were generally comparable forthe total and subscale scores. There were no statistically signifi-cant differences between participants with a principal diagnosisof GAD, an additional diagnosis of GAD, a principal diagnosis ofany other anxiety disorder, or a principal diagnosis of MDD (i.e.,

Author's personal copyR

.N.

Carleton

et

al.

/

Journal

of

Anxiety

Disorders

26 (2012) 468– 479473

Table 2Descriptive statistics.

IUS-12and GAFScore

Undergraduates (n = 428) Community (n = 571) SAD (n = 120) PDA (n = 89) GAD (n = 63) OCD (n = 60) MDD (n = 26)

M (SD) S (.12)/K (.24) M (SD) S (.10)/K (.20) M (SD) S (.22)/K (.44) M (SD) S (.26)/K (.51) M (SD) S (.30)/K (.60) M (SD) S (.31)/K (.61) M (SD) S (.46)/K (.89)

Item 1 2.14(1.08)

.70/−.24

2.25(1.22)

.71/−.52

3.69(1.13)

−.47/−.78

3.25(1.30)

−.25/−.98

3.64(1.32)

−.62/−.83

3.27(1.29)

−.37/−.86

3.54(1.07)

−.43/−.17

Item 2 3.26(1.07)

−.16/−.71

3.34(1.21)

−.23/−.93

3.87(1.07)

−.78/−.12

3.47(1.35)

−.46/−.98

4.00(1.11)

−1.03/.59

3.88(1.09)

−.82/.04

3.96(.96)

−1.10/2.11

Item 3 2.02(1.15)

.92/−.09

2.36(1.39)

.63/−.88

3.82(1.24)

−.76/−.48

3.54(1.41)

−.53/−1.07

3.71(1.26)

−.78/−.41

3.82(1.23)

−.94/.11

3.88(1.14)

−.80/−.01

Item 4 2.49(1.15)

.40/−.69

2.79(1.26)

.19/−.95

3.32(1.35)

−.20/−1.11

2.99(1.24)

.21/−.97

3.62(1.41)

−.68/−.84

3.52(1.17)

−.47/−.50

3.62(1.10)

−.52/−.23

Item 5 2.19(1.14)

.71/−.36

2.34(1.27)

.56/−.80

3.22(1.39)

−.15/−1.22

2.85(1.43)

.12/−1.35

3.16(1.32)

−.13/−1.15

3.25(1.35)

−.30/−1.16

3.38(1.33)

−.34/−.84

Item 6 1.80(.97)

1.09/.48

2.01(1.19)

1.05/.12

3.36(1.34)

−.41/−.98

2.67(1.43)

.33/−1.20

2.71(1.34)

.18/−1.12

3.15(1.42)

−.20/−1.28

3.46(1.14)

−.51/−.03

Item 7 2.15(1.01)

.70/

.022.36(1.26)

.55/−.80

3.69(1.08)

−.63/−.28

3.06(1.33)

.04/−1.19

3.17(1.34)

−.16/−1.22

3.58(1.29)

−.63/−.77

3.85(.88)

−.44/−.27

Item 8 2.77(1.17)

.11/−.84

2.80(1.35)

.18/−1.12

3.35(1.27)

−.36/−.88

3.21(1.34)

−.23/−1.13

3.54(1.50)

−.49/−1.24

3.40(1.36)

−.27/−1.15

3.69(1.09)

−.74/.10

Item 9 1.97(1.08)

.91/−.00

2.23(1.26)

.73/−.59

3.15(1.28)

−.19/−1.09

2.87(1.37)

.19/−1.17

3.06(1.49)

−.05/−1.45

2.92(1.23)

.11/−.92

3.50(1.17)

−.64/−.20

Item 10 2.14(1.16)

.78/−.36

2.26(1.26)

.71/−.60

3.64(1.23)

−.53/−.74

3.07(1.29)

.03/−1.08

3.21(1.32)

−.18/−1.08

3.53(1.32)

−.58/−.69

3.38(1.17)

−.51/−.45

Item 11 2.69(1.19)

.18/−.91

2.80(1.25)

.10/−.98

3.38(1.25)

−.35/−.79

3.19(1.35)

−.16/−1.17

3.57(1.34)

−.66/−.76

3.38(1.43)

−.28/−1.33

3.54(1.21)

−.47/−.93

Item 12 1.89(1.05)

1.10/.55

2.01(1.21)

1.04/.03

3.18(1.25)

−.26/−.96

2.84(1.26)

.18/−.82

2.98(1.39)

.07/−1.24

3.08(1.32)

−.25/−1.14

3.23(1.27)

−.22/−.80

Pro 17.51(5.68)

.42/−.15

18.54(6.50)

.37/−.53

23.97(6.48)

−.23/−.56

21.83(7.49)

.07/−.96

24.60(6.84)

−.22/−.92

23.62(6.45)

−.28/−.88

25.23(5.32)

.15/−1.24

Inh 10.00(4.30)

.94/

.4010.99(5.23)

.81/−.23

17.69(4.75)

−.64/−.42

15.18(5.55)

.02/−1.08

15.79(5.19)

−.16/−.58

17.17(5.28)

−.63/−.44

17.81(4.51)

−.60/.90

Total 27.52(9.28)

.67/

.1629.53(10.96)

.59/−.32

41.65(10.23)

−.38/−.50

37.01(12.45)

.07/−.92

40.38(11.26)

−.18/−.70

40.78(10.71)

−.46/−.35

43.04(9.20)

−.10/−.64

GAF – – – – 59.52(7.45)

– 60.08(8.77)

– 63.06(7.28)

– 63.17(7.95)

– 59.52(8.12)

–

Notes: MDD, major depressive disorder; PDA, panic disorder with or without agoraphobia; GAD, generalized anxiety disorder; SAD, social anxiety disorder; OCD, obsessive-compulsive disorder; S, skew; K, kurtosis; IUS-12,Intolerance of Uncertainty Scale-12; IUS-12 Pro, prospective IU subscale; IUS-12 Inh, inhibitory IU subscale; GAF, global assessment of functioning.



all ps > .10). In contrast, all of the aforementioned clinical group-ings reported statistically significantly higher scores than both theundergraduate and community samples (i.e., all ps < .01). Therewere no statistically significant differences between the under-graduate and community samples. Given that no differences werefound from this first ANOVA, only the second ANOVA results arepresented in detail; however, the results of the first ANOVA areavailable from the authors upon request.

In the second ANOVA, the comparisons of response meansbetween each of the groups (using principal diagnoses for theclinical sample, but not excluding participants with additional diag-noses) revealed statistically significant differences for the totalscore, F(6, 182.90) = 58.59, p < .001, �2 = .20, the prospective IU sub-scale, F(6, 183.03) = 35.87, p < .001, �2 = .14, and the inhibitory IUsubscale, F(6, 182.69) = 73.45, p < .001, �2 = .24. The results of theTukey post hoc comparisons were generally comparable for thetotal and subscale scores (i.e., all but two ps > .10). Specifically,there were no statistically significant differences between any ofthe diagnostic groups within the clinical sample (see Table 2)despite a power analysis indicating sufficient sample sizes to detectdifferences associated with a medium effect size (i.e., total sam-ple n > 200; moderate effect size of f2 = .25, ̨ = .05 (1-tailed), andpower = .85). Similarly, there were no differences between theundergraduate and community samples. In contrast, the SAD, PDA,GAD, OCD, and MDD diagnostic groups all reported statisticallysignificantly higher total scores than both the undergraduate andcommunity samples. Overall, the results suggest that persons witha diagnosis of SAD, PDA, GAD, OCD, or MDD report higher levels ofIU than undergraduate and community samples. The exceptionsto the Tukey comparisons involved people diagnosed with SADreporting slightly higher scores on the inhibitory IU subscale (meandifference = 2.51; p < .01) and therein the IUS total score (mean dif-ference = 4.65; p < .05) relative to people diagnosed with PDA.

The presence of comorbid diagnoses could have masked differ-ences in IU associated with any single disorder. Consequently, athird ANOVA was performed assessing IU across individuals withonly one diagnosis; however, this was not an a priori analysis.In short, the second ANOVA was re-run, but excluded partici-pants with a comorbid diagnosis. The additional criterion furtherrestricted comparative power (i.e., SAD, n = 23, PDA, n = 28; GAD,n = 12; OCD, n = 12; MDD, n = 2; MDD was excluded from the com-parisons because of an insufficient sample size of n = 2), so no Type Icorrections were used; nevertheless, no statistically significant (allps > .05) differences were found between the diagnostic groups. Assuch, the details of this third ANOVA, including the means and stan-dard deviations, have not been presented but can be requested fromthe authors.

2.3. Factor analytic results

The fit indices generally supported the 2-factor structure formen and women in each of the three samples (i.e., clinical, under-graduate, and community); moreover, the 2-factor structure wasstatistically superior to a unitary structure for men and womenin all three samples (Table 3). The fit indices also generally sup-ported the 2-factor structure across men and women in each ofthe three samples; however, relative to the other samples, the clin-ical sample as a whole had the poorest fit indices, possibly dueto the relatively higher variability. In any case, the 2-factor struc-ture was, again, statistically superior to a unitary structure in allthree samples as measured by chi-square difference tests (Table 3).The sample sizes associated with some of the diagnostic groupswithin the clinical sample would typically be considered small (i.e.,the GAD and OCD groups) or insufficient (i.e., the MDD group) forCFA; nevertheless, in an effort to thoroughly present available dataand determine whether a 2-factor or unitary solution would be

superior, all were tested with CFA. The fit indices associated withthe 2-factor structure were inconsistent and generally did not meetthe cut-off indices, as should be expected from relatively smallsamples. Despite that inconsistency, in all cases the 2-factor struc-ture was consistently significantly superior to the unitary structure,except for MDD where no difference was found.

2.4. Invariance analyses results

There were no differences between men and women based onmeasurement weights or structural covariances for any of the threesamples (Table 4). In other words, the response patterns for menand women were comparable in all three samples (i.e., clinical,undergraduate, and community). Despite all three samples demon-strating comparable response patterns for men and women, theclinical sample demonstrated the least difference between menand women, whereas the undergraduate sample demonstrated themost. As such, men and women were assessed together for thesubsequent invariance analyses.

There were no differences between the undergraduate andcommunity samples based on measurement weights (Table 4), sug-gesting the response patterns were largely similar; however, thestructural covariances were significantly different, indicating atleast some variance. In contrast, there were significant and sub-stantial differences in responses patterns based on measurementweights or structural covariances between the clinical sample andeach of the undergraduate and community samples. Even thoughthe fit indices associated with the 2-factor structure assessed ineach diagnostic group were insufficient, likely due to insufficientsample sizes, there were no differences across the diagnostic groups(i.e., SAD, PDA, GAD, OCD, MDD) based on measurement weightsor structural covariances. In any case, if there was a robust dif-ference in the response patterns across each diagnostic group, therelatively poor group fit indices and small sample sizes should haveincreased, not decreased, such differences, particularly given howmany groups were compared simultaneously. As such, it appearsthat within the clinical sample the response patterns are invariantacross each of the diagnostic groups.

3. Discussion

The present study makes three important contributions to theIU literature. First, the study presents the first direct comparativeanalyses of IU response patterns and empirical distributions – asmeasured by the IUS-12 – across clinical samples of people endors-ing criteria for a principal diagnosis of GAD, OCD, SAD, PDA, orMDD relative to undergraduate and community samples. Doing soprovided an empirical assessment of how IU may not be specificto any given anxiety disorders or depression, which would shedlight on a construct that appears to represent a cognitive vulner-ability transdiagnostic across numerous anxiety disorders (Garber& Hollon, 1991; Gentes & Ruscio, 2011; Ingram, 2003; McEvoy &Mahoney, 2011). Second, the same data allowed for the first pre-sentation of normative IUS-12 data for several clinical samples, andfurther assessments of the proposed 2-factor model for the IUS-12.Third, the results lend further psychometric support to the IUS-12as being invariant across sex with a robust factor structure whenadministered as an independent measure (rather than as a subsetof the original IUS).

Previous assessments with the IUS-12 (Carleton, Norton, et al.,2007; Khawaja & Yu, 2010; McEvoy & Mahoney, 2011) and withother scales for measuring IU (Carleton, Gosselin, et al., 2010;Gosselin et al., 2008; Robichaud, Dugas, & Conway, 2003) havedemonstrated broadly normal distributions with negligible differ-ences in endorsement of IU between men and women, which is

Author's personal copyR

.N.

Carleton

et

al.

/

Journal

of

Anxiety

Disorders

26 (2012) 468– 479475

Table 3Confirmatory factor analyses fit indices.

2-Factor Undergraduate sample Community sample Clinical sample Clinical diagnostic groups

Total Men Women Total Men Women Total Men Women SAD PDA GAD OCD MDD

n 428 103 325 571 187 384 376 146 230 120 89 63 60 26�2 173.933 92.498 145.234 182.462 97.550 140.339 244.935 106.484 188.220 133.143 106.853 128.570 80.781 81.759df 53 53 53 53 53 53 53 53 53 53 53 53 53 53�2/df 3.282 1.745 2.740 3.443 1.841 2.648 4.621 2.009 3.551 2.512 2.016 2.426 1.524 1.543CFI .948 .925 .949 .963 .953 .966 .919 .930 .917 .875 .925 .805 .916 .813RMSEA .073 .085 .073 .065 .067 .066 .098 .083 .106 .113 .107 .152 .094 .147RMSEA CI .061; .085 .055; .114 .059; .088 .055; .076 .046; .088 .053; .079 .086; .111 .060; .106 .089; .122 .089; .137 .078; .137 .118; .185 .049; .134 .079; .208SRMR .046 .065 .046 .040 .057 .037 .058 .064 .061 .072 .052 .106 .095 .119ECVI .524 1.397 .603 .408 .793 .497 .786 1.079 1.040 1.539 1.782 2.880 2.217 5.270ECVI CI .440; .627 1.173; 1.698 .504; .725 .342; .487 .665; .964 .415; .599 .667; .926 .903; 1.309 .874; 1.240 1.285; 1.857 1.491; 2.163 2.405; 3.479 1.872; 2.696 4.448; 6.411

1-Factor Undergraduate sample Community sample Clinical sample Clinical diagnostic groups

Total Men Women Total Men Women Total Men Women SAD PDA GAD OCD MDD

�2 258.146 112.116 212.712 308.559 133.882 232.134 361.531 153.031 260.288 172.295 121.557 139.106 121.106 82.546df 54 54 54 54 54 54 54 54 54 54 54 54 54 54�2/df 4.780 2.076 3.939 5.714 2.479 4.299 6.695 2.834 4.820 3.191 2.251 2.576 2.243 1.529CFI .912 .890 .913 .927 .916 .930 .870 .870 .873 .815 .906 .780 .797 .814RMSEA .094 .103 .095 .091 .089 .093 .123 .112 .129 .136 .119 .159 .145 .145RMSEA CI .083; .106 .076; .130 .082; .109 .081; .101 .070; .108 .081; .105 .111; .135 .092; .134 .114; .145 .113; .159 .091; .148 .127; .192 .111; .180 .077; .206SRMR .055 .071 .055 .052 .066 .050 .063 .072 .064 .082 .056 .104 .093 .124ECVI .717 1.570 .805 .626 .978 .731 1.092 1.386 1.346 1.851 1.927 3.018 2.866 5.222ECVI CI .609; .843 1.310; 1.906 .678; .955 .535; .729 .815; 1.182 .618; .864 .942; 1.236 1.158; 1.668 1.143; 1.582 1.549; 2.217 1.606; 2.335 2.517; 3.643 2.390; 3.474 4.397; 6.365Comparing 2-factor and unitary structures

�2 84.213 19.618 67.478 126.097 36.332 91.795 116.596 46.547 72.068 39.152 14.704 10.536 40.325 .787p <.05 <.05 <.05 <.05 <.05 <.05 <.05 <.05 <.05 <.05 <.05 <.05 <.05 >.05

Notes: CFI, Comparative Fit Index; SRMR, Standardized Root Mean Square Residual; RMSEA, Root Mean Square Error of Approximation; ECVI, Expected Cross-Validation Index; CI, 90% confidence intervals; MDD, major depressivedisorder; PDA, panic disorder with or without agoraphobia; GAD, generalized anxiety disorder; SAD, social anxiety disorder; OCD, obsessive-compulsive disorder.



Table 4Invariance analyses.

Sample(s) Comparison groups Measurement weights Structural covariances†

Clinical (n = 358) Men–women �2(10) = 10.05, p = .43 �2(3) = .85, p = .84 InvariantUndergraduate (n = 428) Men–women �2(10) = 17.70, p = .06 �2(3) = .87, p = .83 InvariantCommunity (n = 571) Men–women �2(10) = 13.43, p = .20 �2(3) = 3.85, p = .28 InvariantUndergraduate and community Undergraduate-community �2(10) = 11.83, p = .30 �2(3) = 16.99, p < .01 Partially variantClinical and undergraduate Clinical-undergraduate �2(10) = 33.92, p < .01 n/a VariantClinical and community Clinical-community �2(10) = 35.44, p < .01 n/a VariantClinical MDD–PDA–GAD–SAD–OCD �2(40) = 34.14, p = .73 �2(12) = 17.66, p = .13 Invariant

Notes: MDD, major depressive disorder; PDA, panic disorder with or without agoraphobia; GAD, generalized anxiety disorder; SAD, social anxiety disorder; OCD, obsessive-compulsive disorder.

† Assumes measurement weights were correct and should not be interpreted it the measurement weights were not invariant; invariance is indicated by a lack of statisticalsignificance.

supported by the current findings. To further assess for poten-tial differences between men and women, the 2-factor structurewas assessed using an invariance analysis (Byrne, 2001, 2004). Theinvariance analyses, which are quite stringent, produced no dif-ferences between men and women, providing evidence that theendorsement patterns for men and women appear comparable irre-spective of whether the sample is broad (i.e., undergraduate orcommunity) or range-restricted (i.e., clinical). As such, IU as mea-sured by the IUS-12 appears to be a ubiquitous construct, generallyindiscriminant based on sex.

People in the clinical sample were expected to endorse higherlevels of IU than people in the undergraduate and community sam-ples. As expected, people with a principal diagnosis of SAD, PDA,GAD, OCD, or MDD endorsed significantly higher levels of IU thandid people in the undergraduate and community samples. Therewere no differences between the undergraduate and communitysamples. Furthermore, the empirical distributions of IU for thosewith a principle diagnosis of SAD, GAD, OCD and MDD were verysimilar, suggesting that the groups of individuals with these disor-ders reported not only similar levels of intolerance of uncertainty,but that IU shares a relatively symmetrical distribution across allof these disorders. Nevertheless, these findings did not general-ize to those with PDA, as they reported a relatively larger varianceof scores compared to those with other disorders and undergradu-ate and community participants. These findings are congruent withprevious findings suggesting that IU may be associated with PDAdifferently than in other anxiety disorders or depression (Norton& Mehta, 2007). Future research may benefit from examining thespecific distribution of IU within a larger PDA sample.

Based on the theoretical position that IU is a characteristic fea-ture unique to GAD or OCD, participants endorsing either disorderas principal were expected to report higher levels of IU rela-tive to participants endorsing other disorders as principal (Dugaset al., 2004; Ladouceur et al., 2000; Lind & Boschen, 2009; Tolinet al., 2003); however, there were almost no statistically significantdifferences between any of diagnostic groups within the clini-cal sample. There were also no statistically significant differenceswhen the Ladouceur et al. (1999) analyses were replicated. Specif-ically, there were no differences in IU scores between persons withprincipal diagnosis of GAD, an additional diagnosis of GAD, a princi-pal diagnosis of any other anxiety disorder, or a principal diagnosisof MDD. There were also no differences in IU scores between theundergraduate sample, and the community sample.

An exception to the comparability of IU across clinical samplesinvolved people with a principal diagnosis of SAD who reportedslightly higher scores on the inhibitory IU subscale than those witha principal diagnosis of PDA. Differential associations between thesubscales and symptoms are not unprecedented. Previous resultshave associated prospective IU with GAD and OCD (i.e., anticipa-tion of uncertainty; McEvoy & Mahoney, 2011), and associatedinhibitory IU with panic, SAD, and depression (i.e., uncertainty

produces inhibition; Carleton, Gosselin, et al., 2010; McEvoy &Mahoney, 2011). As speculated implicitly by some (Miranda et al.,2008; Yook et al., 2010) and explicitly by McEvoy and Mahoney,“while IU appears to be a transdiagnostic maintaining mechanism,[prospective IU and inhibitory IU] may play different roles acrossdifferent disorders” (2011, p. 121). In any case, the general compa-rability of endorsement rates, patterns, and distributions, suggestsagainst IU as an underlying feature of only worry-specific anxietydisorders such as GAD and some OCD. Instead, the results lendfurther support to postulate that IU may be a broad vulnerabil-ity factor for most anxiety disorders and depression (Boelen et al.,2010; Butzer & Kuiper, 2006; Carleton, Sharpe, et al., 2007; McEvoy& Mahoney, 2011; Yook et al., 2010).

Previous research has supported the proposed 2-factor struc-ture of the IUS-12 as being robust across undergraduate samples(Carleton, Norton, et al., 2007; Carleton, Sharpe, et al., 2007) andrecently supported the same structure in a broad sample of peo-ple seeking treatment for anxiety disorders (McEvoy & Mahoney,2011). In the current study, the 2-factor structure (i.e., prospectiveIU and inhibitory IU) was supported in all three samples and thefit indices were superior to a unitary alternative. The sample sizesfor each of the diagnostic groups within the clinical sample wouldtypically be considered insufficient for robust CFA; however, thepreliminary analyses nonetheless supported the 2-factor structurerelative to a unitary structure for each diagnostic group within theclinical sample, except for MDD where no difference was found.

The results of the multi-group invariance analyses across thesamples (i.e., clinical, undergraduate, and community) and acrossthe diagnostic groups with the clinical sample (i.e., SAD, PDA, GAD,OCD, MDD) were relatively definitive. The response patterns ofundergraduate and community participants were largely compa-rable. In contrast, the response patterns of clinical participantswere substantially different from each of the undergraduate andcommunity participants, with the differences themselves beingcomparable in magnitude. The results suggest not only do clin-ical participants endorse higher rates of IU than undergraduateand community participants, but the pattern of their endorsement(i.e., across prospective and inhibitory IU subscales) is also differ-ent. When comparing diagnostic groups within the clinical sample,there were no differences in the pattern of responding. It could beargued that the multi-group invariance analyses for the diagnos-tic groups within the clinical sample may not be robust because(1) even though the two-factor structure was superior to a uni-tary alternative, it was not robustly supported and (2) the samplesizes for many of the diagnostic groups were relatively small; how-ever, if there was a robust difference in the response patternsacross each diagnostic group, the somewhat poor group fit indicesand small sample sizes should have increased, not decreased, suchdifferences, particularly given how many groups were comparedsimultaneously. The fact that the diagnostic groups were invari-ant despite these concerns serves to underscore the comparability



of the response patterns across the diagnostic groups within theclinical sample.

The current results suggest a need to reevaluate the conceptthat IU may be specific primarily to worry and GAD. The rela-tionship between these variables is well-supported (Buhr & Dugas,2006, 2009; Sexton et al., 2003); however, rather than being a dis-positional construct that works to differentiate diagnoses of GADfrom other disorders (Dugas, Marchand, et al., 2005; Ladouceuret al., 1999; Sexton et al., 2003), there is growing evidence thatIU is an important component of several disorders (Boelen &Reijntjes, 2009; Buhr & Dugas, 2009; Carleton, Collimore, et al.,2010; Carleton, Sharpe, et al., 2007; Gentes & Ruscio, 2011; McEvoy& Mahoney, 2011; Norton & Mehta, 2007; Norton et al., 2005).Indeed, prior findings of IU being specific to GAD may well bethe result of the original IUS having GAD-specific items (Carleton,Gosselin, et al., 2010; Gentes & Ruscio, 2011).

The theoretical rationale for the importance of uncertainty inpsychopathology (Brown & Barlow, 2009; Carleton, Sharpe, et al.,2007; McEvoy & Mahoney, 2011), the implied necessity for uncer-tainty in anxiety (Suárez et al., 2009), along with evidence ofcomparable response patterns and endorsements of IU across anxi-ety disorders and depression, suggest IU may fit well within currenttransdiagnostic perspectives (Brown & Barlow, 2009; Norton &Philipp, 2008). Transdiagnostic approaches to psychological dis-orders suggest that they are facilitated or maintained by similarunderlying vulnerabilities that should be represented across diag-nostic categories. The most basic of underlying vulnerabilitiesshould be inherently noxious (Reiss & McNally, 1985; Taylor, 1993)and evolutionarily supported as threatening (Confer et al., 2010;Epstein, 1972), thus requiring no prior learning. The IU constructseems to meet the aforementioned criteria and uncertainty hasbeen theorized as a key distinguishing feature at least for anxiety;accordingly, there appears to be growing support for the notion thatIU may be a transdiagnostic vulnerability factor. Nevertheless, thereis a paucity of research examining the specific nature of IU withinanxiety-disorders and depression, and if individuals with these dis-orders experience greater intolerance of uncertainty in general, orspecific only to their disorder (e.g., to bodily sensations in panic).

There are empirically supported treatments, developed specif-ically for GAD-related worry that focus on IU (Koerner & Dugas,2006; Robichaud & Dugas, 2006). Given the seemingly ubiquitousnature of IU, incorporating IU-specific treatment components intotherapeutic protocols may result in pervasive benefits, and not onlyfor those with GAD or OCD, but for people with any anxiety disorderor with depression. There is already evidence that GAD symptomreporting shifts up or down based on increases and decreases inIU (Ladouceur et al., 2000); accordingly, additional research onthe application of IU-specific treatments (e.g., in vivo exposure touncertainty) across the anxiety disorders seems theoretically andclinically warranted.

The current investigation includes several limitations that war-rant consideration in interpretation of findings. These limitationsmay also provide directions for future research. First, the clinicalsample, while relatively large, included diagnostic groups that wererelatively less represented (e.g., MDD, which may limit the general-izability of these findings to such samples). The differences in sizesof diagnostic group were not prohibitive for most of the currentanalyses; however, the normative values and CFA fit indices asso-ciated with the smaller diagnostic groups may be less robust thanthose from larger diagnostic groups. Similarly, several diagnosticcategories were excluded due to sample size limitations (i.e., spe-cific phobia, ADNOS) or because the disorder was not treated atthe treatment center at which participants sought treatment (i.e.,PTSD). Recent research has provided evidence for the role of IU ineating disorders (Konstantellou, Campbell, Eisler, Simic, & Treasure,2011; Sternheim, Startup, & Schmidt, 2011), complicated grief, and

PTSD (Boelen, 2010); as such, future research should assess thesecategories as well.

Second, there was no interrater reliability information availablefor diagnosticians using the SCID-I; however, the diagnostic meth-ods, as described in Section 1, are in line with current best practiceand assessors received extensive training on the administrationof the SCID-I. Third, making comparisons based on the principaldiagnosis, while epidemiologically valid, may have masked dif-ferences between diagnostic groups that would only appear withlarger samples of people with only one diagnosis. In other words,it is possible that the presence of additional diagnoses (as summa-rized in Table 1) within each diagnostic group may have affectedresults (e.g., people with SAD may have higher IU scores than peoplewith panic because of higher rates of comorbid GAD). The prelim-inary attempt at analyzing “pure” diagnostic samples (i.e., thosewith only one diagnosis) presented in the current study involvedrelatively small samples and, while the results continue to indi-cate no substantive differences in IU between anxiety disordersand depression, larger “pure” samples or more dynamic assess-ments of comorbidity may yet produce different results; however,such a possibility is seemingly unlikely. Fourth, there is a singletaxometric investigation of latent worry (Olatunji, Broman-Fulks,Bergman, Green, & Zlomke, 2010) and a single taxometric inves-tigation of IU as measured by the IUS-12 (Carleton et al., 2012);however, the construct may yet benefit from further taxometricanalyses within individual diagnostic groups. Fourth, neither theundergraduate sample nor the community sample was diagnosti-cally assessed. As such, there may have been a disproportionatenumber of persons with clinically significant symptoms in both ofthose samples; however, if that were the case and of sufficient sizeto warrant concern, it would have increased the probability thatthe community sample endorsement rates would be statisticallycomparable to the clinical sample. Instead, the community sampleendorsement rates were significantly lower than the clinical sam-ple and indeed comparable to the undergraduate sample. As such,we suggest the presence of some persons with clinically signifi-cant anxiety in the undergraduate and community samples did notsubstantially impact the current results.

Fifth, determining a specific causal role of IU in the develop-ment of anxiety disorders and depression awaits a comprehensivelongitudinal investigation. Sixth, investigations of the manner inwhich IU relates to each of the disorders were not possible. Suchinvestigations may be particularly important given implicationsthat each subscale, prospective IU and inhibitory IU, may interactdifferently with each disorder. Seventh, despite indications of thecomparability of the IUS and the IUS-12, there remains a paucityof direct comparisons between the two measures (Khawaja & Yu,2010). Researchers have speculated that the IUS, like the new Intol-erance of Uncertainty Index, may better account for symptomsof worry (Gentes & Ruscio, 2011), whereas the IUS-12 focusesexclusively on measuring the core IU construct (Carleton, Gosselin,et al., 2010); however, such speculation remains to be empiricallyassessed. Eighth, the samples were primarily Caucasian, which maylimit the generalizability of the results; however, the evidenceto date suggests against differences based on ethnicity (Norton,2005).

Despite the limitations, the current study accomplishes sev-eral important incremental goals associated with IU research thatallow for conclusions that advance this area. First, there were nodifferences in endorsement patterns across a variety of diagnos-tic groups despite the use of stringent procedures and samplesizes that should have exacerbated differences; however, therewere differences in endorsement patterns between clinical andnonclinical samples. Second, the 2-factor structure of the IUS-12appears to be robust in clinical, undergraduate, and communitysamples, with no differences in endorsement rates or patterns for



men and women. Third, the results provide clinical norms for IUacross a variety of diagnostic groups. Fourth, persons with diag-noses of SAD, GAD, OCD, or MDD reported comparable levels of IUthat were significantly and substantially higher than undergrad-uate and community samples; however, the distribution of IU inPDA may differ from the other disorders, and warrants additionalresearch.

Overall, the results indicate that IU represents a construct thatmay be transdiagnostic across anxiety disorders and depression,and that these disorders share relatively similar levels and distribu-tions of IU. The results suggest against IU having a disorder-specificpattern of endorsement and suggest against IU being specific toGAD or OCD. The current results do not necessarily contrast withprevious research exploring differences in IU across disorders usingeither the IUS or the IUS-12. Findings using the IUS have, to date,been mixed and provide evidence of IU being associated with a vari-ety of symptoms and disorders (Boelen et al., 2010; Butzer & Kuiper,2006; Carleton, Sharpe, et al., 2007; Dugas, Marchand, et al., 2005;Norton & Mehta, 2007; Norton et al., 2005; Sexton et al., 2003;Yook et al., 2010). Previous results with the IUS-12 also indicateIU is broadly applicable to anxiety and possibly mood disorders(McEvoy & Mahoney, 2011). The substantial overlap in measure-ment and comparable psychometric substantiation for the IUS andIUS-12 (Carleton, Norton, et al., 2007; Carleton, Sharpe, et al., 2007)supports speculation that previous inconsistencies in IU across dis-orders may be the result of (1) IUS items that are serendipitouslyGAD-specific (e.g., “My mind can’t be relaxed if I don’t know whatwill happen tomorrow”) that have been excluded from the IUS-12in favour of measuring only the core IU construct, (2) using smallerclinical samples, or (3) the use of nonclinical samples. Irrespectiveof the reason, and until a similarly large-scale clinical comparison ofthe IUS and IUS-12 is conducted, the current study appears to pro-vide additional broad evidence supporting the importance of IU inseveral disorders. Indeed, IU may represent a ubiquitous vulnerabil-ity factor that fits well within current transdiagnostic perspectivesof psychopathology.

References

American Psychiatric Association. (2000). Diagnostic and statistical manual of mentaldisorders (4th ed., text revision ed.). Washington, DC: Author.

Asmundson, G. J. G., & Carleton, R. N. (2005). Fear of pain is elevated in adultswith co-occurring trauma-related stress and social anxiety symptoms. CognitiveBehaviour Therapy, 34, 248–255. doi:10.1080/16506070510011557

Berenbaum, H., Bredemeier, K., & Thompson, R. J. (2008). Intolerance of uncertainty:exploring its dimensionality and associations with need for cognitive closure,psychopathology, and personality. Journal of Anxiety Disorders, 22, 117–125.doi:10.1016/j.janxdis.2007.01.004, pii:S0887-6185(07)00019-9

Boelen, P. A. (2010). Intolerance of uncertainty and emotional distress follow-ing the death of a loved one. Anxiety, Stress, and Coping, 23, 471–478.doi:10.1080/10615800903494135

Boelen, P. A., & Reijntjes, A. (2009). Intolerance of uncertainty and social anxiety.Journal of Anxiety Disorders, 23, 130–135. doi:10.1016/j.janxdis.2008.04.007

Boelen, P. A., Vrinssen, I., & van Tulder, F. (2010). Intolerance of uncertainty in adoles-cents: correlations with worry, social anxiety, and depression. The Journal of Ner-vous and Mental Disease, 198, 194–200. doi:10.1097/NMD.0b013e3181d143de

Brown, T. A., & Barlow, D. H. (2009). A proposal for a dimensional classificationsystem based on the shared features of the DSM-IV anxiety and mood disor-ders: implications for assessment and treatment. Psychological Assessment, 21,256–271. doi:10.1037/a0016608, pii:2009-12887-003

Browne, M. W., & Cudeck, R. (1989). Single sample cross-validation indicesfor covariance structures. Multivariate Behavioral Research, 24, 445–455.doi:10.1207/s15327906mbr2404 4

Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In: K. A.Bollen, & J. S. Long (Eds.), Testing structural equation models. Newbury Park, CA:Sage Publications Inc.

Buhr, K., & Dugas, M. J. (2006). Investigating the construct validity of intolerance ofuncertainty and its unique relationship with worry. Journal of Anxiety Disorders,20, 222–236. doi:10.1016/j.janxdis.2004.12.004, pii:S0887-6185(05)00022-8

Buhr, K., & Dugas, M. J. (2009). The role of fear of anxiety and intolerance of uncer-tainty in worry: an experimental manipulation. Behaviour Research and Therapy,47, 215–223. doi:10.1016/j.brat.2008.12.004

Butzer, B., & Kuiper, N. A. (2006). Relationships between the frequency ofsocial comparisons and self-concept clarity, intolerance of uncertainty,

anxiety, and depression. Personality and Individual Differences, 41, 167–176.doi:10.1016/j.paid.2005.12.017

Byrne, B. (2001). Structural equation modeling with AMOS: basic concepts, applications,and programming. Mahwah, NJ: Erlbaum.

Byrne, B. (2004). Testing for multigroup invariance using AMOS graph-ics: a road less traveled. Structural Equation Modeling, 11, 272–300.doi:10.1207/s15328007sem1102 8

Carleton, R. N., Abrams, M. P., Asmundson, G. J. G., Antony, M. M., & McCabe, R. (2009).Pain-related anxiety across anxiety and depressive disorders. Journal of AnxietyDisorders, 23, 791–798.

Carleton, R. N., Collimore, K. C., & Asmundson, G. J. G. (2010). It’s not justthe judgements – it’s that I don’t know: intolerance of uncertainty asa predictor of social anxiety. Journal of Anxiety Disorders, 24, 189–195.doi:10.1016/j.janxdis.2009.10.007

Carleton, R. N., Gosselin, P., & Asmundson, G. J. G. (2010). The intolerance of uncer-tainty index: replication and extension with an English sample. PsychologicalAssessment, 22, 396–406. doi:10.1037/a0019230

Carleton, R. N., Norton, M. A., & Asmundson, G. J. G. (2007). Fearing the unknown: ashort version of the intolerance of uncertainty scale. Journal of Anxiety Disorders,21, 105–117. doi:10.1016/j.janxdis.2006.03.014, pii:S0887-6185(06)00051-X

Carleton, R. N., Sharpe, D., & Asmundson, G. J. G. (2007). Anxiety sensitiv-ity and intolerance of uncertainty: requisites of the fundamental fears?Behaviour Research and Therapy, 45, 2307–2316. doi:10.1016/j.brat.2007.04.006,pii:S0005-7967(07)00095-2

Carleton, R. N., Weeks, J. W., Howell, A. N., Asmundson, G. J. G., Antony, M. M., &McCabe, R. E. (2012). Assessing the latent structure of the intolerance of uncer-tainty construct: An initial taxometric analysis. Journal of Anxiety Disorders, 26,150–157.

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.).Mawah, NJ: Erlbaum.

Confer, J. C., Easton, J. A., Fleischman, D. S., Goetz, C. D., Lewis, D. M., Perilloux, C.,et al. (2010). Evolutionary psychology. Controversies, questions, prospects, andlimitations. The American Psychologist, 65, 110–126. doi:10.1037/a0018413

Curran, P. J., West, S. G., & Finch, J. F. (1996). The robustness of test statistics to non-normality and specification error in confirmatory factor analysis. PsychologicalMethods, 1, 16–29.

Davison, A. C., & Hinkley, D. V. (2006). Bootstrap methods and their application. Cam-bridge, UK: Cambridge University Press.

de Bruin, G. O., Rassin, E., & Muris, P. (2007). The prediction of worry in non-clinical individuals: the role of intolerance of uncertainty, meta-worry, andneuroticism. Journal of Psychopathology and Behavioral Assessment, 29, 93–100.doi:10.1007/s10862-006-9029-6

Dugas, M. J., Buhr, K., & Ladouceur, R. (2004). The role of intolerance of uncertainty inthe etiology and maintenance of generalized anxiety disorder. In: R. G. Heimberg,C. L. Turk, & D. S. Mennin (Eds.), Generalized anxiety disorder: advances in researchand practice (pp. 143–163). New York, NY: Guilford Press.

Dugas, M. J., Gosselin, P., & Landouceur, R. (2001). Intolerance of uncertainty andworry: investigating specificity in a nonclinical sample. Cognitive Therapy andResearch, 25, 551–558. doi:10.1023/A:1005553414688

Dugas, M. J., Hedayati, M., Karavidas, A., Buhr, K., Francis, K., & Phillips, N. A.(2005). Intolerance of uncertainty and information processing: evidence ofbiased recall and interpretations. Cognitive Therapy and Research, 29, 57–70.doi:10.1007/s10608-005-1648-9

Dugas, M. J., Marchand, A., & Ladouceur, R. (2005). Further validation ofa cognitive-behavioral model of generalized anxiety disorder: diagnos-tic and symptom specificity. Journal of Anxiety Disorders, 19, 329–343.doi:10.1016/j.janxdis.2004.02.002

Dugas, M. J., & Robichaud, M. (2007). Cognitive-behavioral treatment for generalizedanxiety disorder: from science to practice. New York, NY: Routledge.

Dugas, M. J., Savard, P., Gaudet, A., Turcotte, J., Laugesen, N., Robichaud,M., et al. (2007). Can the components of a cognitive model predict theseverity of generalized anxiety disorder? Behavior Therapy, 38, 169–178.doi:10.1016/j.beth.2006.07.002, pii:S0005-7894(06)00088-8

Epstein, S. (1972). The nature of anxiety with emphasis upon its relationship toexpectancy. In: C. D. Spielberger (Ed.), Anxiety: current trends in theory andresearch (pp. 291–337). New York, NY: Academic Press.

First, M., Spitzer, R., Gibbon, M., & Williams, J. (1996). Structured clinical interviewfor DSM-IV Axis I disorders – patient edition. New York, NY: New York StatePsychiatric Institute, Biometrics Research Department.

Freeston, M., Rhéaume, J., Letarte, H., Dugas, M. J., & Ladouceur, R. (1994).Why do people worry? Personality and Individual Differences, 17, 791–802.doi:10.1016/0191-8869(94)90048-5

Garber, J., & Hollon, S. D. (1991). What can specificity designs say aboutcausality in psychopathology research. Psychological Bulletin, 110, 129–136.doi:10.1037/0033-2909.110.1.129

Gentes, E. L., & Ruscio, A. M. (2011). A meta-analysis of the relation of intoleranceof uncertainty to symptoms of generalized anxiety disorder, major depres-sive disorder, and obsessive-compulsive disorder. Clinical Psychology Review, 31,923–933. doi:10.1016/j.cpr.2011.05.001

Gosselin, P., Ladouceur, R., Evers, A., Laverdiere, A., Routhier, S., & Tremblay-Picard,M. (2008). Evaluation of intolerance of uncertainty: development and valida-tion of a new self-report measure. Journal of Anxiety Disorders, 22, 1427–1439.doi:10.1016/j.janxdis.2008.02.005

Grenier, S., Barrette, A.-M., & Ladouceur, R. (2005). Intolerance of uncertainty andintolerance of ambiguity: similarities and differences. Personality and IndividualDifferences, 39, 593–600. doi:10.1016/j.paid.2005.02.014



Hock, M., & Krohne, H. W. (2004). Coping with threat and memory for ambiguousinformation: testing the repressive discontinuity hypothesis. Emotion, 4, 65–86.doi:10.1037/1528-3542.4.1.65

Holaway, R. M., Heimberg, R. G., & Coles, M. E. (2006). A comparison of intoler-ance of uncertainty in analogue obsessive-compulsive disorder and generalizedanxiety disorder. Journal of Anxiety Disorders, 20, 158–174. doi:10.1037/1528-3542.4.1.65

Hu, L., & Bentler, P. M. (1999). Fit indices in covariance structure modeling: sensi-tivity to underparameterized model mis-specification. Psychological Methods, 3,424–453.

Ingram, R. E. (2003). Origins of cognitive vulnerability to depression. Cognitive Ther-apy and Research, 27, 77–88. doi:10.1023/A:1022590730752

Judd, C. M., McClelland, G. H., & Culhane, S. E. (1995). Data analysis: continuing issuesin the everyday analysis of psychological data. Annual Review of Psychology, 46,433–465.

Khawaja, N. G., & Yu, L. N. H. (2010). A comparison of the 27-item and12-item intolerance of uncertainty scales. Clinical Psychologist, 14, 97–106.doi:10.1080/13284207.2010.502542

Koerner, N., & Dugas, M. J. (2006). A cognitive model of generalized anxiety disorder:the role of intolerance of uncertainty. In: G. C. Davey, & A. Wells (Eds.), Worry andits psychological disorders: theory, assessment and treatment. Chichester: Wiley.

Konstantellou, A., Campbell, M., Eisler, I., Simic, M., & Treasure, J. (2011). Testing acognitive model of generalized anxiety disorder in the eating disorders. Journalof Anxiety Disorders, 25, 864–869. doi:10.1016/j.janxdis.2011.04.005

Ladouceur, R., Dugas, M. J., Freeston, M. H., Rheaume, J., Blais, F., Boisvert, J. M., et al.(1999). Specificity of generalized anxiety disorder symptoms and processes.Behavior Therapy, 30, 191–207. doi:10.1016/S0005-7894(99)80003-3

Ladouceur, R., Gosselin, P., & Dugas, M. J. (2000). Experimental manipulation of intol-erance of uncertainty: a study of a theoretical model of worry. Behaviour Researchand Therapy, 38, 933–941. doi:10.1016/S0005-7967(99)00133-3

Lei, M., & Lomax, R. G. (2005). The effect of varying degrees of nonormal-ity in structural equation modeling. Structural Equation Modeling, 12, 1–27.doi:10.1207/s15328007sem1201 1

Lind, C., & Boschen, M. J. (2009). Intolerance of uncertainty mediates the relation-ship between responsibility beliefs and compulsive checking. Journal of AnxietyDisorders, 23, 1047–1052. doi:10.1016/j.janxdis.2009.07.005

McEvoy, P. M., & Mahoney, A. E. J. (2011). Achieving certainty about thestructure of intolerance of uncertainty in a treatment-seeking samplewith anxiety and depression. Journal of Anxiety Disorders, 25, 112–122.doi:10.1016/j.janxdis.2010.08.010

Miranda, R., Fontes, M., & Marroquin, B. (2008). Cognitive content-specificity infuture expectancies: role of hopelessness and intolerance of uncertainty indepression and gad symptoms. Behaviour Research and Therapy, 46, 1151–1159.doi:10.1016/j.brat.2008.05.009

Nevitt, J., & Hancock, G. R. (2001). Performance of bootstrapping approachesto model test statistics and parameter standard error estimation instructural equation modeling. Structural Equation Modeling, 8, 353–377.doi:10.1207/S15328007SEM0803 2

Norton, P. J. (2005). A psychometric analysis of the intolerance of uncertaintyscale among four racial groups. Journal of Anxiety Disorders, 19, 699–707.doi:10.1016/j.janxdis.2004.08.002

Norton, P. J., & Mehta, P. D. (2007). Hierarchical model of vulnerabili-ties for emotional disorders. Cognitive Behaviour Therapy, 36, 240–254.doi:10.1080/16506070701628065, pii:787623975

Norton, P. J., & Philipp, L. M. (2008). Transdiagnostic approaches to the treat-ment of anxiety disorders: a quantitative review. Psychotherapy, 45, 214–226.doi:10.1037/0033-3204.45.2.214

Norton, P. J., Sexton, K. A., Walker, J. R., & Norton, G. R. (2005). Hierarchical modelof vulnerabilities for anxiety: replication and extension with a clinical sample.Cognitive Behaviour Therapy, 34, 50–63. doi:10.1080/16506070410005401

Obsessive Compulsive Cognitions Working Group. (1997). Cognitive assessment ofobsessive-compulsive disorder. Behaviour Research and Therapy, 35, 667–681.doi:10.1016/S0005-7967(97)00017-X

Olatunji, B. O., Broman-Fulks, J. J., Bergman, S. M., Green, B. A., & Zlomke, K. R. (2010).A taxometric investigation of the latent structure of worry: dimensionality andassociations with depression, anxiety, and stress. Behavior Therapy, 41, 212–228.doi:10.1016/j.beth.2009.03.001, pii:S0005-7894(09)00076-8

Osborne, J. W. (Ed.). (2008). Best practices in quantitative methods. Los Angeles: SagePublications Inc.

Reiss, S., & McNally, R. J. (1985). The expectancy model of fear. In: S. Reiss, & R. R.Bootzin (Eds.), Theoretical issues in behaviour therapy (pp. 107–121). New York,NY: Academic Press.

Robichaud, M., & Dugas, M. J. (2006). A cognitive-behavioral treatment targetingintolerance of uncertainty. In: G. C. Davey, & A. Wells (Eds.), Worry and its psy-chological disorders: theory, assessment and treatment (pp. 289–304). Chichester:Wiley.

Robichaud, M., Dugas, M. J., & Conway, M. (2003). Gender differences in worryand associated cognitive-behavioral variables. Journal of Anxiety Disorders, 17,501–516. doi:10.1016/S0887-6185(02)00237-2

Salgado-Ugarte, I. H., & Perez-Hernandez, M. A. (2003). Exploring the use of variablebandwidth kernel density estimators. Stata Journal, 3, 133–147.

Salgado-Ugarte, I. H., Shimizu, M., & Taniuchi, T. (1994). Exploring the shape ofunivariate data using kernel density estimators. Stata Technical Bulletin, 3,8–19.

Scott, D. W. (1979). On optimal and data-based histograms. Biometrika Trust, 66,605–610. doi:10.1093/biomet/66.3.605

Sexton, K. A., & Dugas, M. J. (2009). Defining distinct negative beliefs about uncer-tainty: validating the factor structure of the intolerance of uncertainty scale.Psychological Assessment, 21, 176–186. doi:10.1037/a0015827

Sexton, K. A., Norton, P. J., Walker, J. R., & Norton, G. R. (2003). Hierarchical model ofgeneralized and specific vulnerabilities in anxiety. Cognitive Behaviour Therapy,32, 82–94. doi:10.1080/16506070302321

Starcevic, V., & Berle, D. (2006). Cognitive specificity of anxiety disorders:a review of selected key constructs. Depression and Anxiety, 23, 51–61.doi:10.1002/da.20145

Sternheim, L., Startup, H., & Schmidt, U. (2011). An experimental explorationof behavioral and cognitive-emotional aspects of intolerance of uncer-tainty in eating disorder patients. Journal of Anxiety Disorders, 25, 806–812.doi:10.1016/j.janxdis.2011.03.020

Suárez, L., Bennett, S., Goldstein, C., & Barlow, D. H. (2009). Understanding anxietydisorders from a triple vulnerability framework. In: M. M. Antony, & M. B. Stein(Eds.), Oxford handbook of anxiety and related disorders (pp. 153–172). New York,NY: Oxford University Press.

Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston,MA: Allyn and Bacon.

Taylor, S. (1993). The structure of fundamental fears. Journal of Behavior Therapy andExperimental Psychiatry, 24, 289–299. doi:10.1016/0005-7916(93)90053-Y

Tolin, D. F., Abramowitz, J. S., Brigidi, B. D., & Foa, E. B. (2003). Intolerance ofuncertainty in obsessive-compulsive disorder. Journal of Anxiety Disorders, 17,233–242. doi:10.1016/S0887-6185(02)00182-2

van der Heiden, C., Melchior, K., Muris, P., Bouwmeester, S., Bos, A. E., &van der Molen, H. T. (2010). A hierarchical model for the relationshipsbetween general and specific vulnerability factors and symptom levelsof generalized anxiety disorder. Journal of Anxiety Disorders, 24, 284–289.doi:10.1016/j.janxdis.2009.12.005

Wilamowska, Z. A., Thompson-Hollands, J., Fairholme, C. P., Ellard, K. K., Farchione,T. J., & Barlow, D. H. (2010). Conceptual background, development, and prelimi-nary data from the unified protocol for transdiagnostic treatment of emotionaldisorders. Depression and Anxiety, 27, 882–890. doi:10.1002/da.20735

Yook, K., Kim, K. H., Suh, S. Y., & Lee, K. S. (2010). Intolerance of uncertainty, worry,and rumination in major depressive disorder and generalized anxiety disorder.Journal of Anxiety Disorders, 24, 623–628. doi:10.1016/j.janxdis.2010.04.003

Zanarini, M. C., & Frankenburg, F. R. (2001). Attainment and maintenance ofreliability of Axis I and II disorders over the course of a longitudinalstudy. Comprehensive Psychiatry, 42, 369–374. doi:10.1053/comp.2001.24556,pii:S0010-440X(01)63326-1

carleton mulvogue thibodeau mccabe antony asmundson 2012

Documents

social anxiety disorder

obsessivecompulsive

anxiety treatment

people withpanic disorder

exception of panic disorder

diagnostic groups

distribution of iu

depression clinical