Gender in Worry

7/28/2019 Gender in Worry

1/99

Examination of the Relationship between Intolerance of Uncertainty and Worry

Kristin E. M. Buhr

A ThesisIn

The DepartmentOf

Psychology

Presented in Partial FuifilIrnent of the RequirementsFor the Degree of Masters of Arts atConcordia UniversityMontreal, Quebec, Canada

O Kristin E.M. Buhr, 2001


2/99

NationalLi'braiy BibliohquenationaleduCanadauisitiansand9 Acquisitions etBib iographicServiees services bibliographiques

The author has granted anon- L'auteur a accord une licence nonexclusive licence alowing the excIusivepermettant laNational L i i fCanaa to Bibliothque nationale duCanada dereprod~ce~oan, distniute or seJi reproduire, prter, distniuer oucopies of this thesis in microform, vendre des copies de cette thse souspaper or electronic fomats. Ia forme de microfiche/fk,dereproduction sur papier ou sur formatIecronique.The auhor retains ownersiup of the L'auteur conserve lapropriet ducopyright inthis thesis. Neither the droit d'auteur quiprotge cette th&.thesisnor substantialextracts h m t Nik hse ni des extraits substantielsmay be prhted or otherwise de celle-cine doivent tre imprimsreproducedwithout the author's ouautrement reproduits sanssonpermission. autorisation.


3/99

AbsractExamination of the Relationship between htolerance of Uncertainty and Worry

Kristin E. M. BuhrThe present paper consists of two studies intended to further the understanding ofintolerance of uncertainty and its relationship to worry. The fmt study examineci thepsychornetric properties of the English version of the intolerance of Uncertainty ScaIe(WS),hich hasalready been vdidated in French. Factor analysis indicated that the IUShas a Cfac tor structure tha represents the idea that uncertainty is stressful and upsetting,uncertainty Ieads to the inability to act, uncertain events are negative and should beavoided, and king uncertain is not fair. The IUS has excellent interna1 consistency, goodtest-retest reliability, and convergent and divergent validity when assessed with symptommeasures of worry, depression, and anxiety. The second study attempted to assess theunique relationship ktween intoleranceof uncertainty and worry, beyond constructsalready associated with worry such as perfectionism and control. Fwthermore, the studyassessed the distinction between intolerance of uncertainty and intolerance of ambiguity.The resuIts suggest that worry has a stronger relationship with intolerance of uncertaintythan perfectionism, controI, and intolerance of ambiguity. Moreover, the resuits indicatethat ntolerance of uncertainty and intolerance of ambiguity are distinct co ns m cs .Overall, this study suggests that the iS is a sound measure of intolerance of uncertaintyand supports the idea that intolerance of uncertainty is an important co ns mct involved inworry.


4/99

Ac know ledgments1would k e o thankmy supervisor,Dr. Michel Dugas, for his direction, support,

and encouragement. 1would also ke to thank my committeemembers AlexSchwartzman and NataIie Phillips. 1would like to extend my appreciation to themembers of the Dugas Lab for their support on rnany different Ievels. Thanks to MelisaRobichaud, Nina Laugesen, Naomi Koerner, Kylie Francis,Mary Hedayati, and AngieKaravidas. Thanks to my classrnatesand friends, for providing me with encouragementdong the way. And a special thanks to Sarah Auchterlonie for ail her help.

1am also extremely grateful to my parents, Ken and Lou-Anne Buhr, for theirlove and support. 1have accomptished this because of you. 1would also like to thankDan Gendreau and Zo or their endless support and for making this journey with me.

This research was fundedby a grant from the Centre for Research in HumanDevelopment(CRDH).


5/99

Table of Contents

................................................................................ist of Tables viList of Appendices ..................................................................... vii

........................................................................enerai Introduction 1Smdy One ............................................................................... 10

Method ........................................................................ 13Results ...................................................................... 16Discussion ........................................................................ 27

Smdy Two .................................................................................30Method ........................................................................ 31Results ................................................................... 34Discussion .......................................................................40

General Discussion ........................................................................ 48References ................................................................................. 52


6/99

viList of Tables

......................able 1:Means andStandard Deviations for aU Study Measures 19.......able 2: Means. StandardDeviations. and Item-Total Correlations of the IS 20

Table 3: Ebmax-Rotated Iterated-Principal-Factor Standardized Regression................oefficients and Final Cornmunality Estimates (hz) f theIS 22

.......................................able 4: Correlations between Factors on the IIJS 24.......................able 5: Correlations among Study Measures. Gender. and Age 25

Table 6: Summary of Hierarchicai Multiple Regression Analysis for Variables.............................................redicting Scores on the PSWQ 2 6..........................able 7: Means and Standard Deviations for Study Measures 37

.......................able 8: Correlations among Study Measures. Gender. and Age 38Table 9: Summary of Hieratchical Multiple Regression Analysis for Variables

....................................................redicting Scores on the PSWQ 39


7/99

List of Appendices...................................................ppendix A: ConsentForm Study One 62...................................................ppendix B: Consent Form Study Two 64.................................................ppendix C: Test Re-Test ConsentForm 66...................................................ppendix D: General InformationForm 68

............................................ppendix E: holerance of Uncertainty Scale 70

............................................ppendix F:Penn State Worry Questionnaire 73..........................................ppendix G: Worry and Anxiety Questionnaire 75

..............................................ppendUrH-Beck-Depression Inventory-II 78.......................................................ppendix 1; Beck Anxiety Inventory 82.....................................ppendix J: Multidimensional Perfectionism ScaIe 84

.......................................................ppendix K: Sense of Control Scde 88.............................ppendix L:Scale of Toleranca-Intoleranceof Arnbiguity 90


8/99

Examination of the Relationship between Intoletance of Uncertainty and WorryInterest in the area of worry is rising and this h a ken reflected in the increase in

research examining bothworry and worry reIaied phenomena (e.g., Freeston, Rhaume,Letarte, Dugas, & Ladouceur, 1994;Mathews, 1993;Tallis& Eysenck, 1994). Worrycan be defined as concern about future events in which there is uncertainty about theoutcome and where the individual experiences feelings of anxiety (see MacLeod,Williams,& Bekerian, 1991). Worry is comrnon in both clinicai and nonclinicalpopulations and research has suggested that as high as 38% of individuais in the generaipopulation worry at Ieast once a day (Tallis, Davey, & Capuzzo, 1994). Further,excessive and uncontrollable worry is the central feature of generaiized anxiety disorderorGAD (DSM-IV; Arnerican Psychiatrie Association, 1994). Given the level of worry inthe general population and the role of excessive worry in the clinical disorder,GAD, t isimportant to identify key constructs related to worry in order to begin to establish howexcessive worry develops and what factors are responsible for maintaining it.

Research into the area of wony has generally focused on worry themes and howmuch time is spent worrying (e-g., Davey, 1993;Dugas, Freeston, Doucet, Lachance, &Ladouceur, 1995). However, attention has shified towards the examination of specificconstnicts related to w o q (e-g., Freeston et al., 1994; Russell& Davey, 1993). Forinstance. researchers have begun to identifyconstructs that may be involved in thedevetopment and maintenance of worry mgas, Gagnon, Ladouceur,& Freeston, 1998;Wells&Carter, 1999). Research has demonstrated tbat the tendency to worry is relatedto positive beliefs about the function of worry, the tendency to avoid upsetting mentalimagery, negative problem orientation, and intolerance of uncertainty (Dugas et ai.,


9/99


10/99

3more dffculties when faced with uncertain situations given that they tend to interprethem in a negative way. Overall, the findings indicate that wocriers have difficultytolerating uncertainty, which provides the initial evidence for a specific consuct relatedto worry: intolerance of uncertainty.

Recently, a number of studies have specifically Iinked intolerance of uncertaintyto worry and have suggested that it may be one of the most significant factors invohed inworry (Dugas et al., 1997; Ladouceur, Talbot, & Dugas, 1997). Studies havedemonstrated that intolerance of uncertainty and worry are highly related and that hisrelationship is not the result of shared variance with anxiety and depression (Dugas et al.,1997; Freeston et al., 1994). Given that anxious and depressive symptornsaresignificantly related to worry (Brown, Antony, & Barlow, 1991), these fmdings point tothe important role intolerance of uncertainty may play in worry.

Furthemore, research has established intolerance of uncertainty as the mostsaiient predictor of worry above positive beliefs about worry, negative problemorientation, and cognitive avoidance (Laugesen& Dugas, 2000, Robichaud& Dugas,2000). These findings provide further support for the strong relationship betweenintokrance of uncertainty and worry, given that previous research has suggested thatworry is highly related to beliefs about worry (Davey, Tallis,& Cappuno, 1996;Wells& Carter, 1999), problem orientation (Ladouceur, Blais, Fteeston, & Dugas, 19981,andcognitive avoidance (Butler, Wells, & Dewick, 1995).

Recent snidies have aiso begun to assesswhether intolerance of uncertainty isspecific to worry or whether it is a cognitive process involved in a number of emotiondor anxiety related phenomena. Dugas and coUeagues(2001)assessed the relationship


11/99

between intolerance of uncertainty, worry, obsessions/compulsions, and panic sensations.The results showed thai, in a nonclinical sample, intolerance of uncertainty is highlyrelated to worry, moderately related to obsessions/compulsions, and weakly related topanic sensations. In addition, research examining generalized anxiety disorder (GAD),where the cardinal feature is excessive worry, has idenWied that level of intolerance ofuncertainty distinguishes GAD patients fiom individuais suffering from other anxietydisorders (Ladouceur et ai., 1999).This research supplies initial support for the idea thatintolerance of uncenainty appears to have a stronger relationship with worry than othermanifestationsof anxiety.

Based on the strength of the relationship between intolerance of uncertainty andworry, researchers are now examining the possible causal role of intolerance ofuncertainty in worry. Studies have shown that targeting intolerance of uncertainty in thetreatment of excessive worry Ieads to changes in level of worry (Dugas & Ladouceur,2000; Ladouceur et ai., 2000). Moreover, changes in intolerance of uncertainty generallyprecede changes in worry, over the course of treatment (Dugas et ai., 1998). A recentlaboratory study has aIso demonstrated that manipulating an individuai's level ofintolerance of uncenainty resulted in changes in their levei of worrisome thoughts(Ladouceur, Gosselin,& Dugas, 2). According to Kraemer and associates (1997),establishing that changes in intolerance of uncertainty precede changes in worry anddemonstrating that experirnentaliy manipulating intolerance of uncertainty results inchanges in wotry, suggest that intolerance of uncertainty may bea causal risk factor forworry. AIthough more fesearch is needed to c o n f i hese initiai findings, the results


12/99

5point to the role intolerance of uncertainty may play in the development and maintenanceof wony.

Given the strong relationship between intolerance of uncertainty and worry andthe effect that intolerance of uncettainty has on worry, it is important to consider whetherthey aredistinct consucts. Worry bas been commonly defmed asconcem aboutnegative future events in which there is uncettainty sunounding the outcome and wherethe individuai experiences feelings of anxiety (MacLeod et al., 199 1). Althoughuncertainty is one aspect of worry, intoIeranct of uncertainty is the overall tendency of anindividual to find it unacceptable that a negative event might occur, however srnall chatprobability. Worry might best be viewed as a mental act where the individual thinksabout the situation and possible outcomes. Whereas intolerance of uncertainty can beseen as a through which individuah view their environment, which might be bestdescribed as a predisposition to find uncertainty unacceptable. If an individuai fin&uncertainty unacceptable, when faced with uncertainty they may engage in excessivewonying. In this sense, worry may be seen as a product of intolerance of uncertainty.

One way to examine the distinction between wony and intolerance of uncertaintyis to investigate their relationship with other factors. For example, Ladouceur andcolleagues (1997) found that although intoletance of uncertainty and worry were highlyrelated, they displayed different patterns of correlations with specific behavioral tasks,The researchers required participants to make decisions that varied on level of ambiguityand diffkulty. The results indicatedthat worry was not correlated with performanceonany of the behavioral tasks regardles of the amount of ambiguity or Ievel of difficulty.


13/99

Aiternatively, intolerance of uncertainty was correlated wih performance on moderatelyambiguous tasks.

Another factor that may help differentiate behveen worry and intolerance ofuncertainty can be found in the examination of possible gender differences on theseconstructs. Researchers have consistently identified gender differences on measures ofworry with women reporting higher levels of worry. However, gender differences havenot been found for intoIerance of uncertainty (Freeston et ai., 1994;Robichaud & Dugas,2000). The differentiating patterns of correlations for worry and intolerance ofuncertainty and the gender differences these constructs display, support the notion thatalthough intolerance of uncertainty and worry are related, they are in fact differentconstructs.

Aithough the research demonstrating the rdationship between intolerance ofuncertainty and worry is beginning to accumulate, research stiII needs to compare thecontributions of intolerance of uncertainty to w o q against other measureschat havealready been establishedas factors related to worry. This step is necessary for assessingwhether the contributions of intolerance of uncertainty to worry are not better explainedby other factors. If intoIerance of uncertainty does not add anything unique to theunderstanding of worry, then the focus on intolerance of uncertainty should be shifted tofactors that play a more prevaient role in worry.

There are a number of factors that have been linked to excessive worry. Forexample, researchen have suggested that personaiity traits such as perfectionismarerelated to anxiety and worry (Pratt,Tallis,& Eysenck, 1997). As stated previousIy,tesearchers have postulatedthat worry may be related to elevated evidence requirements


14/99

(Talliset al., 1991). The need for additional information may be related to the attempt tofind the "perfect" solution. individuaIs with perfectionist personality styles are IikeIy toexperience anxiety or wony when attempting to discover perfect solutions or outcomes,given that such outcomes are rare. Research has supponed this link i d emonstratedthat worry is significantly related to perfectionism (Kawamura, Hunt, Frost,& BiBartolo,2001;Stoeber& Joormann, 2001).

Research has aiso shown that worry and anxiety may be related to specificdimensions of perfectionism. Self-oriented perfectionism, or the tendency to placespecific demands and expectations on oneself, is associated with adjustment problemsincluding anxiety (Flen, Hewitt,&Dyck, 19891. Other research has linked anxiety withboth self-oriented perfectionism and midly-prescribed perfectionism, which is theattempt to meet the expectations of others (Flett & Hewitt, 199 1). This indicates thatindividuais who place high demands on themselves or feel that others have placed suchdemands on them are likely to feei anxious and worried about meeting those demands.More ment findings have shown that worry is retated to sociaily-prescribedperfectionism or the need to meet the expectations of others (Flen, Hewin, Endler,&Tassone, 1995). Again, when an individual feels pressure to live up to certain standardsthis can generate high levels of worry. -41thongh previous research has not found a clearIink between worry or anxiety and other-oriented perfectionism, or the tendency to placehigh standards on others, furtherdata is stilI needed.

Worry has alo been linked to perived conml (Davey, 1994). Morespecificaily, worry hasbeen shown to be relaieci to a lack of perceived control overproblern solving. This suggests that womers be1ieve that they have no control over the


15/99

problem-solving process. A perceived lack of personal control has been stiown to bestmngly related to worry (Sebb& Beck, 1998)- Moreover, the lack of personal controlhada stronger relationship with worcy than to somatic anxiety. It is Iikely that a Iack ofperceived control would result in heigbtened levelsof worry. if someone believes thatthey have no control over what is happening this could increase their concerns andworries over the situation.

Although intoIerance of uncertainty is a relatively new constnrct, the concept ofintolerance towards ambiguity is not new. Part of establishing the role of intolerance ofuncertainty in worry will also require researchers to demonstrate a distinction betweenthe newer constnict, intoleranceof uncertainry, and the concept of tolerance ofarnbiguity. This c m be a dificuit task given that "uncertainty"and "ambiguity"appear tosharemany features.

Tolerance or intolerance for ambiguity is an idea that has generated interest formaay years. In fact, research from as early as the late 1940s examined the relationshipbetween intolerance of ambiguity and authoritarian prsonaiity styles (see FrenkeI-Bmnswik, 1948, 1949). This early research defined the consmct as %e tendency toperceive ambiguous situations as sources of threat" [Budner, 1962). Ambiguoussituationswere thought to represent novel, cornplex, or insoluble situations. Althoughinitial examination of the definition suggests that it is compatiite with the definition ofintoleranceof uncertainty,which States that the individual fin& the possibiiity of anegative outcome occuningunacceptable, they are different to somedegree.

For instance, Fucnham (1944) reviewed a number of measures assessingintolerance of ambiguityand his ovemew of theexistingmasures of that consmct


16/99

suggest that it is a much broder concept In fact, his review suggested that the mostconmon measures of intolerance of ambiguity contain underiying factors that includesuch things as conservative points of view, anxiety induced from ambiguity,adventurousness, variety, originality, clarity, and regularity. in addition, the specificfactors identified Vary from measure to measure. Given the broad areas assessed bymesures of intolerance of ambiguity, it is not swprising to find that these mesures wereused to examine a broad range of concepts that included religious beliefs, attitudestowards censorship, career choices, rigidity,conservatism, and hostility (see Furnharn,1994fora review). Researchers examining intolerance of uncertainty believe that thisconstruct is assessing sornething quite specific and varies greatly from the originaIconcept of intolerance of ambiguity (Freeston et al., 1994). It wiII be important toestablish intolerance of uncertainty as a specific construct related to worry and to separateit from the traditional broad concept of intolerance of ambiguity.

Until recently, the research focusing on intolerance of uncertainty has beencamied out exclusively in French-speaking populations, using a French measure ofintolerance of uncertainty. In order to assess the relationship between intolerance ofuncertainty and worry in English populations, an English version of the intolerance ofUncertainty Scale needs to be developed and validated. Furthemore, the concept ofintolerance of uncertainty is still fairly new and additional research is needed to betterdeiineate its relationship to worry. The present study consists of two separate studies thatwiiI attempt to m e r he understanding of intoIerance of uncertainty. The fmt snidywiii examine the psychomehic properties of an English translation of the Intolerance ofUncertainty Scale (IUS) in order to estabiish its reiiabiiity and vaiidity. The econd study


17/99

10

will attempt to ascertain whether intolerance of uncertainty adds to our understanding ofwony beyond what is explained by other constructs associated with worry such asperfectionism and control. Finally, the study will attempt to distinguish intolerance ofuncertainty from the broader concept of intolerance of arnbiguity, which has been used tostudy very different concepts.

StudyOneThe original French version of the Intolerance of Uncertainty Scde(US) as

developed to assess emotional, cognitive, and behavioral reactions to ambiguoussituations, implications of being uncertain, and attempts to control the future (Freeston etai., 1994). Items on the WS were devised from a pool of 74 statements that weregenerated to reflect different aspects of intolerance of uncertainty such as theconsequences of king uncertain, how uncertainty reflects on a person, expectationsabout the predictability of the future,atternpts to control the future, frustration arounduncertainty, and "all-or-nothing responses" to uncertainty. Items were assessed on facevalidity by four judges and items thaiwere deemed irrelevant or redundant werediscarded.

The remaining44 items were administered to a group of 110university students.The students were divided into three p u p s depending on whether they metGADdiagnostic criteria based on their responses to the Generalized Anxiety DisorderQuestionnaire - Modified version (GADQ-M;Roemer, Posa,& Borkovec, 1991). Thethree groups included those meeting the criteria forGAD by questionnaire, those meetingoniy the somaticMiteriaforGAD by questionnaire, and fmally those participants whomet neither the full nor somatic criteria forGAD. Statistical analysis was used to


18/99

11

identify the items that correctly distinguished between these three groups, Twenty-threeitems met this requirement and an additionai4 items were kept because of their highcorrelation with the Penn Sate Worry Questionnaire(PSWQ; eyer, Miller, Metzger&Borkovec, 1990), which is a general measure of the tendency to worry. The final 27items on the IUS reflect the idea that uncertainty is unacceptable, reflects badly on aperson, and leads COfrustration, stress, and the inability to take action.

The original study (Freeston et al., 1994) examined the psychometric properties ofthe French version of the lCTS and examined the relationship between intolerance ofuncertainty and memres of worry, anxiety, and depression. Factor anaiysis identified a5-factor solution that included: beliefs that uncertainty is unacceptable and should beavoided, king uncertain reflects badly on a person, uncertainty results in stress,fmstration, and prevents action. The internai consistency of the scale was excellent(a=-91) and its test-retest reliabiiity over a five-week p e n d was good (L= -78; test-retestfromDugas et ai., 1997). The scaiewas able to differentiatebetween groups of high andIow womers in a nonclinicai sarnple, demonstrating criterion-related validity. Further,the iS was highly correlated to measures of worry and to a Iesserextent with measuresof anxiety and depression, which supports the rneasure's convergent and divergentvaiidity. in addition, once the shared variance of depressed and anxious symptoms waspartiaied out, the retationship to worry remained strong, suggesting that intolerance ofuncertainty is specificdIy related to wony.

The IntoIerance of Uncertainty Scde (IUS)was translated from French to Englishusing aweU estabiished method (see Vailerand, 1989). Two ndependent translatowtranslated the i S into English. It was back translatedby another independent translater,


19/99


20/99

Metbodctuq~& Two hundred and seveniysix =276) participants were recruited

through various undergraduate courses. There were 213 femaie participants and 62males. Information regarding gender was missing for one participant. Themean age ofparticipants was 22.6 (m=5.05). Students were invited to participate at the start of aregular undergraduate course and participation was voluntary.

instruments, The participants completed the foIIowing questionnaires in randomorder: the Intolerance of Uncertainty Scde (IUS), the Penn State Worry Questionnaire(PSWQ),he Worry and Anxiety Questionnaire (WAQ), the Beck Depression InventoryII (BDI-II), and the Beck Anxiety Inventory (BAI). In addition, participants were askedto complete a demographic fom.

The Intolerance of Uncertainty Scde (iS: Freeston et al., 1994) includes 27items reIating to the idea that uncertainty is unacceptable, reflects badly on a person, andleads to frustration, stress, and the inability to take action. Participants rate items on a 5-

. .oint Likert scale ranging from 1 ="not at al1 chiuact~strcf me' to 5= "gntirely. .haractensticof me". Examples of items include "Uncertainty makes me uneasy,

anxious, or stressed and "My mind can't be relaxed if 1dont know what will happentomorrow". The French version of the measure has excellent intemal consistency (a =-9I), good test-retest reliability over a five week period = 78) and demonstratedconvergent and discriminant validity (test-retest fromDugas et ai., 1997; Freeston et al.,1994).

The Penn State Wory Ouestio- (PSWQ:Meyer et a[., 1990) consists of 16items that meamre the tendency to engage in excessive, uncontrollable, and generalized


21/99

14worry. Participants rate items on a 5-point Likert s a l e ranging from 1= "Dtur>ical" to5= "veryW. xamples of items inciude 'My womes overwhelm me"and "Once1start worrying, 1can't stop". The questionnaire has demonstrated reliabilityand validity (Brown et ai., 1992; Davey, 1993; Meyer et al., 1990). The PSWQ s aunifactorial measure with excellent intemal consistency(a .86 to .95) and test-retestreliability (r= .74 to .93; Molina& Borkovec. 1994). The questionnaire has good knowngroups vaiidity and substantial convergent and divergent vaiidity dernonstrating greatercorrelationswith measures of worry than anxiety and depression (Molina& Borkovec,1994).

The Worrv and Anxiety Questionnaire(WAQ: Dugas, Freeston, Provencher,Lachance, Ladouceur,& Gosselin,2001) contains 11 items that cover worry themes andDSM-IV diagnostic criteria forGAD. It examines both the cognitive criteria, such asexcessive worry, and the somatic cntena, which includes physiological symptoms suchas muscle tension. The WAQ can be used to identify whether individuais do not meet thecriteria for GAD, meet onIy the somatic criteria for GAD, or meet al1 of the criteria forGAD, which can be referred to as GAD by questionnaire. Previous research hasdemonstrated that individuals tend to fa11 into those three categories and seldom meetonly the cognitive criteria (Freeston et al., 1994). TheWAQ hows good test-retestreiiability afteraCweek periodO= 76; Beaudoin et ai., 1997)and excellent criterion-related validity for discriminating between GAD patients and matched controIs (Dugas etai., 2001).

The Beck Depression InventoryB BDI-II: Beck, Steer,& Brown, 1996) is a 2 1-item self-reportquestionnaire, each item reflecting depressive symptoms. Participants


22/99

15indicate whether ikms are characteristic of how they have been feeling during the past 2weeks. Examples of themes covered by the BDI-II include: sadness, pessimism, loss ofinterest, suicidal thoughts, sleeping problems, and agitation. The m asure hasexceptional intemal consistency in a coliege sample (a = .92) and excellent test-retestreliability over a one-week period for an outpatient sample (t=.93;Beck et al., 1996).In addition, the measure has demonstrated convergent and divergent vaiidity (see Beck etai., 1996; Steer & Clark, 1997). Cornparisons with the original version of the BDI (Beck,Ward, Mendelson, Mock, & Erbaugh, 1961) suggest that the BDI-IZ is strongly correlatedwith the original version (r= 93; Beck et ai., 1996) but has a stronger factor structure(Dozois, Dobson, & Ahnberg, 1998).

The Beck W e t y nventory @Ai: Beck, Epstein, Brown,& Steer, 1988) is a 21-item measure that examines state anxiety with each item correspondhg to commonanxiety symptoms. Participants rate each item, according to how often the symptornshave bothered them in the previous week, on a 4-point Likert scaIe ranging from O ="maltadl" to 3="m.xamples of symptoms assessed by the BAI include: feeling hot,nervous, shaky, scarexi, faint, and flushed. The BAI has excellent intemal consistency(a = .92), high test-retest reliability over a I-week period (I=-75). nd demonstratedconvergent and divergent validity in an outpatient sampIe (see Becket ai., 1988).Creamer, Foran, and Be11 (1995) have established the reliability and vaiidity for thismeasure in a nonciinical sample.

Procedu= Participants were asked to complete the five questionnaires andsupply demographic information. Thequestionnaires were compietea during one 30-minute testing period and groups of participants were nuion several separate occasions.


23/99

16Participants were told that thepurpose of the expriment was to assess the relaionshipbetween worry and other emotional responses such as anxiety and depression.Participants were informed that they could discontinue the study at any time. In addition,a group of 66 participants that were previously tested were asked to complete theIntolerance of Uncertainty Scale (IUS) for a iweek retest of the measure.Results . .Overview of Statistical Analvsis. To examine the reliability of theWS,coefficient alpha was used in conjunction with an item analysis. Furthemore, therneasure was assessed for test-retest retiability atera five-week interval, using acorrelation between initial IUS scores and subsequent scores.

To evaluate the factor structure of the IUS, he Kaiser (1970) measure of sarnplingadequacy was employed to determine whether the data was appropriate for factoranalysis. Following this initiai test, principal components extraction and Cartell's screetest (1966) were used to determine the number of appropriate factors. Principal factorsextraction with Promax (oblique) cotation, which takes into account the correlationbetween factors, was performed and the final rotated factors were assessed for internaiconsistency.

A correlation rnatrix was used to determine the relationships between studymeasures and partial correlations were utilized to examine the unique relationshipbetween intolerance of uncertainty and worry, once variance shared with anxiety anddepression was removed. Moreover, a hierarchical regression was performed to assessthe predicted variance of worry(PSWQ),y entering demographic information (age and


24/99

17

gender) in the first step, foiiowed by measures of anxiety (BAI)and depression (BDI-II),and finay the measure of intolerance of uncertainty (IUS).

Finally, a one-way between groups Analysis of Variance (ANOVA)wasperforrned using intolerance of uncertainty scores. Individuais were grouped accordingto their responses on the WAQ. This anaiysis was used to test the final hypothesis thatthe IUSwould be able to distinguish between groups of participants who met the fulldiagnostic criteria for GAD, those who met only the somatic criteria, and those who metnone (neither the cognitive nor the somatic) criteria for GAD.

. .1 Data Prior to any specific staistical analysis the data werescreened to determine whether statistical assumptions were met and to ascertain if thedata was appropriate for further statistical analysis (see Tabachnick and Fidell, 1996 for areview of data screening procedures). The data, excluding demographic information,were transfonned into z-scores to evduate the presence of extreme scores that were morethan 3.29 standard deviations fiom the mean in either direction. Seven participantswereidentified as univariate outliers for having extreme scores on study measures and wereremoved fiom further analysis. Mutlivaiate outliers were assessedby examiningMahalanobis distance and Cook's distance. For this analysis,al1 study measures,excluding demographic information, were included and thePSWQwas identified as thedependent variable. The resulting Mahalanobis distance for each participant wascompared against a criticalX' value. Two participants were identified as muItivariateoutiiers due to scores exceeding this criticai value. However, neiiher of these casesproducedaCook's distance that was greater than the criterion of 1, suggestingha t thecases were not significantly effecting the regression; therefore, they were not deleted.


25/99

18Thedatawere aiso assessecifor normality by exarnining the skewness and

kurtosis of the distribution for each measure. Only the htoleraoce of Uncertainty Scale(IUS)was identifiedas being significantly skewed; therefore, it was transformed usinglogarithms. This process resulted in a normal distribution on theIUS.The assumption oflinearity and homoscedasticity was verified through the examination of the bivariatescatterplots between PSWQ nd al1 other measures (IUS,BAI,BDI-II,ndWAQ). Theassumption of linearity was considered to be violated if a nonlinear relationship wasfound. In addition, the assumption of homoscedaticity was met if the pattern on thescatterplot suggested variance was normally distributed. However, the BAI and BDI-IIappeared to violate these assumptions and the scatterplots indicated heteroscedasticityand skewness. In an attempt to rectiQ these violations, both measures were transformedusing square mots and the resulting scatterplots suggest that the assumption of linearityand homoscedasticity were met. Finaily, the data were examined to determine whetherthe assumption of multicollinarity and singularity were met and the analysis indicatedthat there was no significant overlap between measures.

Statistical Analvsis, Means and standard deviations for the measures arepresented in TabIe 1. Themeans and standard deviationsare consistent with those foundfor the validation of the French version of the IUS (Freeston et al., 1994). Moreover, theinterna1 consistency of the IUS was excellent(a 94) and item-totd correlations rangedfrom .36 to -77 and are displayed in Table 2. A group of 66 participants were ce-tested onthe iUS after5 weeks, and the reliability coefficient was 1= -74.


26/99

Table 1

PSWQ 47.22 13.82BDI-II 10.54 7.84BAI 14.15 10.74

IUS = Intoleranceof UncertaintyScale:PSWQ = Penn StaceWorry Questionnaire;BDI-II Beck Depressionhventory-II;BAI=Beck Anxiety Inventory.


27/99

Table 2Means. Standard Deviations. and Corrected I t e w o t a l Correlations of the lLTSM=2761

No. Item M Se L,Uncertainty stops me from having a strong opinion.Being uncertain means that a person is disorganized,Uncertainty Makes H e intolerable.Tt's unfair having no guarantees in life.My mind can't be relaxed if 1 don't know what willhappen tomorrow.Uncertainty makes me uneasy, anxious, or stressed.Un foreseen events upset me greatly.It frustrates me not having al1 the information 1need.Uncertainty keeps me from living a full life.One should always look ahead so as to avoid surprises.A small unforeseen event can spoil everything, evenwith the best planning.W h e n t's time to act, uncertainty paralyses me.Being uncertain means that 1 am not first rate.When 1am uncertain, 1 can't go forwardWhen 1 am uncertain, 1 can't hinction very well.Unlike me, others seem to know where they are goingwith their lives.Uncertainty makes me vulnerable, unhappy, or sad.1 always want to know what the future has in store forme.1can't stand ki n g taken by surprise.The smallest doubt can stop me fiom acting.1should be able to organize everything in advance.Being uncertain means that 1Iack confidence.I think it's unfair that other people seem to be sure aboutthe future.Uncertainty keeps me from sleeping soundly.1must get away from al1 uncertain situations.The ambiguities in H e stress me.

27 I an't stand k i n g undecided about my future. 2.38 1.23 .52m. =Corrected item-total correlations.


28/99

2 1

Factor analysis was used to idenm the factor structure of theWS. Kaiser'smeasure of samplingadequacy for the intercorrelation matrix was -94, which Kaiser(1970) considered "marvelous" and appropriate for factor anaiysis. Cattell's (1966) screetest was used to help identify how many factors should be considered for extraction.Principal components analysis using SPSS version 10.0 was used to assess the factorstructureof the 27 items on the IUS. The first 10 eigenvalues were 10.94, 1.94, 1.32,1-13,1.04, -89, -84, -74, .71, and .68.A review of the eigenvahes suggests an initial fivefactor solution which is consistent with the French version and accounted for 60.7%ofthe variance; however, an examination of the scree test suggests that a more appropriatefactor solution may include les than 5 factors.

An iterated principd-factor analysis was then performed in which squaredmultiple correlations were used for the initiai commonaiity estimates. Furthemore, aRomax (oblique) rotation was employed to identify the underlying factor structure. Itemloadings for a 5-factor, dfactor, and 3-factor soIutions were examined. The scree testand item loadings were used to idenm a4-factor solution as the best representation ofthe results. Four eigenvalues were identified for this solution which were 8.07,8.71,6.10and 7. I 1 and the solution accounted for 56.8% of the variance.

The pattern mauix of the standardized regression coefficients for the 4 factors isprovided in Table 3. Keeping with the factor analysis of the French version, loadings of.30or pa t e r were considered for inclusionof items on factors. Factor 1 consisted of IOitems and represents the idea that uncertainty leads to the inability to act. Factor 2consisted of 12 items indicating that uncertainty is stressfui and upsetting. Seven items


29/99

Table 3R e m i o n Coefficients and Final

Comrnunality Estimates (h2lof the IUSM -2761NO. ~tem I n m rv h2Uncertainty stops me h m aving a smng opinion.Being uncertain means that a person is disorganized,Uncertainty Makes He intolerable.It's unfair having no guarantees in life.My mind can't be relaxed if 1 don't know what willhappen tomomw.Uncertainty makes me uneasy, anxious, or stressed.Unforeseen events upset me greatly.It frustrates me not having al1 the information 1need.Uncertainty keeps me h m Living a full life.One shouId always look ahead so as to avoid surprises.A smail unforeseen event can spoil everything, evenwith the bestplanning.When it's time to act, uncertainty paraIyses me,Being uncertain means that 1am not first rate.When 1am uncertain, 1 can't go forward.When 1am uncertain. 1 can't function very well.Unlike me, others seem to know where they are going

with their tives.Uncertainty makes me vulnerable, unhappy, or sad.1 always want to know what the future has in store forme.1can't stand king taken by surprise.The srnailest doubt can stop me h m cting.1should be able to organize everything in advance.Being uncertain means that 1 lack confidence.1think it's unfair that other people secm to bc sureabout k i r unire.Uncertainty keeps me h m leeping sonndly.1 must gct away h m l1 uncettain situations.The ambiguities in He stress me.1 can't stand king undecided about my future.Eigenvaimes 8.07 8.71 6.10 7.11m. alient regression coefficients are those 2.30 and appear in boldface. Factor 1=uncertainty leads to the inability to act; Factor II= uncertainty is stressfui and upsetting;Factor iii=unexpected events are negative and should be avoided; Factor IV =k in guncertain about the future is unfair.


30/99

23loaded on Factor 3, refer to the idea that unexpected events are negative, and should beavoided. FinaUy, Factor 4 consisted of 5 items that suggest that king uncertain is unfair.The correlations between the factors ranged from .42 to .69@ c O01) and are presentedin Table 4, thus verifying the use of oblique rotation. Finally, ali 4 factors were highlycorrelated with the overall IUS score and the correlations ranged from .82 to -94.

Correlation coefficientswere caiculated between the IUS and the other measures.The correlation matrix is presented in Table 5. The highest correlation for the IUSoccurred with thePSWQ L=60, p c .01); however, it was not significanrly higher thanthe corrdation between the IUS and the BDI-II and BAI. Results indicated significantpartid correlations between the IS andPSWQ, hen controlling for the BAI = -41,p< .mi),controlling for the BDI-II (r = .38, p c .@Il), and control1ing forboth the BAIand BDI-II Q= .30, < .01). These results show that the relationship betweenintolerance of uncertainty and worry remains after partiaiing out anxiety and depression.

A hierarchicai regression was performed to assess the predicted variance of worry(PSWQ)y entering demographic information (age and gender) in the first step. foltowedby measures of anxiety (BAI)and depression (BDI-E), and finally the measure ofintolerance of uncertainty (JUS). Intolerance of uncertainty continued to predict worryafter demographics andmoodstatehadbeen entered in and accounted for an additionai5%of the variance. Table6presents the results of the hierarchicai regression. Thebetacoefficients reported in the table were derived after ai1 the stepshad ken entered.

Finaiiy, a one-way between groupsANOVA was performed using intoIeranceofuncertainty scores. Individuals were grouped according to their responses on theWAQ.There were 45 (16%)participants who met the criteria for GAD by questionnaire,97


31/99

Table 4Correlation betwee~ actors on the SN- 2761

Factor 1 Facior 2 Factor 3 Factor4Factor 1 1-00Factor2 .69*** 1.O0Factor 3 .42*** .58*** 1.O0Factor4 .65*** .63*** .53*** 1.O0m. actor 1= uncertainty Ieads to the inability to act; Factor II=uncertainty isstresshl and upsetting;Factor =unexpected events are negative and should beavoided; Factor V =k i ng uncertain about the funire is unfair.


32/99

25

TabIe 5Correlations among Studv Measures. Gender. an- - 2761

Variable EUS PSWQ BDI-II BAI GENDER' AGErus - a*** .59*** .55*+* -.IO -.O6PSWQ - .61*** .59*** -.39*** -.O6BDI-II - .59*** -.14* -. 5*BAI - - . I F * -.15*GENDERa - -.O3AGE -m.US = Intolerance of Uncertainty Scale; PSWQ=Penn State Wony Questionnaire;BDI-II = Beck Depression hventory -II;BAI = Beck Anxiety Inventory."ender coding: 1 =Male; O = Female.* pc.05. or. ***pdo I .


33/99

Table 6arv ofHierarchicd MuItide Reeress'on Analvsis for ariables Predicune:Scores.1

on the PSWO (N=2761Variables B2 -R2 B 2Z.B BStep 1 .16** .16***

Gender" -9.27 1.34 -.28***

Step 2 .52*** .36***BAI 2.10 .49 .-3***BDI-Ti 2.83 .58 .26***

Step 3 .57*** .OS***lClS 29.64 5.36 .29***

PSWQ = Penn Stace Worry Questionnaire;BAI = Beck Anxiety Enventory; BDI-II= Beck Depression Inventory-U; iS = Intoleranceof Uncertainty Scde.a Gender coding: 1 =Maie;O =Femde.* pc05. *%.01. ***pcI.


34/99

(35%)participants who met the somatic criteziaody, and 121 (44%)participantswhomet none of the critena forGAD. Thirzeen (5%)participants were unclassifiable becauseof missing data andwere not included in the andysis, The resultsof the one-wayANOVA revealed that the groups differed significantly on intolerance of uncertainty [F(2,260)=41.18, pc .001]. Further, the Scheff test for group cornparisons indicated thatparticipantswho met the cnteria for GAD by questionnaire scored significantIy higher onthe nTS than those who met only the somatic criteria and those who met none of thecriteria for GAD. Moreover, those participants meeting onIy the somatic criteria scoredsignificantlyhigher on the IUS han those who met none of the criteria forGAD.Discussioq

The results confirm the study's predictions. The English version of theIntolerance of Uncertainty Scale (JUS) bas excellent internal consistency and good test-retest reliability. A Cfactor structurewas identified which suggests that the items on theIUS eptesent the idea that intolerance of uncertainty is stressful and upsetting,uncertainty leads to the inability to act, uncertain events are negative and should beavoided, and being uncertain isnot fair. Although the French version of the iS has a S-factor solution, the ideas represented by the factors are similar enough to support theconsistency of theU S cross the French and English versions. However, the 4-factorsolution identified in this study, appears to more cIeariy capture the underlying factors ofthe US. Whiie this may aIIow researchers to assessdifferentaspects of an individuai'sintolerance of uncertainty and better understand the underlying themes, it does not seemappropriate to use the factors as subscales due to the apparent overlap of factors anditems on those factors. Futther, aii the factors are signifcantiy related to the ovedl score


35/99

on the IUSand there areno significant differences between those relations. At this pointin time, although the four factors allow mearchers to get a fuller idea of the breadth ofintolerance of uncertainty, the data suggest that the overailIUS core should be used.

Although the correlation mak foUows an expected pattern of results with thehighest correlation occurring between intoIeraace of uncertainty and worry, thiscorrelation was not significantly higher than the correlation between intolerance ofuncertainty and anxiety and depression. Research has aIready demonstrated that worry isclosely related to mood States such as anxiety and depression; therefore, it is notsurprising to find high correlations between hese constructs (Andrews& Borkovec,1988;Borkovec, Robinson, bzinsky, &DePree, 1983;Zebb& Beck, 1998). However,significant partial correlations indicate a unique reiationship between intolerance ofuncertainty and worry that goes beyond the shared variance with negative affect. Inaddition, regression analysis indicated that worry continued to predict intolerance ofuncertainty beyond demographics and mood state. This supplies further evidence for theunique relationship between intolerance of uncertainty and wony.

The IUSwas able to distinguish between groups of participants who met thecriteria forGAD by questionnaite, those who met the somatic criteria only, and thosewho met none of the criteria forGAD by questionnaire. Specifically, participants whomet the criteria forGAD by questionnaire scored significaniiy higher on the IUS hanparticipants who met only the somatic criteria and those who met none of the criteria forGAD. Moreover, those who met thesomatic criteria forGAD by questionnaire scoredsignificantty higher on the IUS han those who met none of the criteria for GAD. Theseresults support the measure's critexionrelatedvaiidity and this suggests that the ltTS can


36/99

29play a discriminant role in the assessment of GAD. Finaily, it is important to note thatalthough a high percentage of individuais meeting the criteria forGGD by questionnairewere identified (16%), this is typical of self-report measures and is consistent withprevious research that found a high rate of faise positives when using questionnaires toassess for the presence of GAD in nonclinicai populations (Roemer et al., 1991).

At this point, the IntoIerance of Uncertainty Scale (TS) has proven to be a validand reliable instrument for the assessment of intolerance of uncertainty. However, thereare some limitations to the present study. The f m t limitation stems from the fact that77% of the participants were female. Although the results revealed no gender differenceson the iUS, and these results are consistent with those found in o ther studies (Robichaud& Dugas, 2000), gender differences were noted for the other measures and this may haveaffected the results.

Secondly, the participants in the study were undergraduate students and the resuttsrnay not generalize to other populations. Aithough research in ch ica l samples using theFrench version of the IS has demonsimted its ability to distinguish between GADpatients, patients suffering from a variety of other anxiety disorders, and normal controls(Dugas, Gagnon et al., 1998;Ladouceur et ai., 1999), further research is needed toreplicate the present findings with the English version in both community and clinicalsamples.

In surnmary, the present study has demonstrated the sound psychometricproperties of the English version of the intolerance of Uncertainty Scale. These findingsare consistent with those found for the French version and support the use of thismeasure. Future research should attempt to focus on validating the Engiish version with


37/99

30

different populations and attempt to establish further the specificity of the relationshpbetween intolerance of uncertainty and worry. However, at this point it seems clear thatthe intolerance of Uncertainty Scale (IS ), which has k e n shown to be a reiiable andvaIid instrument, wiii play a key role in the iurther exploration of the relationshipbetween intolerance of uncertainty and worry.

Study TwoThe present study examined the specificity of the relationship between intolerance

of uncertainty and worry. Aithough research has begun to lay the foundation forunderstanding the role of intolerance of uncertainty in excessive worry, it is unclearwhether the relationship between intolerance of uncenainty and worry is not alreadyaccounted for by factors aiready believed to play a role in worry, such as perfectionismand perceived control. The present study exam ined the relationship between worry andperfectionism, perceived controt, and intoIerance of uncertainty in an attempt todemonstrate whether intolerance of uncertainty and worry share a unique relationship.Moreover, this study attempted to differentiate intolerance of uncertainty from thetraditional broader concept of intolerance of ambiguity by assessing their reIationship toworry.

The study had three hypotheses. First, based on the findings suggesting a smngrelationship between intolerance of uncertainty and worry, it is predicted that worry willbe more highiy related to intolerance of uncertainty than to perfectionism, perceivedcontrol, and intolerance for ambiguity. Second, it is proposed that the relationshipbetween intolerance of uncertainty and worry wiii not be accounted for by the other studyvariables. Finally, based on responses to a questionnaire assessingGAD riteria, it is


38/99

believed that intolerance of uncertaintywiU distinguish between participants meeting ail,some, or none of the diagnostic criteria forGAD, controlling for perfectionism, control,and intolerance of arnbiguity.Method

Participants, One h u n M and ninety-seven = 197) participants were recruitedthrough various undergraduate courses. There were 152 femaie participants and 45males. The mean age of participants was 22.56 =5.5). Students were invited toparticipate at the start of a regular undergraduate course and participation was voluntary.

Instruments. The participants completed the following questionnaires in randomorder: the intolerance of Uncertainty Scale (IIJS), the Penn State Worry Questionnaire(PSWQ), the Worry and Anxiety Questionnaire (WAQ), the MultidimensionalPerfectionism Scale (MPS), the Senseof Control Scale (SC), and the Scale of Tolerance-intolerance of Ambiguity(TM),.n addition, abjects completedademographicinformation form.

The Lntolerance of Uncertaintv Scale(WS:reeston et al., 1994) includes 27items relating to the idea that uncertainty is unacceptable, reflects badly on a person, andleads to frustration, stress, and the inability to take action. As demonstrated in StudyOne, the Engiish version of the N S as excellent internai consistency (a .94), andgood test re-test reliability over a five-week period = .74). Moreover, the measure hasdemonstrated convergent and divergent validity when assessed with symptom measuresof worry, depression, and anxiety.

The Penn State Wory Questionnaire(PSWQ: Meyer et ai.,1990) measures thetendency to engage in excessive, uncontrollabIe, and generalized worry. As stated


39/99

32previously, the PSWQ hasexcellent internai consistency,good test re-test reliability anddemonstrated validity. See page 13 for a full description of the PSWQ.

The Wocry and Anxiety Ouestionnaire (WAQ: Dugas et al., 2001) assesses worrythemes and the DSM-IV iagnostic criteria forGAD. As stated previously, the WAQshows good test-retest re1iabiIit.y anddemonstrated vaiidity. See page 14 for a fulldescription of the WAQ.

The Wdimensionai Perfectio- Scale. (MPS:Hewitt& Fiett, 1989) is a45-item measure of personal characteristics and traits associated with perfectionism. TheMPS has three subscaies: self-oriented perfectionism (SOP), which examines self-directed perfectionism; sociaily-prescribed perfectionism (SPP),which assesses the needto meet the expectations of others, and other-oriented perfectionism (OOP), which tapsthe expectations about the capabilities of others. Participants rate items on a 7-pointLiiert scale ranging fiom 1= " m v is-" to 7=" Items assessingself-oriented perfectionism include 'When 1work on something, 1cannot relax until it isperfect". Items exarnining socially-prescribed perfectionism include "The people aroundme expect me to succeed at everything I do". Finaily, other-oriented perfectionism isassessed through items such as "If I ask someone to do something, 1expect it to be doneflawlessly". The MPS has excelient internai consistency in a student sarnple (SOP:a =.89; SPP: a = .86; OOP: a =-79; Hewitt& Flett, 1991). Fially, the MPS subscaleshave demonstrated convergent and divergent validity (Hewin& Flett, 199 1; Hewitt, Fiett,Turnbull-Donovan,&Mikail, 199 1).

The Sense of Control Scde(SC: Lachman&Weaver, 1998) is a l2-itemquestionnaire assessingan individual'ssenseof conml through two dimensions: personal


40/99

mastery and perceived constraints. Personal mastery reflects one's personal sense ofeficacy in reaching goals, whiIe perceived constraints assesses the belief in obstaclesbeyond one's co nm l that may interfere with achieving goals. Participants rate each itemon a 7-point Likert scale ranging from 1= "disamee stronely" to 7 = "stronoly amee".Examples of items include "1can do just about anything I really set my mind to" and" m a t happens to me in the future mostly depends on me". High scores are indicative ofa strong sense of conm l. Factor anaiysis supporh the two dimensions and analysisindicates that the measure has high interna1 consistency (Personal mastery: a = .70;Perceived constraints: ot= 86; Lachman & Weaver, 1998).

The Scale of ToIerance-Intolerance of A m b w. (TIA: Budner, 1962) is a 16-item questionnaire that assesses intolerance of ambiguity. The scale examines thetendency to perceive ambiguous situations as a source of threat and items refer to threefeatures of ambiguity: novelty, complexity, and insotubility. Participanis are asked torate each item on a &point Likert scaie ranging from O ="WYi e " o 5 ="s troncl~ g&'. Examples of items include "A good job is one where what is to be doneand how it is to be done are always clear" and "W hatwe areused to is always preferableto what is unfamiliar". The scde has good test-retest retiability over a 2 month period &= .85) but only moderate internai consistency (a = 49 to .59; Budner, 1962; Fumham,1994). However, the measure has demonstrated vaiidity (Budner, 1962; Funiham, 1994).

Procedure. Participants were asked to comptete the six questionnaires and supplydemographic information. The questionnaires were completd during one 30-minutetesting period and groups of participants were run on several separate occasions.Participants were told that the purpose of the experiment was to assess the telationship


41/99

34between worry and constmcts related to worry. In addition, participants were infonnedthat they could discontinue the study at any tirne.Results

Overview of Statistical Analvsis. A correlation matrix was used to assess therelationship between study measures and test the hypothesis that worry would have ahigher correlation with intolerance of uncertainty than dimensions of perfectionism,control, and intolerance of ambiguity. Furthemore, a partiai correlation was utilized toassess the unique relationship between intolerance of uncertainty and worry, once sharedvariance with the other measureswas removed. In addition, partial correlations betweenworry and dimensions of perfectionism, contrd, and intolerance of arnbiguity wereassessed to determine if a relationship remained once variance shared with intolerance ofuncertainty was partialed out.

A hierarchicai cegression was performed to assess the predicted variance of wony(PSWQ),by entering demographic information (age and gender) in the first step,followedby measures of perfectionism (SOP, SPP, OOP), control (SC), and intoleranceof ambiguity (TIA) in the second step, and the measure of intolerance of uncertainty(IUS) was entered in the finally step.

Finaily, a one-way between groups Analysis of Covariance (ANCOVA)wasperformed using intolerance of uncertainty scores, controiling for the other studymeasures (SOP, SPP, OOP, SC,TIA). Individuais were grouped according to theirresponses on theWAQ. This anaIysis was used to test the finai hypothesis that the IUSwould be able to distinguishbetween groupsof participants who met the Midiagnostic


42/99

criteria forGAD, those who met onIy the somatic criteria, and those who met none of thecriteria for GAD, controI1ing for the effects of the other study mesures.. .imnarv Data Prior to any specific statistical andysis the data werescreened to determine whether statistical assumptions were met and to ascertain if thedata was appropriate for further statistical anaiysis (see Tabachnick and Fidell, 1996 for areview of data screening procedures). The data, excluding demographic information,were transformed into z-scores to evaiuate the presence of extreme scores that were morethan 3.29 standard deviations from the mean in either direction. Two participants wereidentified as univariate outliers for having extreme scores on study measures.Mutlivariate outiiers were assessed by examining Mahalanobis distance. For thisanalysis, ail study measures, excluding demographic information, were included and thePSWQ was identified as the dependent variable. The resulting Mahalanobis distance foreach participant was compared against a criticalX2value. The two participants that werealso identified as univariate outiiers were identified as multivariate outliers due to scoresexceeding this critical value and were removed from further statistical analysis.

The data were also assessed for normaiity by examining the skewness andkurtosis of the distribution for each measure. Al1 study measures were deemed normallydismbuted. The assumption of linearty and homoscedasticity was verified through theexamination of the bivariate scatterplots between PSWQ and ail other measures (IUS.SOP, OOP, SPP, SC,and TIA). The assumption of Iinearity was considered to beviolated if a nonlinear relationship was found. in addition, the assumption ofhomoscedaticitywas met if thepattern on the scatterpIot mggested variance wasnormaIIy distributed These assumptionswere met for aDmeasutes. FialIy, the data


43/99

were examined to determine whether the assumption of multicoIIinarity and smgularitywere met and the anaiysis indicated that there was no significant ovedap betweenmeasures.

Statistical Anal- Means and standard deviations for the measures arepresented in Table 7. Correlation coefficients were calcuIated between the I S and theother measures and the correlation matrix is presented in Table 8. The strongestcorrelation occurred ixtween the IUS and PSWQ L= .63, < 001). This correlationwas significantIy higher than the correlations between the PSWQ and the SPP (f = 37. p


44/99

Table 7Means and Standard Deviations forStudv Measures +J= 197)

Variable Mean Standard DeviationIUS 61.25 18.98PSWQ 48.68 14.00SOP 68.88 14.99SPP 50.76 12.70

TIA 32.09 8.64Note. S= intolerance of Uncertainty Scde; PSWQ = Penn State Worry Questionnaire:SOP = Self-riented Perfectionism; SPP=Socidly-Prescribed Perfectionism; OOP=Other-Oriented Perfectionism; SC =Sense of Conml Scale; TIA =Tolerance-intoletance


45/99

Table 8Correlationsm g tudv Mwures. Gender.and ee -197)

Variable PSWQ S SOP SPP OOP SC TIA Ciender* AGE

N S - .33*** .52*** - 1 1 -.48*** .43*** -.O7SOP - .38*** .38*** .O2 .I8* -.I7*SPP - .I6* -.36*** .35*** .O4OOP - .O -.15* .13SC - -.32*** .O5T U .O3Gender'AGE

Note. KJS = Intolerance of Uncertainty Scaie; PSWQ = Penn State Wony Questionnaire;SOP =Self-Oriented Perfectionism; SPP=Socially-Prescribed Perfectionism; OOP =Other-Oriented Perfectionism; SC =Sense of Control Scaie;TIA=Tolerance-htoleranceof Ambiguity Scaie.

Gender coding: 1 = Maie;O =Femaie.* pL05. **p


46/99

Table 9Summarv of Hierarchical MmleR em sio n Analvsis for Variables Predichn~. Scoreson the PSWO (N=197l

Variables Bf bR2 B SEI$ l3Step 1 .12*** .12***

Gende? -6.62 1-87 -.20***Age - 25 -14 -.0

Step 2 .34*** =***SOP .15 .O6 .16*SPP - .O3 .O8 .O3

TIA - .O5 .10 -.O3Stcp 3 .47*** .14***

l'US .36 .O5 .50***Note: PSWQ= Penn State Wony Questionnaire; SOF =Self-Oriented Perfectionisrn:SPP: Socially-Prescxibed Perfectionism;OOP=mer-riented Perfectionism; SC =Sense of Control Scaie;TIA=Tolerance-Intolerance of Ambiguity Scale; IUS =Intolerance of Uncertainty Scale.'Gender coding: 1=Male;O =Female;


47/99

40Finally, a one-way between groups ANCOVA was perfonned on intolerance of

uncertainty scores, controiiing for the other study measutes (SOP, SPP, OOP, SC,TM).Participants were grouped according to their responses on the WAQ. This analysis wasused to test the final hypothesis that the IUS would be able to distinguish between groupsof participants who met the diagnostic criteria for GAD by questionnaire, those who metonly the somatic criteria, or those who met none of the criteria for GAD, controlling forthe effects of the other study measures. There were 11 (5.7%) participants who met thecriteriaforGAD by questionnaire,65 (33.7 8)participants who met the somatic criteriaonly, and 11 1 (57.5%) participants who met none of the criteria forGAD byquestionnaire. Six (3.1%) participants were unclassifiable because of missing data andwere not included in the analysis. The results of the one-way between groups ANCOVArevealed a significant group effect[E 2, 179)=6.416, p = .002]. Further, the Scheff testfor group cornparisons indicated that participants who met none of the criteria forGADby questionnaire scored significantly lower on the S han those who met the hl1criteria for GAD and those meeting the somatic criteria. However, those meeting thesomatic criteria did not score signiF1cantiy lower on the IUS than those meeting the fullcriteria forGAD by questionnaire.Ilhusim

The results of the study confm the initial predictions. Correlations arnong studymeasures demonstrated that intolerance of uncertainty had the strongest relationship withworry. Taken a step further, when variance shared with other measures was removed,intolerance of uncertainty continued to be related to worry. These fmdings suggest thatintolemce of uncertainty and worry share variance that is not explained by measures of


48/99

intolerance of ambiguity, petfectionism, or perceived control.Furthermore, the relationship between intolerance of uncertainty and worry was

significantly stronger than the relationship between intolerance of ambiguity and worry.Sirnilarly, the correlation between intolerance of ambiguity and intolerance ofuncertainty, although significant, does not suggest that they are measunng the sameconstruct, Taken together, these findings clearly point to the idea that intolerance ofuncertainty and the broader concept of intolerance of ambiguity are different constructsthat maintain distinct relationships with worry.

Conceming the relationship between dimensions of perfectionism and worry,aithough self-oriented perfectionism and sociaily-prescribed perfectionism weresignificantly related to worry, other-oriented perfectionism was not. This is consistentwith previous findings that show that self-oriented perfectionism is related tomaladjustment problems such as anxiety (Flett et ai., 1989) and that socially-prescribedperfectionism is rdated to worry (Flett et al., 1995). Furthermore, the study re-established the relationship between worry and perceived control (Davey, 1994).However, despite the findings confuming a relationship between worry andperfectionismand control, the hypothesis that the strongest relationship would emergebetween intolerance of uncertainty and worry was confinned.

The importance of the relationship between worry and intolerance of uncertaintywas firrtherestablished by exarnining partial correlations between worry andperfectionisrn, perceived controi, and intolerance of ambiguity, by removing varianceshared with intolerance of uncertainty. Once intolerance of unceaainty was partialed out,ody the relationship between worry and seif-oriented perfectionism remained. In a


49/99

42fucrher attempt to clarify the nile of intolerance of uncertainty, study masures were usedto predict woq . Intolerance of uncertainty emerged as the strongest predictor,continuing to predict worry above and beyond ail other study measwes. In fact,intolerance of uncertainty continued to predict a . dditional 14%of the variance inworry. These are strong findings that support the central role of intolerance ofuncenainty in worry.

Finally, intolerance of uncertainty was able to distinguish between individualswho met none of the criteria forGAD by questionnaire, from those meeting the fullcriteria forGAD nd those meeting only the somatic criteria forGAD, ontrolling forintolerance of ambiguity, perfectionism, and perceived control. This suggests thatintolerance of uncertainty can play a discriminate role in assessing individualsexperiencing excessive levels of worry and somatic anxiety.

The present study suggests that there is a significant relationship between worryand intolerance of ambiguity. However, the relationship between worry and intoleranceof uncertainty was significantly stronger. Moreover, the correlation between intoleranceof ambiguity and intoIerance of uncertainty suggests that, although they may be related,they appear tobe measuring different things. Furnham (1994) reviewed the literature onintolerance of ambiguity starting at its origins in the late 1940s and suggests thatintolerance of ambiguity has been used to assess a number of different constnicts andoutcome measures. He cites research that Linked intolerance of ambiguity to re!igiousbeiiefs, attitudes towards censotship, career choices, and conservatism. Recent researchcontinues to examine the relationship between intolerance of ambiguity and a variety offactors, such as fearof the paranormal (Houran& Lange, 1996) and political orientation


50/99

43(Fibert& Ressler, 1998). The factors examined in relation to intolerance of ambiguityhave not been commonly associated with wony; therefore, it is not surpriskg thatintolerance of ambiguity itself is not as highly correlated with worry as intolerance ofuncertainty. intolerance of uncertainty maintains a stronger relationship to worry becauseit appears to be measuring something that is not captured by the broadly used construct ofintolerance of ambiguity.

The significant relationship behveen worry and perfectionism is consistent withthe idea that individuds who place high expectations or standards on themselves mayexperience womsome thoughts amund meeting those expectations. However, the linkbetween worry and perfectionism was not as strong as the relationship between worryand intolerance of uncenainty. These findings may be a result of the underlying aspectsof perfectionism. Hamacheck (1978) postulated that perfectionism is composed of twodimensions and he clearly differentiates between normal or adaptive perfectionism andmaladaptive or pathological perfectionism Hamacheck ( 978) suggests that the formermay be differentiated from the latter by the ability to derive pleasure from one's efforts.individuals who have n omai perfectionistic tendencies are more likely to be successfulachievers who gain a sense of pleasure from reaching ttieir goals. AItemativeIy,individuais with maiadaptive perfectionism are likely to be engaged in efforts to obtainimpossible goaIs and betieve that things are never quite perfect. These individuais mayexperience distress and concern regarding their attempts for perfectionism. in this sense.the results of the present study are not surprising given that there are positive andnegative aspects of perfectionism. Some aspects of perfectionism are adaptive andtherefore are not k e l y to be associated with high Ievels of worry. These differing


51/99

dimensions of perfectionism may have resulted in only a moderate, albeit simcant,correlation between worry and perfectionism.

To lend further support for this explanation of the present findings, Frost andcolleagues (1993) performeda factor analysis on two perfectionism scales including themeasure used in the present study. They identified two underlying factors: maiadaptiveevaiuation concems and positive striving. These two factors are consistent withHamacheck's (1978)description of maiadaptive and normal perfectionism. Cornmonsense suggests that maladaptive perfectionism may result in distress and thetefore berelateci to worry. However, the positive aspects of perfectionism may have detractedfrom the overall relationship between wony and perfectionism, and may be responsiblefor the lower correlation between wotry and perfectionism compared to worry andintolerance of uncertainty. Altematively, intolerance of uncertainty is not adaptive andmost aspects of it can beconsidered negative. Someone who is intolerant of uncertaintymay tend to worry regardless of whether things are perfect now, because there isuncertainty around how things will be tomorrow and that uncertainty is likely to causethem distress. Therefore, it is reasonable to expect that one would find a strongerrelationship between worry and intolerance of uncertainty.

The r d t s of the present study also demonstrated a significant relationshipbetween worry and perceived control. It is easy to imagine how a lack of perceivedcontrol may beconnected to level of worry. If someone determines that they have nodirect control over situations or events they may become distressed or womed about howthose situationswill turn out. Other research supports this idea and has linked a kick ofperceived conuol to a variety of mentai health outcomes and constnicts (see Skimer,


52/99

45

1996for a review). However, despite conftrminga Iink between worry and perceivedcontrol, the present study points to a stronger connection between worry and intoleranceof uncertainty.

Perceived control can be considered a stable personaiity trait and as an unstable,situation-specific state. One can imagine that there are individuais who approach lifewith an overall sense of control. These people rnay believe that they have the necessaryskilIs and abilities to have an effect on their environment. It is just as likely that there arespecific situations where individuals rnay feeI that they have more or less control. In onesituation, individuais rnay feel that they have the necessary skills and abilities toaccomplish a specific goai and further believe that there are no extemal obstacies in theirway. in an alternative situation, those individuals rnay feel that they are lacking thenecessary skiIIs to accomplish a goal and beiieve that there are a number of extemalobstacles that rnay interfere with obtaining that goal. This suggests that there aresituations where people will feel a pa t e r sense of control and other situations wherethey will feel a lack of control. Therefore, the lower correlation between worry andperceived control in the present study rnay be due to the variations in perceived control asa resuit of diferent situations.

On the other han& uncertainty can be found in everyday iife given that mostsituationsare not straightforward and generally contain some element of uncertainty.One can never be certain how situations wiii resolve. Someone who is intolerant ofuncertainty, according to the definition,wiU find any uncertainty unacceptable.Therefore, it is not surprising to find a stronger relationship between worry andintolerance of uncertainty, becauseuncertainty is aiways present and someone who Fmds


53/99

uncertainty unacceptable wilI likely experience a great deal of concem and distressaround that uncertainty.

Another possible explmation for the stronger relationship between worry andintolerance of uncertainty as opposed to worry and perceived control, may be the result offocusing on perceived control rather than other aspects of control. According to SchuIzand Heckhausen (1999), research tends to focus exclusively on perceived control andignores other aspects of control. They believe that a number of processes, functions, andbehaviours are not captured by perceived conml. Moreover, they suggest that justbecause a person has a low estimate of personai control does not rnean that they are notactually engaging in behaviours that are aimed at exerting control over their environment.Perceived control may not be directly connected to objective control. Therefore,perceived control rnay not be the best indicator of control and accordingly a perceivedlack of personal control may not mean that the person experiences worry or distress.

Finally, one must consider how people who have a lack of perceived controlinterpret that lack of conirol. It is possibIe that individuais who estimate that they have alow level of control rnay not experience anxiety or worry related to that lack of conuol.These individuais may adopt the attitude ihat because they have no control over thesituation they are not responsible for the outcomeand consequently feel no distress orconcern over the situation. Whereas someonewho is intolerant of uncertainty is unlikelyto adopt an attitude where they are unconcemed about the situation because someonewho is intolerant of uncertainty by definition fin& uncertainty unacceptable. Individualswho are intolerant of uncertainty are likely to experience distress or worry in situationsthat they interpret as uncertain.


54/99

Although important findings have e m e w from this study, t is not without itslimitations. Similar to the f m t study, the majority of participants were femaie. Theresults suggest that there were gen&r differences on the measureof worry (PSWQ) andon one of the dimensions of perfectioaism(SOP). Due to the gender differences on thosespecific measures. the discrepancy in the number of male and female participants mayhave affected the results. in addition, the study was conducted using a composite ofundergraduate srudents, the majority of which were psychoIogy students. Therefore, thegeneralizability of the results to the general population should be done so with caution.Future studies should strive to include a sample that incorporates anequal distribution ofmales and females and should examine alternative nonclinical populations.

AIthough research has suggested that simiIar pmcess may be involved in b thclinical and nonclinical worry (see Dugas & Ladouceur, 1998), the present study wasconducted on a nondinical sample and the resuIts may not generalize to cIinicalpopulations. AIthough mearch in din ical populations has aiready incorporated theconcept of intoIerance of uncertainty (Le. Dugas, Gagnon et d.,1998), hiture researchmay want to replicate the present findings in regards to the reiationship between worry,and intolerance of uncertainty, erfectionism, control, and intolerance of ambiguity in adi ni cd sample.

Finally, the Tolerance-Intoleranceof Ambiguity Scale (TIA) eveloped byBudner (1962)was chosen for inclusion in the present study. However, as noted earIierthere are a number of masures that assess the broad concept of intolerance of ambiguity(see Furnham, 1994). These rneasures appear to have a varey of different underlyingfactors and the use of an alternative measure may have produced different resuIts,


55/99

especially if the measure chosen was more closeiy related to intolerance of uncertainty.However, Furaham's (1994) review of the tolerance of ambiguity measures indicates thatonly a few of the factorsappearto be related to intolerance of uncertainty and thereforeregardless of the measure used ihe results would liely be consistent with the presentfindings.

In conclusion, the present study suggests that intolerance of uncertainty isspecifically related to worry. Tite study showed that intolerance of uncertainty had thestrongest relationship withworry when compared to factors already associated withworry such as perfectionism and perceived control. Furthemore, the study was able todemonstrate the difference between intoIerance of uncertainty and intolerance ofambiguity by demonstrating their distinct relationship with worry. Finally, the resultspoint to a unique relationship between intolerance of unceaainty and worry that cannotbe explained by factors already related to worry. At this time, it appears clear thatintolerance of uncertainty is a key construct in understanding excessive worry.

Generai DiscussionTaken together, StudiesOne and Two show that the Intolerance of Uncertainty

Scale (TUS) is a sound measure of intolerance of uncertainty and that the relationshipbetween intoleranceof uncertainty and worry shows evidence of sensitivity andspecificity. The relationshipbetween intolerance of uncertainty and worry is notaccounted for by shared variance with anxiety, depression, perfectionism, and control. Inaddition, the findings indicate that intolerance of uncertainty and intolerance of ambiguityare distinct constnicts. Using the present results as a base, research can continue to


56/99

49

investigate the specificity of the relationship between intolerance of uncertainty andwony.

Although research, including the present expriment, has now established thatintolerance of uncertainty and worry are highly related, it is stilI unclearhow exactlyintolerance of uncertainty might Iead to elevated levels of worry. It has been proposedthat intolerance of uncertainty is a filter through which individuals view their world. Itwould be interesting to determine how the filter functions by examining intolerance ofuncertainty and information processing. Does intolerance of uncertainty result in anattentional bias for uncertainty, an enhanced memory for uncertain information, or is it amatter of interpretation?

Two recent studies, conducted by our research team, have attempted to shed sometight on the relationship between intolerance of uncertainty and information processing.The first study examined whether people who are intolerant of uncertainty have a bias inhow they process uncertain information (Karavidas, Dugas, & Buhr, 2001). Morespecifically, the study assessed whether people who are intolerant of uncertainty have abias towards words representing uncertainty compared to matched control words. Thestimuli for the study were generated froma list of words believed to contain elements ofuncertainty, and included words such as "unknown", "unpredictable", and "uncertainty".A set of conml words, which included "identifiable", "career", and "unitary", wasmatched to the uncertain words on a number of characteristics including neutraiity, partof speech, familiarty,concreteness, and frequency of use. The words were shown toparticipants and they were later asked to recall as many words as they could.


57/99

The results suggest that individuals who are highly intolerant of uncertainty havea bias towards uncertain information. SpecificaiIy, individuals high on intolerance ofuncertainty recalled a greaterproportion of uncertain words, compared to individuals tawon intolerance of uncertainty. Although these results suggest that individualswho areintolerant of uncertainty display a bias in processing uncertain information, it is unclearwhether they selectively attend to, or have an enhanced memory for, uncertain words.

The second study attempted to ascertain how people who are intotemnt ofuncertainty interpret uncertain or ambiguous situations (Hedayati, Dugas, & Francis,2001). Participantswere given diary entries that were positive, negative, or ambiguous innature. For example, "1went io Amanda's party last night, it was fun!" (positive),"1wentto the hairdressets this morning, my new hairstyle is atrocious, 1 look awful" (negative),and "1phoned the doctor today and was surprised to heu he results of last week's check-up" (ambiguous). The participants were asked to rate their level of concern for eachentry.

The resutts demonstrated that individuds who were identified as intolerant ofuncertainty tended to interpret ambiguous entries more negatively than individu& lowon intolerance for uncertainty. in addition, the tendency to interpret ambiguous situationsas h a t e n i n g was more closeIy related to intokrance of uncertainty than worry, anxiety,and depression. Overail, these two studies suggest that people who are intoIerant ofuncertainty have a bias in how they processuncertain information.

Aithough these studies provide inte res thg initial fmdings regarding intolerance ofuncertainty and n f o d o n processiag, more research is needed. Futwe research shouldattempt to replicate thesepreliminary ikdin gs and begin to tease apart the relationsfp


58/99

between intolerance of uncertainty and information processing. Such information canenhance the way we understand intolerance of uncertainty and may provide answersregarding who is at risk for developing excessive worry and what should be targeted inintervention strategies. RegardIess of where future research endeavors lead, the presentstudy has demonstrated that the Intolerance of Uncertainty Scale (IUS) is a vad tool formeasuring intolerance of uncertainty and has further established intolerance ofuncertainty as process involved in wony.


59/99

ReferencesAndrews, V. H., & Borkovec, T, D. (1988). The differential effects of inductions

of wony, somatic anxiety, and depression on emotionai experience. Journal of Behavior'hera~v nd ExperimentaI Psvc- & 21-26.

American Psychiatrie Association. (1994). -ostic and statistrcd manual O. . fmental disorciers (4th edition). Washington,DC:Author.

Beaudoin,S., Tremblay, M., Carbonneau, C., Dugas, M. J., Rovencher,M.. &Ladouceur, R. (1997, October). Vali&on d'un instrument diamostique !out le troubled'anxit ~nraliseVdidation of a diagnostic measure for generalized anxietydisorder]. Poster presented at the annual meeting of the Socit qubecoise pour larecherche en psychologie, Sherbrooke, Canada.

Beck, A. T., Epstein, N., Brown, G., & Steer,R A. (1988). An inventory for. .measuring clinicai anxiety: Psychometric properties. Journal of Co and Clinical

Psvchoiogv. 56,893-897.Beck, A. T., Steer, R.A., & Brown,G. K. f 1996). Beck Deuression inventory

Manual (2nd Ed.). San Antonio. TX:Psychological Corporation.Beck, A. T., Ward, C.H.,Mendelson, M., Mock, J., & Erbaugh, J. (1961). An

inventory for measuring depression. Archives of General Psvchiatrv,IS, 7-62.Borkovec, T. D. (1985). Worry: A potential valuable concept Behaviour

Wearch and The- 481-482.Borkovec, T. D. & Inz, J. (1990).The nature of worry in generaiized anxiety

disorder: A predorninance of thought activity.&&tvioar Research and The- 28,153-158.


60/99

53Borkovec,T. D., Robinson, E., M n s k y , T., DePtee, J. A. (1983). Preliminary

exploration of worry: Some characteristics and processes. Behaviour Research and'hera~v,a - 16.

Borkovec, T. D., Shadick,R. & Hopkins, M. (1990). The nature of paihologicaland normal worry. in D. Barlow,& R. Rapee, (Eds),Chronic an-d GeneriikwlAnxietv Disorder. New York: Guilford Press.

Brown, T, A., Antony, M. M., & Barlow,D. H. (1992). Psychometric propertiesof the Penn State Worry Questionnaire in a clinical disorder sample. Behaviour Researchand Therapv, a33-37.

Budner, S. (1962). Intolerance of ambiguity as a personality variable. Journal ofPersonalua 9-59.

Butler,G., &Mathews, A. (1983). Cognitive processes in anxiety. Advances inBehavior Researd andThe- L 51-565.

Butler, G., Wells, A. &

Gender in Worry

Documents

Gender in Worry