This article was downloaded by: [McMaster University]On: 29 October 2014, At: 12:53Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: MortimerHouse, 37-41 Mortimer Street, London W1T 3JH, UK

Cognitive Behaviour TherapyPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/sbeh20

Expectations and Attributions in Social AnxietyDisorder: Diagnostic Distinctions and Relationship toGeneral Anxiety and DepressionSteven Taylor & Jaye Walda University of British Columbia , Vancouver, CanadaPublished online: 17 May 2010.

To cite this article: Steven Taylor & Jaye Wald (2003) Expectations and Attributions in Social Anxiety Disorder: DiagnosticDistinctions and Relationship to General Anxiety and Depression , Cognitive Behaviour Therapy, 32:4, 166-178, DOI:10.1080/16506070310020315

To link to this article: http://dx.doi.org/10.1080/16506070310020315

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose ofthe Content. Any opinions and views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be reliedupon and should be independently verified with primary sources of information. Taylor and Francis shallnot be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and otherliabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to orarising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Expectations and Attributions in Social AnxietyDisorder: Diagnostic Distinctions and

Relationship to General Anxiety and Depression

Steven Taylor and Jaye WaldUniversity of British Columbia, Vancouver, Canada

Abstract. Contemporary cognitive models suggest that social anxiety disorder arises from a number ofcognitive factors, including tendencies to form pessimistic (rather than optimistic) attributions andexpectations for socially-related events. These models also assume that the strengths of such attributions andexpectations are more closely linked with social anxiety than with general anxiety or depression. To testthese assumptions, a battery of self-report measures was completed by participants with a primary diagnosisof generalized social anxiety disorder (n = 75), panic disorder with agoraphobia (n = 44), or post-traumaticstress disorder (n = 59). To examine differences on these cognitive variables, group comparisons wereperformed controlling for general anxiety, depression and medication status. Social anxiety disorder,compared with panic disorder with agoraphobia and post-traumatic stress disorder, was characterized bylower expectations for positive social events and higher expectations for negative social events. There wasno difference among the groups on expectations for non-social positive or negative events. Stable and globalattributions for social negative events were more closely associated with social anxiety disorder than withpanic disorder with agoraphobia and post-traumatic stress disorder. Correlational analyses also revealedspecific relationships among social-cognitive measures and social anxiety, even after controlling for generalanxiety and depression. The results are consistent with cognitive models of social anxiety disorder.Keywords: social anxiety disorder; social phobia; expectations; attributions; cognitive specificity

Received September 23, 2002; Accepted October 18, 2002

Correspondence address: Steven Taylor, Department of Psychiatry, University of British Columbia, 2255Wesbrook Mall, Vancouver, BC, V6T 2A1, Canada. E-mail: taylor@unixg.ubc.ca

Over the past 20 years there has been increasing interest in social anxiety disorder (SAXD).SAXD is characterized by marked and persistent fear of social or performance situations in whichembarrassment may occur (American Psychiatric Association [APA], 2000). Also known associal phobia, SAXD is a more appropriate term for this clinical disorder because it better conveysthe fact that social fears are typically pervasive and often debilitating (Liebowitz, Heimberg,Fresco, Travers, & Stein, 2000). SAXD is quite prevalent, with lifetime estimates ranging from 3to 13% (APA, 2000).

There are several psychological and biological models of SAXD. No single modelpredominates, although a growing body of research suggests that cognitive factors play animportant role (e.g. Amir, Foa, & Coles, 1998a,b; Stopa & Clark, 1993, 2000; Williams, Watts,MacLeod, & Mathews, 1997). There are several contemporary cognitive models of SAXD (e.g.Beck & Emery, 1985; Clark & Wells, 1995; Rapee & Heimberg, 1997), which have furtherclarified the variables underpinning socially-related anxiety. In turn, this has led to promisingtreatments (Fedoroff & Taylor, 2001).

Although the cognitive models differ in detail, they all emphasize the importance ofdysfunctional socially-based beliefs, acquired early in life, such as “My worth as a person dependson what others think of me,” “People tend to be critical of others,” and “I am sociallyincompetent.” These beliefs are said to give rise to expectations and attributions that the person

Cognitive Behaviour Therapy Vol 32, No 4, pp. 166–178, 2003

� 2003 Taylor & Francis ISSN 1650-6073DOI 10.1080/16506070310020315

Dow

nloa

ded

by [

McM

aste

r U

nive

rsity

] at

12:

53 2

9 O

ctob

er 2

014

holds about their interpersonal world, which in turn contribute to social anxiety. The cognitivemodels incorporate Beck’s (1976) cognitive specificity hypothesis, which proposes thatemotional disorders differ from one another in terms of their cognitive mechanisms. SAXD isthought to differ from other emotional disorders in that it is characterized by distinct types ofexpectations and attributions. The present study examined socially-related expectations becausemuch remains to be learned about their association with SAXD.

SAXD is thought to be associated with tendencies to expect that unpleasant social events (e.g.criticism by an employer) are likely and that pleasant social events (e.g. receiving approval fromothers) are unlikely. Social anxiety is also thought to be associated with specific types ofattributions in explaining causation of socially-related events (e.g. the tendency to blame oneselffor unpleasant social events and the tendency to see oneself as having little or no role in bringingabout pleasant social events). These types of attributions can perpetuate anxiety about one’sability to gain approval from others and thereby maintain social anxiety.

Only a handful of studies have examined expectations and attributions in SAXD. Lucock andSalkovskis (1988) found that compared with normal controls, people with SAXD tend tooverestimate the probability of unpleasant social events and underestimate the probability ofpleasant social and non-social events. The groups did not differ in their probability estimates ofnon-social negative events. When social anxiety declined after cognitive-behavioural therapy, sodid expectations for social but not non-social events. These results suggest that SAXD isassociated with specific types of expectations (e.g. those for social events) rather thanexpectations in general.

Foa, Franklin, Perry, and Herbert (1996) also compared a SAXD group with normal controlsand found that SAXD was associated with greater expectations for negative social events. Peoplewith SAXD, compared with controls, also overestimated the cost (“badness”) of negative socialevents. SAXD and controls did not differ in their expectations about the probability or cost of non-social negative events. Expectations for positive events were not assessed. When social anxiety inthe SAXD group was reduced via cognitive-behavioural therapy, the negative expectations alsodeclined. Similar findings were reported by Gilboa-Schechtman, Franklin and Foa (2000), whocompared SAXD with non-anxious controls and obsessive compulsive disorder (OCD). People inthe SAXD group, compared with those in the other groups, expected a lower occurrence ofpositive social events and anticipated a higher likelihood of negative social events.

Little is known about the role of attributions in SAXD. Depression research (e.g. Abramson,Seligman, & Teasdale, 1978) suggests that pessimistic attributions (in general) play a key role indepression. Teglasi and Fagin (1984) proposed that the attributions for social events might bemore specific to social anxiety. These investigators assessed college students classified as havinglow, medium, or high social anxiety. A measure of attributional style for negative social eventswas constructed to assess 3 dimensions of causality for events; the extent to which the causes wereinternal (due to the respondent, vs due to external factors); the extent to which the events were dueto stable (vs unstable) causes; and the extent to which the causes wereglobal in their effects (vsparticular to a given situation). Severity of social anxiety was correlated with stability andglobality, but not with internality. This finding suggests that at least some attributions for negativesocial events are associated with social anxiety. The authors did not assess attributions for positivesocial events and so it is not known whether these types of events are also related to social anxiety.

Further work is needed to investigate whether there are specific types of socially-relatedexpectations and attributions in SAXD. Little is known about how SAXD differs from otheremotional disorders in terms of the strength of expectations and attributions for social events. Noris it known whether the strength of these cognitive processes are specific to social anxiety, orwhether they are simply correlates of general distress. If the latter is the case then contemporarycognitive models would need to be revised to better explain why some people develop SAXDwhile other people develop other emotional disorders.

VOL 32, NO 4, 2003 Expectations and attributions in social anxiety 167

Dow

nloa

ded

by [

McM

aste

r U

nive

rsity

] at

12:

53 2

9 O

ctob

er 2

014

The present study had 3 complementary aims. The first was to explore the differences inexpectations and attributions between SAXD and other anxiety disorders. People with SAXDwere compared with people with either panic disorder and agoraphobia (PDA) or post-traumaticstress disorder (PTSD) to determine whether people with SAXD, as compared with the otherdiagnostic groups, would show higher scores for measures of pessimistic social attributions andexpectations and lower scores for measures of optimistic social attributions and expectations. Tofurther examine the specificity of these cognitive factors in social anxiety, non-social measureswere also included. PDA and PTSD were selected as comparison groups because both disorderstend to be associated with intense, non-specific distress (e.g. depression and general anxiety)(Taylor, 2000; Taylor et al., 2001). If the attributions and expectations associated with SAXD aresimply an index of the person’s global severity of psychopathology, then people with SAXDshould score no higher on measures of the strength of these socially-related cognitive variablesthan people with PDA or PTSD. On the other hand, if there are expectations and attributionsspecific to social anxiety, then people with SAXD should score higher than the other groups, evenwhen matched for general anxiety and depression.

The second aim was to conduct a dimensional (correlational) analysis of the relationshipbetween social anxiety and socially-related expectations and attributions. That is, by pooling datafrom the 3 diagnostic groups, to determine whether the strength of these socially-related cognitivevariables is correlated with social anxiety, after controlling for general anxiety and depression. Athird, related aim was to determine whether these cognitive factors were more strongly correlatedwith social anxiety than with general anxiety and depression using partial correlations. Such testscan reveal whether there are specific attributions and expectations in social anxiety.

MethodParticipants

Three diagnostic groups were recruited from 3 different treatment outcome studies:Generalized SAXD (n = 75; Taylor et al., 1997), PDA (n = 44; Sochting et al., 1998) and PTSD(n = 59; Taylor et al., 2001). Participants were assigned to each group on the basis of their primary(most severe) disorder, diagnosed according to DSM-IV criteria (APA, 2000). Demographiccharacteristics of the groups are described in Table 1. The table shows that the participants werepredominantly female and Caucasian, with a mean age in the mid-30s.

All participants were recruited in the same manner, primarily from advertisements in the localmedia, along with a smaller number of referrals from primary care physicians or mental healthpractitioners. Not all participants in the present study actually entered the outcome studies; somedropped out or were excluded beforehand (see the original studies for their inclusion andexclusion criteria). In the present study, the inclusion criteria were fluency in written and spokenEnglish, and provision of informed consent for their data to be used for research involving groupaverages and trends, without the identification of specific individuals. Exclusion criteria includedthe presence of psychotic, substance-use, or neurologic disorders. Participants on psychotropicmedication were accepted into the study so long as their doses were stable.

MeasuresDemographics

A short questionnaire was assembled to assess each participant’s demographic features andmedication status (type and dose of medication). The following demographic variables wereassessed: Age, gender, employment status, marital status, ethnicity and education level. For theanalyses reported later in this article, education level was scored on an 8-point scale(1 = completed less than grade 7; 2 = grades 7–12, without graduating high school; 3 = graduated

168 Taylor and Wald COGNITIVE BEHAVIOUR THERAPY

Dow

nloa

ded

by [

McM

aste

r U

nive

rsity

] at

12:

53 2

9 O

ctob

er 2

014

high school; 4 = partial college education; 5 = graduated from a 2-year college program;6 = graduated from a 4-year college program; 7 = partial graduate or professional training;8 = completed graduate or professional school).

Expectation measuresThe Subjective Probability Scale (Lucock & Salkovskis, 1988) was used to assess expectations

for social and non-social events. This measure consists of 4 6-item scales. For each item theparticipant is asked to estimate the likelihood of a commonplace hypothetical event happening tothem. Ratings are made on a 5-point scale, ranging from 1 (not at all likely) to 5 (extremelylikely). The scales and sample items are as follows: (1) Subjective probability of positive socialevents (e.g. “If you met someone at a party, you would become good friends”); (2) subjectiveprobability of negative social events (e.g. “If you asked a stranger for directions, you would beignored”); (3) subjective probability of positive non-social events (e.g. “When you next go onholiday the weather will be fine”); and (4) subjective probability of negative non-social events(e.g. “You will have a serious illness in the next 5 years). Higher scores on this measure representstronger expectancies.

Attribution measuresSocially-related attributions were measured by the Attributional Style Questionnaire for Social

Events (Taylor et al., 1997). This is an adaptation of the Attributional Style Questionnaire(Peterson et al., 1982), which is widely used in depression research. The adapted version differs

Table 1.Group comparisons on demographic and clinical variables.

Social AnxietyDisorder

(SAXD; n = 75)

Panic Disorderwith

Agoraphobia(PDA; n = 44)

Post-traumaticStress Disorder(PTSD;n = 59)

Omnibus test:Group comparison

Student-Newman-Keulscomparisons forsignificantF

M or % SD M or % SD M or % SD F or �2 df tests (p � 0.05)

% Female 55 76 71 6.47 2 –

% Unemployed 20 24 40 6.94 2 –

% Married orcohabiting

35 38 42 0.84 2 –

% Caucasian 79 94 80 3.76 2 –

% Takingpsychotropicmedication

7 55 34 29.91*** 2 –

Age (years) 36.3 9.7 33.2 7.8 36.4 9.6 1.40 2,162 –

Education level(8 point scale)

4.7 1.4 3.9 1.0 4.6 1.6 4.08 2,163 –

Social Phobia &AnxietyInventory

140.9 20.2 66.4 35.2 62.7 33.2 149.75*** 2,174 SAXD� PDA,PTSD

Beck AnxietyInventory

14.3 10.5 25.9 12.0 27.6 13.9 23.88*** 2,174 SAXD� PDA,PTSD

Beck DepressionInventory

12.4 6.7 15.8 8.4 25.6 11.6 36.21*** 2,174 SAXD� PDA,PTSD;PDA � PTSD

* p � 0.01, ** p � 0.005, *** p � 0.001.

VOL 32, NO 4, 2003 Expectations and attributions in social anxiety 169

Dow

nloa

ded

by [

McM

aste

r U

nive

rsity

] at

12:

53 2

9 O

ctob

er 2

014

from the original in that the 6 scenarios described in the questionnaire all specifically refer tosocial situations. Scenarios in the original questionnaire referred to a mix of social and non-socialsituations. Apart from this modification, the adaptive version is identical to the original. Theadaptation describes 3 positive social scenarios (e.g. “You are complimented by a friend”) and 3negative social scenarios (e.g. “You are reprimanded by your boss”). For each scenario therespondent is asked to write down the one major cause (e.g. “I was reprimanded because I did notfollow company policy”). The respondent then makes 3 ratings on 7-point scales about the originsof the cause they have written down: (1) the extent to which the cause is regarded as having aninternal vs external origin (ranging from 1 = totally due to other people or circumstances, to7 = totally due to me); (2) the stability of the cause (1 = will never again be present, 7 = willalways be present); and (3) the extent to which the cause is global vs specific in its effects onevents in the respondent’s life (1 = influences this particular situation, 7 = influences all situationsin my life). Thus, the Attributional Style Questionnaire for Social Events yields 6 scores,representing the extent to which the respondent makes internal, stable and global attributions forpositive and negative social events. For this measure, higher scores reflect stronger attributions.

Symptom severity measuresThe Social Phobia and Anxiety Inventory (SPAI: Turner, Beidel, Dancu, & Stanley, 1989) was

used to measure the severity of social anxiety. Following the recommendations of Turner andcolleagues, scores on this self-report scale were calculated by computing the total score for thesocial anxiety/phobia items and then subtracting this score from the total score of the SPAI’sagoraphobia scale. This method increases the ability of the SPAI to assess social anxiety, asdistinct from agoraphobia (Turner, Beidel, Dancu, & Stanley, 1989). Previous studies have shownthat the SPAI has good reliability and validity (Beidel, Turner, & Cooley, 1993; Beidel, Turner,Stanley, & Dancu, 1989; Herbert, Bellack, & Hope, 1991; Turner, Beidel, Dancu, & Stanley,1989; Turner, Stanley, Beidel, & Bond, 1989).

General anxiety – that is, anxiety arising from any of a variety of sources rather than beinglinked to a specific source such as social situations – was measured by the Beck Anxiety Inventory(Beck & Steer, 1990). Depression was assessed by the Beck Depression Inventory (Beck & Steer,1987). These self-report measures have been shown to have good reliability and validity (Beck &Steer, 1987, 1990; Beck, Steer, & Garbin, 1988).

Diagnostic interviewsDiagnoses were established by using a structured clinical interview. Participants in the SAXD

group were diagnosed by the Anxiety Disorders Interview Schedule-Revised (ADIS-R: DiNardo& Barlow, 1988). Although this instrument was designed for DSM-III-R, the diagnostic criteriafor SAXD are essentially unchanged in DSM-IV, and so the ADIS-R can be used to diagnoseDSM-IV SAXD. Participants in the PDA group were diagnosed by the Anxiety DisordersInterview for DSM-IV (DiNardo, Brown, & Barlow, 1994) and those in the PTSD group werediagnosed by the Structured Clinical Interview for DSM-IV (First, Spitzer, Gibbon, & Williams,1996).

ProcedureParticipants were assessed prior to treatment by a structured clinical interview followed by the

self-report questionnaires. Written informed consent was obtained beforehand. A trained researchstaff member administered the interviews, supervised by a doctoral level psychologist. Interviewswere audiotaped and independently reviewed by another doctoral-level psychologist to assessinter-rater reliability. This form of reliability, defined in terms of�, was assessed for the SAXDand PTSD samples and found to be “excellent” according to the criteria of Landis and Koch(1977); diagnosis of SAXD:� = 0.91, diagnosis of PTSD:� = 0.70 (Taylor et al., 1997, 2001).

170 Taylor and Wald COGNITIVE BEHAVIOUR THERAPY

Dow

nloa

ded

by [

McM

aste

r U

nive

rsity

] at

12:

53 2

9 O

ctob

er 2

014

Inter-rater reliability was not assessed for PDA in the So¨chting et al. (1998) study. However, suchreliability was assessed in our previous PDA treatment study, conducted in the same clinic usingthe same assessment procedures (Taylor et al., 1996). Inter-rater reliability in that study wasexcellent (� = 0.86).

ResultsWith regard to Type I and II errors, a stringent correction in Type I error (e.g. a Bonferroni

correction) increases the number of Type II errors. Some editorials have argued against the use ofBonferroni and similar corrections, because of their effects on Type II error (Perneger, 1998;Rothman, 1990). Given the large number of tests conducted in the present study, we sought tostrike a balance between Type I and II errors by setting the� level for all tests at 0.01 (all tests 2tailed), with trends identified asp � 0.05. The exception was for error-protected tests (e.g.Student-Newman-Keuls post-hoc comparisons), which were conducted at� = 0.05 if theirpreceding ominbus test was significant at� = 0.01. In most cases the significant results reported inthis article were associated withp values�0.001, suggesting that the pattern of significantfindings would remain largely unchanged if another approach to constraining Type I and II errorshad been adopted.

Preliminary analysesTable 1 shows that the diagnostic groups did not differ in terms of demographic variables,

although there were significant differences in the proportion of people from each group takingpsychotropic medication. (Degrees of freedom for the F tests reported in this and the followingtables vary because of incomplete data.). Given these results, subsequent analyses reported in thisarticle explore the possible effects of medication on expectations and attributions.

Table 1 also shows that SAXD, as compared with PDA and PTSD, was characterized bysignificantly higher scores on the Social Phobia and Anxiety Inventory. This is consistent with ourgoals of classifying participants according to their primary (most severe) disorder. Some PDA andPTSD participants had comorbid SAXD, although the latter was mild compared with PDA andPTSD (detailed comorbidity data was not available). The table also shows that SAXD comparedwith the other groups tended to be associated with milder depression and general anxiety. Thus,SAXD reflected more severe social anxiety rather than severe general distress. Accordingly,subsequent analyses, reported later in this article, controlled for group differences in depressionand general anxiety (as well as controlling for medication use).

Expectations and attributions across diagnostic groupsTable 2 shows the group comparisons on the measures of expectations and attributions when

medication, general anxiety and depression were used as covariates. Note that higher scoresindicate greater expectancies or stronger attributions. Compared with the other groups, SAXDwas associated with lower expectations for the occurrence of positive social events and higherexpectations for the occurrence of negative social events. The groups did not significantly differ interms of expectancies for non-social events. The table also shows that SAXD, compared withPDA and PTSD, was associated with stronger stable and global attributions for negative socialevents, and weaker internal attributions (e.g. more external attributions) for positive social events.The groups did not significantly differ in the strength of internal and global attributions fornegative social events, or in the strength of stable attributions for positive social events. However,the group differences on these variables do suggest trends (e.g. trend toward stronger internalattributions for negative events in SAXD, compared with PDA and PTSD). In summary, theseresults provide some evidence of disorder-specific differences in social expectations andattributions when medication, general anxiety and depression are controlled.

VOL 32, NO 4, 2003 Expectations and attributions in social anxiety 171

Dow

nloa

ded

by [

McM

aste

r U

nive

rsity

] at

12:

53 2

9 O

ctob

er 2

014

Correlational analysesA complementary means of examining the relationship between social anxiety and these

cognitive variables is to conduct correlational analyses. This approach can be more powerful thangroup comparisons because it takes into consideration the fact that people within the SAXD groupdiffer from one another in the severity of social anxiety, and some people in the other diagnosticgroups also had elevated social anxiety. Between-group comparisons are compromised to somedegree by these within-group variations. Correlational analyses, pooling data from all 3 groups, donot suffer from this problem. However, a concern with such an approach is that artifacts (spuriouscorrelations) can be introduced when groups are aggregated, particularly when the groups greatlydiffer from one another in their scores on one or more variables. To explore the possibility of such

Table 2.Group comparisons on cognitive variables, controlling for general anxiety, depression, and medicationstatus.

SocialAnxietyDisorder(SAXD)

Panic Disorderwith

Agoraphobia(PDA)

Post-traumatic

StressDisorder(PTSD)

Omnibus test:Group

comparison(ANCOVA)

Significantcovariate-adjustedpairwisecomparisons(p � 0.05),computed forsignificant omnibus

Madj SD Madj SD Madj SD F df tests

Subjective Probability Scale

Positive non-socialevents

18.1 3.5 18.8 3.1 18.4 4.0 0.30 2,152

Negative non-socialevents

15.5 3.3 16.0 4.5 17.4 4.9 2.28 2,152

Positive socialevents

15.0 3.4 19.6 4.0 21.1 4.4 33.69*** 2,152 SAXD� PDA,PTSD

Negative socialevents

16.7 3.0 13.1 3.4 12.9 5.0 17.15*** 2,152 SAXD�PDA,PTSD

Attributional Style Questionnaire for Social Events

Internal attributionfor positiveevents

15.7 2.8 17.1 2.4 17.9 2.9 7.58** 2,145 SAXD� PTSD

Internal attributionfor negativeevents

17.6 2.7 16.0 3.1 15.9 3.5 3.77 2,145

Stable attributionfor positiveevents

15.8 2.9 16.9 9.7 16.9 2.3 2.27 2,145

Stable attributionfor negativeevents

16.3 3.0 14.3 3.4 13.0 3.8 10.02*** 2,145 SAXD� PDA,PTSD

Global attributionfor positiveevents

13.3 4.1 15.7 4.0 15.0 4.7 3.00 2,145

Global attributionfor negativeevents

16.3 3.1 12.3 3.9 11.7 4.8 18.10*** 2,145 SAXD� PDA,PTSD

Madj = covariate-adjusted mean. * p � 0.01, ** p � 0.005, *** p � 0.001.

172 Taylor and Wald COGNITIVE BEHAVIOUR THERAPY

Dow

nloa

ded

by [

McM

aste

r U

nive

rsity

] at

12:

53 2

9 O

ctob

er 2

014

an artifact, the correlations among the variables were initially calculated based on simply poolingdata from the 3 groups. The pooled correlations were then compared with averaged within-groupcorrelations (not reported). The latter are calculated by: (1) computing correlations separately foreach diagnostic group; (2) transforming thers into Fisher’sZ; (3) computing, for each pair ofvariables, the weighted meanZ across the 3 groups (weighted according to sample size); and then(4) converting eachZ back tor. The matrix of correlations for the pooled data was then subtractedfrom matrix of averaged within-group correlations. The absolute value of the difference betweeneach corresponding correlation was then calculated. The pooled and averaged within-groupcorrelations tended to be quite similar. Accordingly, correlations based on the simpler (pooled)method are reported in this article and are shown in Table 3.

Based on the pooled correlation calculations, the table shows that social anxiety, as assessed bythe Social Phobia and Anxiety Inventory, was significantly correlated with expectancies forpositive and negative social events but not with expectancies for non-social events. Social anxietywas also significantly correlated with all of the attribution measures, except for stable attributionsfor positive events. In contrast, general anxiety and depression were correlated with expectanciesfor non-social positive and non-social negative events, and were not correlated with any of thesocial attribution measures. General anxiety and depression were correlated with expectations forsocial negative events, but not for social positive events.

The measures of socially-related expectancies and attributions were generally unrelated todemographic variables and medication status, with no systematic patterns of significance (resultsavailable on request). This suggests that the correlations between these cognitive variables andsocial anxiety were not due to some confounding variable such as medication status or educationlevel.

Table 3 also shows the tests of the significance between pairs of correlations, using theprocedure described by Meng, Rosenthal, and Rubin (1992). This table also shows trends(0.01� p � 0.05) to facilitate the interpretation of the results. The table shows that the measuresof these socially-related cognitive variables tended to be more strongly correlated with social

Table 3.Correlations and tests of significance of differences between correlations.

SPAI vsBAI:

SPAI vsBDI:

rSPAI rBAI rBDI ZDIFF ZDIFF

Subjective Probability Scale

Positive non-social events 0.01 �0.22*** �0.31*** 2.03† 3.03**

Negative non-social events �0.19 0.39*** 0.45*** 4.88*** 5.61***

Positive social events �0.54*** 0.00 �0.16 5.06*** 3.81***

Negative social events 0.34*** 0.23*** 0.38*** 0.89 0.52

Attributional Style Questionnaire for Social Events

Internal attribution for positive events �0.25*** 0.00 �0.10 2.16† 1.36

Internal attribution for negative events 0.35*** �0.06 �0.04 3.23** 3.12**

Stable attribution for positive events �0.11 0.15 0.08 2.11† 1.62

Stable attribution for negative events 0.33***�0.05 0.02 3.04** 2.59*

Global attribution for positive events �0.26*** 0.05 0.01 2.61* 2.34†

Global attribution for negative events 0.37*** 0.08 0.13 2.38† 2.01†

ns range from 169 to 177. BAI = Beck Anxiety Inventory, BDI = Beck Depression Inventory, SPAI = Social Phobia & AnxietyInventory, rSPAI = correlation with SPAI, rBAI = correlation with BAI, rBDI = correlation with BDI.† p� 0.05, * p � 0.01, ** p � 0.005, *** p � 0.001.

VOL 32, NO 4, 2003 Expectations and attributions in social anxiety 173

Dow

nloa

ded

by [

McM

aste

r U

nive

rsity

] at

12:

53 2

9 O

ctob

er 2

014

anxiety than with general anxiety and depression. Conversely, the measures of non-socialcognitive measures tended to be more strongly correlated with general anxiety and depressionthan with social anxiety. Many of the differences were trends (p � 0.05); not all reachedsignificance (defined as� = 0.01).

Finally, Table 4 shows the partial correlations of the cognitive measures. Generally consistentwith the previous analyses, here it can be seen that these socially-related expectations andattributions were significantly correlated with social anxiety, even after partialling out the effectsof general anxiety and depression. Such expectations and attributions were largely uncorrelatedwith general anxiety and depression once the effects of social anxiety were partialled out.Conversely, partial correlations for the non-social expectations and attributions were significantonly for depression. These results further suggest some degree of specificity in the relationshipbetween social anxiety and these cognitive variables.

DiscussionExpectations and attributions are thought to be important cognitive factors in the etiology and

maintenance of SAXD. Research to date suggests that some types of expectations and attributionsare more closely related to SAXD than other anxiety disorders (Foa et al., 1996; Lucock &Salkovskis, 1988). To date, no other study has directly compared expectations and attributionsbetween SAXD, PDA and PTSD. Furthermore the relationship of potential disorder-specificcognitive variables with different types of events (social vs non-social, positive vs negative), andgeneral anxiety and depression has not been previously investigated in detail. Continuedinvestigation of cognitive mechanisms and their relationships to other variables has importantimplications for understanding the etiological underpinnings of SAXD and for developing moreeffective treatments.

To address these issues, this study had 3 complementary aims. The first objective of this studywas to investigate the differences in expectations and attributions between SAXD and 2 otheranxiety disorders (PDA and PTSD) after controlling for general distress (anxiety and depression)

Table 4.Partial correlations of cognitive variables with social anxiety, general anxiety and depression.

Correlation withSPAI, controlling forBAI and BDI

Correlation withBAI, controlling forSPAI and BDI

Correlation withBDI, controllingfor SPAI and BAI

Subjective Probability Scale

Positive non-social events �0.04 �0.06 �0.23**

Negative non-social events �0.08 0.13 0.30***

Positive social events �0.58*** 0.06 �0.28***

Negative social events 0.43*** 0.12 0.35***

Attributional Style Questionnaire for Social Events

Internal attribution for positive events �0.25*** 0.04 �0.15

Internal attribution for negative events 0.33*** 0.01 0.01

Stable attribution for positive events �0.06 0.12 �0.02

Stable attribution for negative events 0.33*** �0.03 0.08

Global attribution for positive events �0.24** 0.02 �0.04

Global attribution for negative events 0.42*** 0.11 0.12

n = 163. BAI = Beck Anxiety Inventory, BDI = Beck Depression Inventory, SPAI = Social Phobia & Anxiety Inventory.* p � 0.01, ** p � 0.005, *** p � 0.001.

174 Taylor and Wald COGNITIVE BEHAVIOUR THERAPY

Dow

nloa

ded

by [

McM

aste

r U

nive

rsity

] at

12:

53 2

9 O

ctob

er 2

014

and medication status. Group differences were examined for positive and negative events (bothsocial and non-social). A second goal was to investigate the relationships between social anxietyand these cognitive variables. Lastly, the study examined whether these cognitive factors weremore strongly correlated with social anxiety than with general anxiety and depression.

Consistent with Beck’s (1976) cognitive specificity hypothesis, the results of this studyrevealed several notable differences in attributions and expectations across the 3 diagnosticgroups. SAXD, compared with the PDA and PTSD, was associated with lower expectations forthe occurrence of positive social events and higher expectations for the occurrence of negativesocial events when general anxiety and depression are controlled. The groups did not significantlydiffer in terms of expectancies for non-social events (positive or negative). These findings areconsistent with prior research (e.g. Foa, et al., 1996; Lucock & Salkovskis, 1988), and furtherdelineate disorder-specific cognitive mechanisms in SAXD and the role of general distress amongthese variables.

With regard to attributions, Teglasi and Fagin (1994) found that the severity of social anxiety incollege students was associated with the stability and globality, but not internality for negativesocial events. Similarly, in the present study, SAXD, compared with the other groups, wasassociated with stronger stable and global attributions for negative social events, but did notsignificantly differ in the strength of internal attributions for negative social events (aftercontrolling for medication, general anxiety and depression). Although the differences on internalattributions for negative social events did not reach statistical significance, the results did suggesta trend towards stronger internal attributions for negative events in SAXD, compared with PDAand PTSD. Thus, people with SAXD have a greater tendency towards attributing negativeoutcomes to dispositional factors (e.g. tendency to blame oneself for negative social experiences)rather than to external factors (e.g. bad luck, task difficulty). In a recent study that appeared in theliterature after our study had been completed, Coles, Turk, Heimberg, and Fresco (2001) foundthat SAXD, compared with normal controls, was associated with more internal, stable and globalattributions as the anxiety levels in a situation increased. These results suggest that attributions fornegative social events in SAXD may be moderated by emotional and contextual factors, andfurther research is needed to clarify these relationships.

Less research has examined attributions for positive social events in SAXD. Results of thisstudy found weaker internal attributions (e.g. more external attributions) for positive social eventsin SAXD, compared with other disorders. This pessimistic attributional style suggests that peoplewith SAXD are more likely to explain the occurrence of positive social experiences by externalreasons, rather than by their own abilities or characteristics. This finding is generally consistentwith the general literature (e.g. Arkin, Appelman, & Burger, 1980; Clark & Wells, 1995; Stopa &Clark, 1993) that has shown that people with SAXD tend to negatively evaluate themselves insocial situations (e.g. “People won’t like me”). The strength of stable and global attributions forpositive social events was not significantly different across the groups. This finding indicates thatthese attributions for positive social events may be less salient in SAXD, but may play a morespecific role for negative social events.

The correlational analyses were used to identify specific relationships between social anxietyand these cognitive variables. These results suggest some degree of specificity in the correlates ofthese measures; the strength of these variables for social events was correlated with social anxietybut not with general anxiety, depression, or most other variables. Conversely, strength of thesecognitive variables for non-social events was significantly correlated with general anxiety anddepression, but not with social anxiety or most other variables. Results from the partialcorrelations of the cognitive measures provided further evidence of the specificity in therelationship between social anxiety and the cognitive measures after controlling for the effects ofgeneral anxiety and depression.

These results are consistent with contemporary cognitive models of social anxiety (e.g. Beck &

VOL 32, NO 4, 2003 Expectations and attributions in social anxiety 175

Dow

nloa

ded

by [

McM

aste

r U

nive

rsity

] at

12:

53 2

9 O

ctob

er 2

014

Emery, 1985; Clark & Wells, 1995; Rapee & Heimberg, 1997), which emphasize the role of suchvariables in the etiology and maintenance of SAXD, as well in its treatment. Previous studiesshow that when social anxiety is reduced by cognitive or cognitive-behavioural therapy, socially-related expectations and attributions similarly decline (Foa et al., 1996; Lucock & Salkovskis,1988; Taylor et al., 1997). In contrast, when social anxiety does not change (during the course ofplacebo treatment), expectations and attributions also do not change (Taylor et al., 1997).

A limitation of this study and a limitation of previous studies of socially-related expectationsand attributions (Foa et al., 1996; Gilboa-Schechtman et al., 2000; Lucock & Salkovskis, 1988;Teglasi, & Fagin, 1984), is that little is known about the reliability and validity of the measuresused in this research. Construction of these scales was based primarily on face validity. Althoughthe pattern of results in these studies and those of the present study are what one would expectfrom reliable and valid scales (e.g. significant correlations with social anxiety), the psychometricproperties of these instruments remain to be further examined.

Another issue that merits consideration is whether there is a circularity in the research on therole of cognitive variables in SAXD. In order to be diagnosed with this disorder, the person musthave “a marked and persistent fear of one or more social or performance situations … Theindividual fears that he or she will act in a way (or show anxiety symptoms) that will behumiliating or embarrassing” (APA, 2000, p. 456). It would be circular and trivial to demonstratethat people with this disorder score higher on scales measuring fear of acting in an embarrassingmanner. The present study does not do this; we assessed expectations about the likelihood offuture positive and negative events, as well as the extent to which people make internal, stable andglobal attributions for positive and negative events. None of these variables are built into thediagnostic definition of SAXD and so there is no circularity. The aim of this study was to identifythe cognitive variables that may (or may not) underlie social fears. Reduction in socially-relatedexpectations and attributions is one possible pathway for reducing social anxiety, although theremay be others (e.g. changes in socially-related self-efficacy). Given the encouraging findings inthis study and others, the next step would be to conduct experimental studies that manipulatespecific types of socially-related expectations and attributions (vs non-social expectations andattributions) to assess the effects on social anxiety.

ReferencesAbramson, L. Y., Seligman, M. E. P., & Teasdale, J. D. (1978). Learned helplessness in humans: critique

and reformulation.Journal of Abnormal Psychology, 87, 49–74.American Psychiatric Association (2000).Diagnostic and Statistical Manual of Mental Disorders (4th edn,

text revision). Washington, DC: American Psychiatric Association.Amir, N., Foa, E. B., & Coles, M. E. (1998a). Automatic activation and strategic avoidance of threat-

relevant information in social phobia.Journal of Abnormal Psychology, 107, 285–290.Amir, N., Foa, E. B., & Coles, M. E. (1998b). Negative interpretation bias in social phobia.Behaviour

Research and Therapy, 36, 945–957.Arkin, R. M., Appelman, A. J., & Burger, J. M. (1980). Social anxiety, self-presentation, and self-serving

bias in causal attribution.Journal of Personality and Social Psychology, 36, 56–71.Beck, A. T. (1976).Cognitive Therapy and the Emotional Disorders. New York: International Universities

Press.Beck, A. T., & Emery, G. (1985).Anxiety Disorders and Phobias: A Cognitive Perspective. New York:

Basic Books.Beck, A. T., & Steer, R. A. (1987).Manual for the Beck Depression Inventory. San Antonio, TX:

Psychological Corporation.Beck, A. T., & Steer, R. A. (1990).Manual for the Beck Anxiety Inventory. San Antonio, TX: Psychological

Corporation.Beck, A. T., Steer, R. A., & Garbin, M. G. (1988). Psychometric properties of the Beck Depression

Inventory: Twenty-five years of evaluation.Clinical Psychology Review, 8, 77–100.Beidel, D. C., Turner, S. M., & Cooley, M. R. (1993). Assessing reliable and clinically significant change in

176 Taylor and Wald COGNITIVE BEHAVIOUR THERAPY

Dow

nloa

ded

by [

McM

aste

r U

nive

rsity

] at

12:

53 2

9 O

ctob

er 2

014

social phobia: validity of the Social Phobia and Anxiety Inventory.Behaviour Research and Therapy,31, 331–337.

Beidel, D. C., Turner, S. M., Stanley, M. A., & Dancu, C. V. (1989). The Social Phobia and AnxietyInventory: concurrent and external validity.Behavior Therapy, 20, 417–427.

Clark, D. M., & Wells, A. (1995). A cognitive model of social phobia. In R. G. Heimberg, M. R. Liebowitz,D. A. Hope, & F. R. Schneier (eds),Social Phobia: Diagnosis, Assessment, and Treatment (pp. 69–93).New York: Guilford.

Coles, M. E., Turk, C. L., Heimberg, R. G., & Fresco, D. M. (2001). Effects of varying levels of anxietywithin social situations: relationship to memory perspective and attributions in social phobia.BehaviourResearch and Therapy, 39, 651–665.

DiNardo, P. A., & Barlow, D. H. (1988).Anxiety Disorders Interview Schedule-Revised. New York:Graywind.

DiNardo, P. A., Brown, T. A., & Barlow, D. H. (1994).Anxiety Disorders Interview Schedule for DSM-IV.New York: Graywind.

Fedoroff, I. C., & Taylor, S. (2001). Psychological and pharmacological treatments for social phobia: Ameta-analysis.Journal of Clinical Psychopharmacology, 21, 311–324.

First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B. W. (1996).Structured Clinical Interview for AxesI and II DSM-IV disorders – Patient Edition. New York: Biometrics Research Department, New YorkState Psychiatric Institute.

Foa, E. B., Franklin, M. E., Perry, K. J., & Herbert, J. D. (1996). Cognitive biases in generalized socialphobia.Journal of Consulting and Clinical Psychology, 105, 433–439.

Gilboa-Schechtman, E., Franklin, M. E., & Foa, E. B. (2000). Anticipated reactions to social events:differences among individuals with generalized social phobia, obsessive compulsive disorder, andnonanxious controls.Cognitive Therapy and Research, 24, 731–746.

Herbert, J. D., Bellack, A. S., & Hope, D. A. (1991). Concurrent validity of the Social Phobia and AnxietyInventory.Journal of Psychopathology and Behavioral Assessment, 13, 357–368.

Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement of categorical data.Biometrics, 33, 159–174.

Liebowitz, M. R., Heimberg, R. G., Fresco, D. M., Travers, J., & Stein, M. B. (2000). Social phobia or socialanxiety disorder: what’s in a name?Archives of General Psychiatry, 57, 191–192.

Lucock, M. P., & Salkovskis, P. M. (1988). Cognitive factors in social anxiety and its treatment.BehaviourResearch and Therapy, 26, 297–302.

Meng, X.-L., Rosenthal, R., & Rubin, D. B. (1992). Comparing correlated correlation coefficients.Psychological Bulletin, 111, 172–175.

Perneger, T. V. (1998). What’s wrong with Bonferroni adjustments.British Medical Journal, 316, 1236–1238.

Peterson, C., Semmel, A., von Baeyer, C., Abramson, L. Y., Metalsky, G. I., & Seligman, M. E. P. (1982).The Attributional Style Questionnaire.Cognitive Therapy and Research, 6, 287–300.

Rapee, R. M., & Heimberg, R. G. (1997). A cognitive-behavioral model of anxiety in social phobia.Behaviour Research and Therapy, 35, 741–756.

Rothman, K. (1990). No adjustments are needed for multiple comparisons.Epidemiology, 1, 43–46.Sochting, I., Taylor, S., Freeman, W., de Koning, E., Segerstrom, S., & Thordarson, D. (1998). In vivo

exposure for panic disorder and agoraphobia: does a cognitive rationale enhance treatment efficacy? InE. Sanavio (Ed.),Behavioural and Cognitive Therapy Today: Essays in Honor of Hans J. Eysenck (pp.293–302). Amsterdam: Elsevier.

Stopa, L., & Clark, D. M. (1993). Cognitive processes in social phobia.Behaviour Research and Therapy,31, 255–267.

Stopa, L., & Clark, D. M. (2000). Social phobia and interpretation of social events.Behaviour Research andTherapy, 38, 273–283.

Taylor, S. (2000).Understanding and Treating Panic Disorder: Cognitive-Behavioural Approaches. NewYork: Wiley.

Taylor, S., Fedoroff, I. C., Koch, W. J., Thordarson, D. S., Fecteau, G., & Nicki, R. (2001). Posttraumaticstress disorder arising after road traffic collisions: Patterns of response to cognitive-behaviour therapy.Journal of Consulting and Clinical Psychology, 69, 541–551.

Taylor, S., Woody, S., Koch, W. J., McLean, P., & Anderson, K. (1996). Suffocation false-alarms andefficacy of cognitive-behavioural therapy for panic disorder.Behavior Therapy, 27, 115–126.

Taylor, S., Woody, S., Koch, W. J., McLean, P., Paterson, R., & Anderson, K. (1997). Cognitive restruc-turing in the treatment of social phobia: efficacy and mode of action.Behavior Modification, 21, 487–511.

VOL 32, NO 4, 2003 Expectations and attributions in social anxiety 177

Dow

nloa

ded

by [

McM

aste

r U

nive

rsity

] at

12:

53 2

9 O

ctob

er 2

014

Teglasi, H., & Fagin, S. S. (1984). Social anxiety and self-other biases in causal attribution.Journal ofResearch in Personality, 18, 64–80.

Turner, S. M., Beidel, D. C., Dancu, C. V., & Stanley, M. A. (1989). An empirically derived inventory tomeasure social fears and anxiety: the Social Phobia and Anxiety Inventory.Psychological Assessment,1, 35–40.

Turner, S. M., Stanley, M. A., Beidel, D. C., & Bond, L. (1989). The Social Phobia and Anxiety Inventory:construct validity.Journal of Psychopathology and Behavioral Assessment, 11, 221–234.

Williams, J. M. G., Watts, F. N., MacLeod, C., & Mathews, A. (1997).Cognitive psychology and emotionaldisorders (2nd edn). New York: Wiley.

178 Taylor and Wald COGNITIVE BEHAVIOUR THERAPY

Dow

nloa

ded

by [

McM

aste

r U

nive

rsity

] at

12:

53 2

9 O

ctob

er 2

014