The Negative Self-Portrayal Scale- Development, Validation, And Application to Social Anxiety

Available online at www.sciencedirect.com

Behavior Therapy 42 (2011) 183–196www.elsevier.com/locate/bt

The Negative Self-Portrayal Scale: Development, Validation,and Application to Social Anxiety

David A. MoscovitchVanessa Huyder

University of Waterloo

The Negative Self-Portrayal Scale (NSPS) is a new ques-tionnaire designed to assess the extent to which individualsare concerned that specific self-attributes they view as beingdeficient will be exposed to scrutiny and evaluation bycritical others in social situations. These concerns have beenproposed to drive symptoms of social anxiety and accountfor individual differences in social fears and avoidancebehaviors (Moscovitch, 2009). Here, we introduce the NSPSand examine its factor structure and psychometric proper-ties across two large samples of North American under-graduate students with normally distributed symptoms ofsocial anxiety. Exploratory and confirmatory factor analy-ses supported a 3-factor solution representing concernsabout (a) social competence; (b) physical appearance; and(c) signs of anxiety. The NSPS was found to have goodinternal consistency and test-retest reliability, strong con-vergent validity, and adequate discriminant validity. Inaddition, NSPS total scores accounted for a significantproportion of unique variance in self-concealment (i.e.,safety) behaviors over and above established symptommeasures of social interaction anxiety, social performanceanxiety, and depression. Results are discussed in relation totheoretical models of social anxiety and the potential utilityof the NSPS for both clinical research and practice.

This research was supported, in part, thanks to funding from theCanada Research Chairs Program to David A. Moscovitch. We aregrateful to Dr. Erik Woody for his helpful suggestions andassistance with the statistical analyses.

Address correspondence to David A. Moscovitch, Ph.D., CanadaResearch Chair in Mental Health Research, Department ofPsychology, University of Waterloo, 200 University Ave. WestWaterloo, Ontario, CanadaN2L 3G1; e-mail: [email protected]/10/183–196/$1.00/0© 2010 Association for Behavioral and Cognitive Therapies. Published byElsevier Ltd. All rights reserved.

THERE IS A GROWING consensus that the perceptionof self plays a crucial role in the pathogenesis andpersistence of social anxiety (cf. Stopa, 2009a).Although specific conceptualizations of the selfdiffer across contemporary cognitive and inter-personal models of social anxiety (Alden &Taylor, 2004; Clark & Wells, 1995; Hofmann,2007; Leary & Kowalski, 1995; Rapee &Heimberg, 1997), there is now strong empiricalsupport for the view that symptoms of socialanxiety arise from the discrepancy produced bythe motivation to convey a desired social impres-sion of oneself to others in combination with theexpectation that one will fail to do so (e.g.,Schlenker & Leary, 1985). Indeed, numerousstudies have shown that highly socially anxiousindividuals significantly underestimate their socialperformance abilities as well as other salientsocial self-attributes, and routinely provide self-ratings that fall well below their perception of thestandard they believe is required by relevant“audience” evaluators (e.g., for reviews of thisliterature, see Moscovitch & Hofmann, 2007;Orr & Moscovitch, 2010a; Moscovitch, Orr,Rowa, Gehring Reimer, & Antony, 2009).Under social threat, highly socially anxious and

phobic individuals allocate increased attentionalresources toward monitoring their internal anx-iety cues and other perceived negative features ofthe self (Rapee & Heimberg, 1997; Spurr &Stopa, 2002). This increase in self-monitoringcoincides with the experience of spontaneous,recurrent, intrusive, and often vivid negativemental self-images that are viewed from anobserver's perspective (Hackmann, Clark, &McManus, 2000; Hackmann, Surawy, & Clark,1998). These images tend to be negatively

http://dx.doi.org/

http://dx.doi.org/

http://dx.doi.org/

mailto:[email protected]

184 moscov i tch & huyder

distorted and exaggerated representations of theself; nevertheless, they are perceived and pro-cessed by socially anxious individuals as beingaccurate. As a result, these images captureprecious emotional and cognitive resources andpossibly prevent the encoding of salient positivesocial and interpersonal cues in the externalenvironment (Clark & Wells, 1995).On the basis of clinical observations and the

burgeoning empirical literature on the central roleof the self in social anxiety, Moscovitch (2009)recently recommended that scientist-practitionersengaged in assessing and treating patients withsocial anxiety disorder (SAD) within an exposure-based cognitive behavioral therapy (CBT) frame-work shift their focus in treatment from targetingpatients' feared social situations, to targeting moredirectly the core feared stimuli in SAD. Moscovitchproposed that the feared stimuli in SAD are specificself-attributes that socially anxious individualsperceive as being flawed or contrary to perceivedsociocultural norms, hypothesizing that it might bepossible to organize these self-attributes into broadthemes that fall across four nonorthogonal dimen-sions: (a) concerns about social skills and beha-viors; (b) concerns about visible signs of anxiety;(c) concerns about physical appearance; and (d)concerns about personality. While research onSAD has historically focused on socially anxiousindividuals' concerns about social skills/behaviorsand signs of anxiety (see Moscovitch, 2009, for areview of this literature), investigators have onlyrecently begun to examine the link between socialanxiety and heightened concerns about physicalappearance (Hart, Flora, Palyo, Fresco, Holle, &Heimberg, 2008; Izgiç, Akyüz, Dogan, & Kugu,2004; Rapee & Abbott, 2006) and personalitycharacteristics (Mansell & Clark, 1999; Wilson &Rapee, 2006).Moscovitch's (2009) model offers a potentially

useful, theory-guided heuristic for conceptualizingsymptom variations and individual differences insocial anxiety. Socially anxious people are hetero-geneous with respect to the kinds of socialsituations they fear and avoid (see Hofmann,Heinrichs, & Moscovitch, 2004 for a review ofthis literature), as well as the types of subtleavoidance or safety behaviors they tend to use(e.g., McManus, Sacadura, & Clark, 2008).According to Moscovitch (2009), variations insocial fears and safety behaviors are functionallyrelated to underlying differences in self-attributeconcerns across the proposed dimensions. Specif-ically, social situations that are endorsed as beinganxiety-provoking are those which individualsperceive as having the capacity to expose their

perceived self-deficiencies for public consumption.In the same vein, subtle avoidance and safetybehaviors represent attempts to conceal or preventthe public exposure of those self-attributes. Thus,core concerns about self-attribute flaws areproposed to drive the constellation of socialanxiety symptoms—subjective distress, avoidance,and self-concealment—that are associated withhigh levels of functional impairment in theemotional, vocational, and interpersonal lives ofindividuals with SAD (Ledley & Heimberg,2005).Although numerous psychometrically validated

measures of social anxiety are now widelyavailable, none specifically assesses self-portrayalconcerns across the proposed dimensions. Most ofthe excellent social anxiety self-report measurescurrently in circulation (e.g., LSAS-SR, SIAS, SPS,SPIN, SPAI, etc.; see Antony, Orsillo, & Roemer,2001) assess the level or type of social anxietysymptoms that individuals may experience acrossa variety of social performance and interactionsituations. Our objective in designing the NegativeSelf-Portrayal Scale (NSPS) was not to addanother questionnaire to this list. Rather, wewished to create a novel measure that directlyassessed the specific types of self-attributes thatMoscovitch (2009) proposed would elicit concernsabout self-presentation for socially anxious indi-viduals in anxiety-provoking social situations.Being able to identify and measure such concernsmight conceivably both enhance our understand-ing of symptom heterogeneity in social anxietyand guide case conceptualization of SAD inclinical practice. This may, in turn, help facilitatethe delivery of CBT for SAD in a more individ-ually customized manner, thereby having thepotential to enhance SAD treatment outcomes.Here, we examine the factor structure and

psychometric properties of the NSPS in two largesamples of individuals with social anxiety symptomsdistributed across the normal spectrum and begin toinvestigate some of the central, but previouslyuntested, theoretical claims of Moscovitch's (2009)model. Despite our theoretical hypotheses aboutthe factor structure of the NSPS across theproposed dimensions, no previous studies havetested these predictions empirically. Thus, we firstpresent an exploratory factor analysis (EFA) of theNSPS in participant Sample 1, as recommended inthe early stages of scale development (e.g., Hurleyet al., 1997). Next, based on the results fromSample 1, we follow up the EFA in Sample 2 withconfirmatory factor analysis (CFA) of the derivedmodel and plausible competing models (Brown,2006).

185negat i v e s e l f - portrayal scale

Methodparticipants

Two groups of nonoverlapping undergraduateparticipants (n=225; n=316) at a large, urbanCanadian university completed a series of onlinequestionnaires in exchange for partial course credit.

Initial Item GenerationAn initial pool of 62 NSPS items was constructed bythe authors. These were then sent via electronic mailto three independent Ph.D.-level scientist-practi-tioners with expertise in social anxiety research andextensive experience treating individualswith SAD inclinical settings. Assessors were given a randomizedlist of the preliminary items alongwith an instructionsheet defining the four hypothesized (Moscovitch,2009) dimensions of self-attribute concerns (i.e.,concerns about social skills and behaviors; concernsabout visible signs of anxiety; concerns aboutphysical appearance; and concerns about personal-ity). Assessors were instructed to eliminate any itemsthat seemed redundant or unclear, add items theythought were appropriate, and indicate the dimen-sion in which they thought each self-attribute itembelonged, according to the definitions given to them.Any items eliminated or misallocated by more thanone assessor were removed from the respectiveinventory. Any new items that were suggested by atleast one assessor were added to the NSPS item pool,provided they were face valid and nonredundantwith other items already on the list. Following thisprocess, 11 items were removed and 6 items wereadded, resulting in a total of 57 NSPS items.

1 A copy of the SCBI is available upon request from the firstauthor.

measures

Negative Self-Portrayal Scale (NSPS)Participants were administered the 57-item NSPS.Instructions to respondents asked them to rate thedegree to which they are concerned about certainaspects of themselves when in anxiety-provokingsocial situations. Response options per item rangedfrom 1 to 5, with descriptive anchors listed as “not atall concerned” to “extremely concerned.” Therewere no reverse-scored items; higher scores repre-sented greater self-portrayal concerns.

Self-Concealment Behaviors Inventory (SCBI)The SCBI is a self-report measure of self-concealment(i.e., safety) behaviors. This measure is essentially anexpanded version of the unpublished Social Behav-iour Questionnaire (SBQ; Clark, Butler, Fennell,Hackmann, McManus, & Wells, 1995; see alsoMcManus et al., 2009). Given that the SBQ is anunpublished scale with unknown psychometricproperties, we chose to add items deemed by

experienced clinicians to be relevant to sociallyanxious individuals. This was done using theexpert-guided process described above for theNSPS. Following this process, 2 items were removedand 9 items were added to our original 46-item pool,resulting in a total of 53 SCBI items. Respondentswere instructed to rate how frequently they typicallyengage in each of the listed behaviors before orduring anxiety-provoking social situations in orderto prevent or alleviate their experience of anxiety.1

Social Phobia Scale (SPS) and Social InteractionAnxiety Scale (SIAS) (Mattick & Clarke, 1998)The SPS and SIAS are companion 20-item self-report scales that assess social performance andinteraction fears, respectively. Each item is rated ona scale from 0 (not at all characteristic or true of me)to 4 (extremely characteristic or true of me), with acombination of sum- and reverse-scored items. Bothscales have been found to have good test-retestreliability (r=0.91-0.93) and convergent validity(r=0.59-0.77 with other measures of social anxi-ety), high levels of internal consistency (Cronba-ch's alpha indices of 0.90 and 0.91), and todistinguish between individuals with and withoutSAD (Mattick & Clarke, 1998; Osman, Gutierrez,Barrios, Kopper, & Chiros, 1998; Peters, 2000).

Liebowitz Social Anxiety Scale–Self Report Version(LSAS-SR; Baker, Heinrichs, Kim,&Hofmann, 2002)The LSAS-SR is a 24-item self-report version of awell-validated clinician-administered instrument(Heimberg et al., 1999) that assesses fear andavoidance, in separate subscales, across a numberof social and performance situations. For the fearsubscale (LSAS-F), each item is rated on a scale of0 (none) to 3 (severe). For the avoidance subscale(LSAS-A), each item is rated on a scale of 0 (never)to 3 (usually). Like the clinician-administeredversion, the LSAS-SR has been shown to havegood test-retest reliability (r=0.83), internal con-sistency (Cronbach's alpha of 0.95), and conver-gent and discriminant validity (e.g., among patientswith SAD, correlations with other measures ofsocial anxiety were found to be significantlystronger than correlations with measures of de-pression) (Baker et al., 2002; Fresco et al., 2001).

Obsessive Compulsive Inventory–Short Version(OCI-SV; Foa et al., 2002)Originally adapted from the longer 42-item Obses-sive-Compulsive Inventory (Foa, Kozak, Salkvoskis,Coles, & Amir, 1998), the OCI-SV is an 18-item


brief self-report scale that measures concerns asso-ciated with obsessive-compulsive disorder (e.g., I getupset if objects are not arranged properly; I collectthings I don't need; etc.). Respondents are asked torate how much each item distresses or bothers themon a scale from 0 (not at all) to 4 (extremely). TheOCI-SV has been found to have excellent test-retestreliability (r=0.82–0.84), good convergent validity(r=0.53–0.85 with other measures of OCD), andhigh internal consistency (Cronbach's alpha indicesof 0.81 to 0.93) (Foa et al., 2002). Of note, however,was Foa and colleagues' (2002) finding that theOCI-SV had a high correlation (r=0.70) with theBeck Depression Inventory, a conceptually distinctmeasure of depression.

Beck Depression Inventory II (BDI-II; Beck, Steer,& Brown, 1996)The BDI-II is a 21-item self-report instrument forassessing the intensity of depression in diagnosedpatients and detecting symptoms of depression innormal populations. Respondents are instructed tochoose the statement amongst a group of state-ments that best applies to how they generally feel.Each group of statements ranges from 0 (e.g., I donot feel sad) to 3 (e.g., I am so sad or unhappy that Ican't stand it). The BDI-II has consistently beenshown to have adequate test-retest reliability(r=.60–.83 in nonclinical samples and r=.48–.86in clinical samples), good convergent and discrim-inant validity (e.g., the BDI-II has been found to besignificantly more strongly correlated with depres-sion relevant measures, such as the Beck Hopeless-ness Scale, than with measures of anxiety, pb .01),and high internal consistency (Cronbach's alphaindices of 0.76 to 0.95 in clinical populations and.73 to .92 in nonclinical populations; Osman,Barrios, Gutierrez, Williams, & Bailey, 2008).

procedure

All questionnaires were administered in onlinesessions via a standardized web-based automatedcomputer program that was designed to protect

Table 1Demographic Characteristics of Participants

Age (N=220 for Group 1; 277 for Group 2)Gender (% female) (N=225 for Group 1; 281 for Group 2)Ethnicity (N=224 for Group 1; 276 for Group 2)CaucasianAsianEast IndianMiddle EasternOther

Note. Due to some missing data that could not be imputed, Ns vary for

participants’ confidentiality and the security of theirdata. Participants in Group 1 completed two onlinesessions. The first session included administrationof the BDI-II, LSAS-SR, OCI-SV, SPS, and SIAS.Due to the anticipated length and monotony ofcompleting multiple questionnaires in one sitting,participants were then asked to complete theNSPS and SCBI in a separate session 2 days afterSession 1. All measures in both sessions werepresented in a randomly counterbalanced order.Our experience from administering the question-

naires to Group 1 indicated that all measures could becompleted in a reasonable amount of time withoutimposing unrealistic demands on the participants.Thus, participants in Group 2 completed all of thecounterbalanced measures—BDI-II, LSAS-SR, OCI-SV, SPS, SIAS, NSPS, and SCBI—in one session. Oneweek prior to this session, we also asked participantsto complete theNSPS (but none of the othermeasures)to enable us to evaluate its test-retest reliability.Preliminary item-by-item frequency analyses for

each measure administered in the study indicatedthat the total proportion of missing data was low(0% to 2% for the vast majority of items) across allmeasures and participants in both groups. Thus,missing data were imputed for all measures usingthe expectation-maximization (EM) estimationmethod in SPSS Statistics 18.0 (2009). Using aniterative process, the EM method estimates themeans, the covariance matrix, and the correlationof quantitative (scale) variables of missing values.When imputing these missing values, we includedonly the measures that were used in each relevantanalysis described below.

results

Descriptive AnalysesDemographic information (reported age, gender,and ethnicity) for participants in Groups 1 and 2are presented in Table 1. A MANOVA with genderas the grouping variable was conducted for Groups1 and 2. There were no significant effects of genderon any of the measures, F(6, 218)=1.67, pN .05, for

Group 1 Group 2

19.9 years (SD=3.35) 20.6 years (SD=3.74)72.9% 68.3%

45.5% 47.8%29.9% 32.6%12.9% 4.3%3.6% 4.3%8.0% 10.9%

Groups 1 and 2.

Table 2Internal Consistency, Means and (Standard Deviations) of Study Measures

Measure Group 1 Group 2

Cronbach's α Mean (SD) 1 Cronbach's α Mean (SD) 1

NSPS 0.96 2 57.95 (21.61) 0.962 60.89 (23.30)SCBI 0.96 101.64 (23.30) 0.95 99.21 (22.65)SPS 0.93 20.07 (14.74) 0.95 20.29 (15.88)SIAS 0.93 29.71 (16.15) 0.90 31.50 (14.16)LSAS-SR-F 0.93 25.41 (12.51) 0.92 26.06 (11.96)LSAS-SR-A 0.91 23.76 (12.93) 0.90 23.35 (12.13)BDI-II 0.93 11.52 (9.74) 0.93 11.47 (9.54)OCI-SV 0.89 16.64 (10.82) 0.89 15.90 (10.73)

Note. Ns=225 (Group1) and 316 (Group 2). NSPS=Negative Self-Portrayal Scale; SCBI=Self-Concealment Behaviors Inventory;SPS=Social Phobia Scale; SIAS=Social Interaction Anxiety Scale; LSAS-SR-F=Liebowitz Social Anxiety Scale – Self Report Version -Fear Subscale; LSAS-SR-A= Liebowitz Social Anxiety Scale – Self Report Version - Avoidance Subscale; BDI-II=Beck DepressionInventory II; OCI-SV=Obsessive Compulsive Inventory – Short Version.1 Based on 27-item NSPS.2 Despite the high internal consistency values, the highest correlation between any two of the retained NSPS items across both samples

was r=.80, suggesting non-redundancy between the items.


Group 1, and F(6, 274)= .972, pN .05, for Group 2.A MANOVA with ethnicity (Caucasian, Asian,East Indian, Middle Eastern, and Other) as thegrouping variable was conducted on the SIAS, SPS,LSAS-F, LSAS-A, NSPS, and SCBI scores forGroups 1 and 2. The results indicated no significanteffects of ethnicity on any of the measures, F(36,933)=1.15, pN .05, for Group 1, and F(36,1162)=1.36, pN .05, for Group 2.2 Mean scores,standard deviations, and Cronbach's alphas for eachmeasure in Groups 1 and 2 are presented in Table 2.

Scale Item Characteristics and Factor AnalysisGroup 1 scale item characteristics and exploratory

factor analysis. EFA with maximum likelihoodextraction was used to examine the underlyingstructure of the NSPS in the first group ofparticipants. As the hypothesized factors of theNSPS were expected to correlate with each other(Moscovitch, 2009), the NSPS items were subjectedto analysis using an oblique (promax) rotation.Items were removed that were cross-loading orunreliable, in accordance with Comrey and Lee's(1992) recommendations for factor analyses in-volving multiple subscales. All retained items withfactor loadings of N .45 on a single factor and b .32on any other factor were retained, with theexception of the one “blushing” item, whichcross-loaded onto two factors but was retainedbecause of its clinical relevance in socially anxious

2 Independent-sample t-tests were also conducted to comparescores across measures for the primary ethnic groups (Caucasianvs. Asian). Results in both Groups indicated that relative toCaucasians, Asian participants reported higher scores on the SIAS(ts N 2.10, ps b .05), and NSPS (ts N 2.15, ps b .05 ). There were noother significant differences.

populations (e.g., Bögels & Reith, 1999; Gerlach,Wilhelm, Gruber, & Roth, 2001). Following thisprocess, 27 items remained, all of which hadcorrected item-total correlations greater than orequal to .43. An inspection of all eigenvaluesgreater than 1, as well as the scree plot, indicatedthat a three-factor solution provided the best fit.The three factors accounted for 55.1% of the totalvariance. Interpreting the factor loadings, welabeled the first factor, which accounted for45.4% of the variance, concerns about socialcompetence. The second factor, accounting for5.6% of the variance, was labeled concerns aboutphysical appearance. Finally, the third factoraccounted for 4.1% of the variance and was labeledconcerns about showing signs of anxiety. Table 3presents the rotated factor loadings from the EFApattern matrix for each of the 27 NSPS items thatwere retained in the first group of participants(please also see the notes below Table 3 for theNSPS respondent instructions).

Group 2 scale item characteristics and confirma-tory factor analysis. As indicated above, the oneblushing item included in the measure for partici-pants in Group 1 did not load cleanly onto onefactor, but instead cross-loaded onto both concernsabout social competence (.29) and concerns aboutshowings signs of anxiety (.24). This finding wasconsistent with those of Schultz and colleagues(2006), who, in their recent psychometric study ofthe Appraisal of Social Concerns Scale (ASC),reported that the one ASC blushing item wasdiscarded because it cross-loaded both onto a factorrelated to social competence and a factor related tosocial anxiety. We chose to retain the blushing item

Table 3Rotated Factor Loadings from Pattern Matrix for Non-Discarded Items of the Negative Self-Portrayal Scale for Group 1

Item Social Competence Physical Appearance Signs of Anxiety Communalities

Social Competence21. Socially awkward 1.00 -.07 -.11 .7724. Lacking social skills .95 -.04 -.09 .7512. Lacking personality .76 .18 -.08 .7110. Interpersonally ineffective .74 .03 .09 .6714. Unable to express myself .72 -.08 .13 .5818. Reserved .66 -.02 -.06 .3717. Humorless .63 .08 .12 .6023. Speaking incoherently .56 .09 .20 .6320. Stupid .55 .24 -.04 .5119. Aloof .53 .13 .04 .433. Boring .50 .13 .09 .44Physical Appearance27. Ugly -.01 .97 -.12 .805. Physically unattractive -.03 .93 -.10 .7311. Weird-looking .07 .80 .02 .7513. Fat -.10 .72 .07 .4726. Unfashionable .17 .66 -.01 .619. Blemished (i.e., my appearance) .08 .64 .05 .5322. Having a bad hair day .00 .58 .19 .512. Poorly dressed .07 .46 .14 .38Signs of Anxiety8. Speaking with a trembling voice -.02 -.02 .82 .6225. Fidgeting -.05 .02 .79 .581. Stuttering -.16 .01 .73 .4115. Twitching (i.e., my facial muscles) .10 -.06 .64 .4616. Frozen .32 .01 .51 .596. Losing control of my emotions .16 .09 .50 .464. Sweating .04 .07 .49 .327. Blushing .29 -.05 .24 .21

55.11%

Note. NSPS instructions to respondents read: “Using the scale provided above the following list of items, please select the circle under theappropriate number/statement to indicate the degree to which you are concerned about the following aspects of yourself when you are inanxiety-provoking social situations (e.g. talking to someone who is a stranger; giving a speech in front of an audience; answering aquestion in class; etc.).” For each item, respondents are asked to consider the following statement: “In social situations (in which I feelanxious), it will become obvious to other people that I am ____.” Response options per item range from 1-5, with descriptive anchorsincluding “not at all concerned,” “slightly concerned,” “moderately concerned,” “very concerned,” and “extremely concerned.” The NSPS totalscore is a sum score of all items. Each subscale can also be summed separately to obtain data on the relative strength of self-portrayalconcerns across dimensions within individuals and across individuals within a larger sample.


because of its clinical relevance for socially anxiousindividuals and our overarching theoretical model(i.e., Moscovitch, 2009). In order to explorewhether blushing might represent a separate fourthfactor, three new blushing items were constructed(i.e., becoming red in the face, becoming hot in theface, and becoming flushed) and added to the NSPS(resulting in a 30-item measure), which we admin-istered to participants in Group 2.CFA was used to analyze this 30-item NSPS

scale for participants in Group 2 using Amos17.0 (a structural equation modeling program;Arbuckle, 2008). Missing data were handled withthe full information maximum likelihood method(Anderson, 1957), as implemented in Amos. Thedata met the criteria for univariate and multi-

variate normality suggested by Kline (1998) andWest, Finch, and Curran (1995); for example,values for univariate skew were all less than|1.57| and values for univariate kurtosis were allless than |1.61|.A 4-factor model was evaluated with CFA, which

included the 3-factor structure from Group 1 and aseparate fourth factor containing the four blushingitems. This model was evaluated and fit moderatelywell, χ2(399)=1579.83, pb .001; comparative fitindex (CFI)= .846; root-mean-square error of ap-proximation (RMSEA)= .097; probability of closefit (pclose)= .000. However, the four blushing itemswere highly correlated with each other (r=.76-.88)and deemed redundant; thus, only the one“blushing”item was retained.


CFA was then conducted to test the originalMoscovitch (2009) 4-factor model with theremaining 27 NSPS items. The model demonstratedgood fit, χ2(318)=948.52, pb .001; CFI= .898;RMSEA=.079; pclose=.000. However, two of thefactors—concerns about social skills and behaviorsand concerns about personality—were highly corre-

FIGURE 1 Confirmatory factor analysis of the NSPS 3-factothe corresponding number from the NSPS items shown in Ta

lated (r=.96), indicating that they were not distinctfactors and would be more accurately representedas one factor (concerns about social competence).Thus, a new model was evaluated via CFA,

consisting of 27 items loading onto the three factorsfound in Group 1. The one “blushing” item wasconstrained to load onto the concerns about signs of

r model for Group 2. Items are labelled with I= item andble 3. All factor loadings had a p-valueb .001 (2-tailed).


anxiety factor. This 3-factor model fit well,χ2(321)=974.58, p b .001; CFI = .895; RMSEA= .080;pclose= .000, and is illustrated in Fig. 1 withstandardized estimates. The three factors arelabeled: (a) Social Competence, (b) Physical Ap-pearance, and (c) Signs of Anxiety. Each item of thescale is identified by number, according to the 27-item measure presented in Table 3.Finally, for comparative purposes, an alternative

3-factor model was tested via CFA, consisting of27 items loading onto the three factors found inGroup 1 and the one “blushing” item constrainedto load onto the social competence factor. Thismodel did not fit as well, χ2(321)=1018.27,pb .001; CFI= .888; RMSEA=.083; pclose= .000.The significance of the difference in model fitbetween this model and the one displayed in Fig. 1could not be tested because they are not nestedmodels. However, the Browne-Cudeck Criterion(BCC) and expected cross-validation index (ECVI)values were inspected for each of these models,with smaller values indicating better model fit(Wicherts & Dolan, 2004). Relative to the 3-factormodel with blushing loading onto the socialcompetence factor, the 3-factor model with blush-ing loading on the signs of anxiety factor(displayed in Fig. 1) demonstrated lower BCC(1158.97 vs. 1202.66) and ECVI (3.63 vs. 3.77)values and was, therefore, retained.Intercorrelations between the three NSPS factors

modeled in Fig. 1 (social competence, signs of anxiety,and physical appearance) are reported in Table 4.

Internal Consistency and Test-Retest ReliabilityThe NSPS demonstrated excellent internal consis-tency. Cronbach's alphas for the 27-item NSPS inGroups 1 and 2 were both .96. Each factor alsodemonstrated strong internal consistency, withCronbach's alpha coefficients for the social com-petence, physical appearance, and signs of anxietyfactors, respectively, of .93 (n=225), .91 (n=225),and .85 (n=225) in Group 1, and .94 (n=316), .93(n=316), and .88 (n=316) in Group 2.

Table 4IntercorrelationsBetween the Factors of theNegative Self-PortrayalScale

Note. Ns=225 (Group 1) and 316 (Group 2). Group 1 correlationsare in the gray-shaded region.*pb .001 (2-tailed).

Though not measured in Group 1, the test-retestreliability of the 27-item NSPS in Group 2 wassatisfactory (r=.75; n=316) over a 1-week interval.

The SCBI also demonstrated excellent internalconsistency. Cronbach's alphas for the 53-itemSCBI in Groups 1 and 2, respectively, were .96(n=225) and .95 (n=316).

Convergent and Discriminant ValidityAs depicted in Table 5, NSPS total scores forGroups 1 and 2 demonstrated significantly higherzero-order correlations with measures of socialanxiety, including LSAS-SR-F, LSAS-SR-A, SPS,and SIAS scores (r=.63–.73) than with measures oftheoretically distinct constructs such as OCD(r=.40–.47) (p-values b .01; n=225 and 316).Although the NSPS in Group 2 was significantlymore strongly correlated with measures of socialanxiety than with the theoretically distinct constructof depression, as measured by BDI-II depressionscores (r=.53) (pdifferenceb .001; n=316), the NSPSin Group 1 was not more strongly correlated withother measures of social anxiety than with the BDI-II (r=.65) (pdifferenceN .05; n=225).ForGroups 1 and2, the 53-itemversion of the SCBI

total score demonstrated significantly higher zero-order correlations with measures of social anxiety,including LSAS-SR-F, LSAS-SR-A, SPS, and SIASscores (r=.71–.77) than with measures of theoretical-ly distinct constructs, including both the OCI-SV(r=.43-.45) (pdifferenceb .001; n=225 and 316) andBDI-II (r=.50-.61) (pdifferenceb .05; n=225 and 316).

Construct ValidityMoscovitch (2009) theorized that self-portrayalconcerns and self-concealment (i.e., safety) beha-viors are functionally related to one another, suchthat self-concealment is used in a strategic mannerby socially anxious individuals to prevent othersfrom observing their negative self-attributes insocial situations. We were, therefore, interested inexamining the hypothesis that the NSPS would be auseful measure that would help account for thewide-ranging use of self-concealment behaviors,over and above established symptom measures ofsocial anxiety and depression. Thus, we predictedthat self-concerns (even in the unselected samples inthe present study) would account for a significantamount of variance in self-concealment behaviorsover and above the influence of symptoms of socialinteraction anxiety, social performance anxiety, anddepression. To this end, we conducted a hierarchicalmultiple regression analysis with SCBI scores as thecriterion variable and four predictor variables(SIAS, SPS, BDI-II, and NSPS scores) entered infour separate steps. Tolerance values were

Table 5Zero-Order Correlations Between the NSPS Total Score, NSPS Subscales, SCBI, and Other Self-Report Measures

Measures NSPSTotal

NSPS SocialCompetence

NSPS PhysicalAppearance

NSPS Signsof Anxiety

SCBI

SCBIGroup 1 .77 ⁎ .73 ⁎ .62 ⁎ .72 ⁎ –Group 2 .73 ⁎ .72 ⁎ .58 ⁎ .64 ⁎ –SPSGroup 1 .69 ⁎ .65 ⁎ .58 ⁎ .64 ⁎ .77 ⁎

Group 2 .73 ⁎ .67 ⁎ .62 ⁎ .68 ⁎ .76 ⁎

SIASGroup 1 .69 ⁎ .71 ⁎ .55 ⁎ .58 ⁎ .71 ⁎

Group 2 .71 ⁎ .73 ⁎ .55 ⁎ .61 ⁎ .75 ⁎

LSAS-SR-FGroup 1 .68 ⁎ .66 ⁎ .56 ⁎ .60 ⁎ .76 ⁎

Group 2 .67 ⁎ .68 ⁎ .52 ⁎ .59 ⁎ .76 ⁎

LSAS-SR-AGroup 1 .63 ⁎ .62 ⁎ .52 ⁎ .57 ⁎ .75 ⁎

Group 2 .65 ⁎ .66 ⁎ .51 ⁎ .56 ⁎ .75 ⁎

BDI-IIGroup 1 .65 ⁎ .60 ⁎ .54 ⁎ .58 ⁎ .61 ⁎

Group 2 .53 ⁎ .50 ⁎ .45 ⁎ .48 ⁎ .50 ⁎

OCI-SVGroup 1 .47 ⁎ .44 ⁎ .33 ⁎ .49 ⁎ .45 ⁎

Group 2 .40 ⁎ .37 ⁎ .35 ⁎ .37 ⁎ .43 ⁎

Note. Ns=225 (Group 1) and 316 (Group 2). NSPS=Negative Self-Portrayal Scale; SCBI=Self-Concealment Behaviors Inventory;SPS=Social Phobia Scale; SIAS=Social Interaction Anxiety Scale; LSAS-SR-F=Liebowitz Social Anxiety Scale – Self Report Version -Fear Subscale; LSAS-SR-A= Liebowitz Social Anxiety Scale – Self Report Version - Avoidance Subscale; BDI-II=Beck DepressionInventory II; OCI-SV=Obsessive Compulsive Inventory – Short Version.All correlations for Groups 1 and 2 are based on the 27-item NSPS.⁎ pb .001 (2-tailed).

Table 6Summary of Hierarchical Regression Analysis for Variables Predicting Levels of Social Self-Concealment on the SCBI

Predictor B SE B ß

Group 1 Group 2 Group 1 Group 2 Group 1 Group 2

Step 1SIAS 1.04 1.02 0.07 0.05 0.72 ⁎⁎ 0.75 ⁎⁎

Step 2SIAS 0.44 0.57 0.10 0.07 0.31 ⁎⁎ 0.42 ⁎⁎

SPS 0.83 0.63 0.10 0.07 0.53 ⁎⁎ 0.45Step 3

SIAS 0.41 0.55 0.09 0.07 0.28 ⁎⁎ 0.41 ⁎⁎

SPS 0.65 0.59 0.11 0.08 0.41 ⁎⁎ 0.42 ⁎⁎

BDI 0.55 0.16 0.12 0.10 0.23 ⁎⁎ 0.07Step 4

SIAS 0.22 0.44 0.09 0.07 0.15 ⁎ 0.32 ⁎⁎

SPS 0.54 0.46 0.10 0.08 0.34 ⁎⁎ 0.32 ⁎⁎

BDI 0.25 0.08 0.12 0.09 0.10 ⁎ 0.03NSPS 0.39 0.24 0.06 0.05 0.36 ⁎⁎ 0.24 ⁎⁎

Note. SIAS=Social Interaction Anxiety Scale; SPS=Social Phobia Scale; BDI-II=Beck Depression Inventory II; NSPS=Negative Self-Portrayal Scale; SCBI=Self-Concealment Behaviors Inventory.Group 1: N=225; R2= .516 for Step 1 (pb .001); R2

change=.108 for Step 2 (pb .001); R2change=.035 for Step 3 (pb .001); R2

change= .050 forStep 4 (pb .001).Group 2: N=316; R2= .565 for Step 1 (pb .001); R2

change= .087 for Step 2 (pb .001); R2change= .003 for Step 3 (pN .05); R2

change= .024 forStep 4 (pb .001).⁎ pb .05.⁎⁎ pb .001.



examined prior to this analysis to assess for thepotential impact of multicolinearity and were foundto be no less than .33 for all predictor variables,which is well above the minimum criteria of .10(Cohen, Cohen, West, & Aiken, 2003).SIAS scores were entered in step 1, SPS scores in

Step 2, BDI-II scores in Step 3, and NSPS totalscores in Step 4. The totalNs were 225 and 316 forGroups 1 and 2, respectively. As shown in Table 6,social interaction anxiety, social performance anx-iety, and depression symptoms each accountedsignificantly and uniquely for variance in SCBIscores in Steps 1, 2 and 3 for Group 1. For Group 2,only social interaction and performance anxiety(but not depression symptoms) accounted signifi-cantly and uniquely for variance in SCBI scores.Importantly, for both Groups 1 and 2, the additionof NSPS scores to the regression equation in Step 4accounted for a significant proportion (2.4%–5.0%) of the variance (in the use of self-conceal-ment behaviors over and above the contributions ofSIAS, SPS and BDI-II scores).3

DiscussionThe results of the present study suggest that theNSPS is a psychometrically strong and potentiallyuseful instrument for measuring concerns thatindividuals have about exposing specific negativeself-attributes to critical others in anxiety-provok-ing social situations. The NSPS represents a novelassessment tool in the social anxiety field that goesbeyond measuring the level or types of social

3We also analyzed the psychometric properties of the NSPS forthe Caucasian (n = 102 in Group 1; n = 132 in Group 2) and Asian(n = 67 in Group 1; n = 90 in Group 2) groups separately:

(1) NSPS total score internal consistency: Group 1 (Caucasians =.96; Asians = .95) and Group 2 (Caucasians = .95; Asians =.97).

(2) NSPS internal consistency for the 3 factors: Group 1(Caucasians = .93 SocCom, .91 PhyApp, .86 SigAnx.;Asians = .93 SC, .92 PA, .84 SA) and Group 2 (Caucasians =.93 SC, .94 PA, .84 SA; Asians = .95 SC, .94 PA, .91 SA).

(3) NSPS test-retest reliability: Group 2 (Caucasians = .74;Asians = .75).

(4) SCBI internal consistency: Group 1 (Caucasians = .96;Asians = .95) and Group 2 (Caucasians = .95; Asians = .96).

(5) NSPS convergent/discriminant and construct validity: Spacelimitations prevent us from reproducing the correlationalmatrices and regression results here, which were conductedfor the Caucasian and Asian groups separately across bothsamples. The overall pattern of results between the groupswas similar. Any differences that emerged between ethnicgroups were not found consistently across the two samplesand may have been affected by sample size considerations,making these post-hoc analyses difficult to interpret. The fullset of results is available by request from the first author.

anxiety symptoms respondents experience acrossdifferent situations. It enables the specific assess-ment of the core negative features of the self that arethought to underlie self-presentational concernsthat drive social anxiety (e.g., Leary & Kowalski,1995; Stopa, 2009a) and to account for observedheterogeneity in situational fears, avoidance, andself-concealment behaviors amongst socially anx-ious individuals (Moscovitch, 2009). Moreover, thedevelopment of this measure represents a crucialfirst step in the process of testing Moscovitch'smodel of social anxiety.Empirical evidence obtained across two indepen-

dent undergraduate samples in the present studyindicated that NSPS items developed to tapconcerns about social skills and concerns aboutpersonality loaded jointly onto a single factor,which we labeled concerns about social compe-tence. Whereas psychologists might distinguishbetween social skills and behaviors, on one hand,and personality, on the other, it is likely that NSPSrespondents simply considered social skill/beha-viors to be observable indices of personality. Thus,respondents may be indicating that concerns aboutappearing “socially awkward” or “speaking inco-herently” (both originally social skills/behaviorsitems) have the same negative social impact asappearing “aloof” or “boring” (both originallypersonality items); in other words, exhibiting flawsin social skills/behaviors or flaws in personalityleads essentially to the same concern: appearingsocially incompetent.Alongside concerns about social competence and

in support of Moscovitch's (2009) theory, concernsabout signs of anxiety and concerns about physicalappearance emerged as separate (nonorthogonal)factors in the EFA. This 3-factor solution wasassociated with good model fit in the subsequentCFA. It was surprising that the blushing item cross-loaded relatively weakly onto both the signs ofanxiety and social competence factors in the EFA,but we chose to retain the blushing item for the CFAdue to its empirically supported relevance in socialanxiety research and practice. In the subsequentCFA, we found reasonably strong model fit whenwe constrained the blushing item to load only ontothe signs of anxiety factor. Future research isneeded to replicate this finding.While the observed factors of the NSPS across

both groups in the present study did not mapperfectly onto the predicted dimensions, theemergent factor structure of the NSPS appears tobe valid, reliable, fairly stable over a 1-weekperiod, and conceptually sound. A test-retestcorrelation of .75 is consistent with our expecta-tions that the NSPS reflects relatively stable


concerns about self-attributes that ought not tochange significantly over short time intervals in theabsence of treatment.The new measure demonstrated adequate con-

current validity across both samples and explainedsignificant variance in the use of self-concealmentbehaviors over and above that accounted for bysocial interaction anxiety, social performanceanxiety, and depression symptoms. Although wehypothesized that the NSPS would correlate morestrongly with measures of social anxiety than withmeasures of depression, the observed correlationsin our study between the NSPS and measures ofboth social anxiety and depression were quiterobust. A possible reason for this finding is that theNSPS and BDI-II capture overlapping componentsof the self-critical cognitive style that is character-istic of—and may serve as a common cognitivediathesis underlying—both syndromes (seeIngram, Ramel, Chavira, & Scher, 2005). Indeed,previous research has highlighted the central roleof negative self-related cognitions in the phenom-enology of both social anxiety and depression (e.g.,Dozois & Frewen, 2006; Moscovitch, Hofmann,Suvak, & In-Albon, 2005). Another possibility isthat the NSPS might have good discriminantvalidity in relation to some measures (e.g., OCI-SV) but poor discriminant validity in relation toothers (e.g., BDI-II). More research is required tobetter understand the unique cognitive features ofsocial anxiety and depression and whether theNSPS is capable of distinguishing between them.The present study was not without its limitations.

Notwithstanding the recent publication of theSubtle Avoidance Frequency Examination (SAFE;Cuming, Rapee, Kemp, Abbott, Peters, & Gaston,2009), no validated, published measure of safetybehaviors existed when our data were collected;thus, we adapted the SCBI from a previouslyunpublished measure of safety behaviors and usedit as our measure of self-concealment behaviors inthe present study. Our results suggested that theSCBI is a valid and reliable measure, with highCronbach's alphas across both samples and ade-quate convergent and discriminant validity. Weassessed the discriminant validity of the NSPS inrelation to the BDI-II and the OCI-SV, but it will beimportant in future studies to determine how theNSPS correlates with measures of constructs—forexample, worry, panic, or higher-order dimensions(e.g., Brown & Barlow, 2009)—that might be evenmore similar to social anxiety both conceptuallyand clinically than depression or OCD per se. Sinceall questionnaires were administered online, it ispossible that participants felt less accountable toanswer questions as carefully as participants who

complete pencil-and-paper measures in the labora-tory might feel. Consequently, although previousstudies comparing online and paper administra-tions of questionnaires have shown that onlinemeasures are highly reliable (e.g., Luce, Winzelberg,Das, Osborne, Bryson, & Taylor, 2007), it wouldbe prudent to replicate our study with paperquestionnaires. In addition, our use of multiplesessions for completion of the study measures mayhave introduced additional variability into thestudy associated with participants’ fluctuatingcircumstances and moods across testing sessions.Furthermore, despite the wide range of reportedsymptoms of anxiety and depression in our study,the overall mean levels of these symptoms werequite low; thus, it is impossible to know whetherand how the NSPS findings reported here wouldgeneralize from the present participants to clinicalsamples. Moreover, because our unselected samplesof participants were comprised predominantly ofyoung, female, university-educated, and socioeco-nomically advantaged individuals, we cannot inferthat the results would generalize to samples ofindividuals with different compositions of demo-graphic characteristics.As highlighted by some of the reported differ-

ences between Asian and Caucasian participants inthe present study, it is crucial to note that, becauseindividuals’ self-portrayal concerns depend on thestandards and norms for social behavior andbecause such standards and norms vary consider-ably across cultures (e.g., Markus & Kitayama,1991), it is likely that the NSPS is a culturally boundmeasure. Thus, whereas the values of one culturemight dictate that its members make humorousremarks or witty conversation during social gather-ings in order to appear socially competent, thevalues of another culture might dictate that socialcompetence is reflected in one's ability to remainsilent during social gatherings, constrain the desireto speak out, and practice listening respectfully tothe elders of the community. In other words, whatmight appear socially incompetent according to thenorms of one society might appear quite competentaccording to the norms of another. Similarly, thereare, no doubt, variations across cultures in stan-dards of physical appearance and norms associatedwith showing signs of anxiety, which in turn woulddictate cross-cultural differences in the relevance ofand responses to the present version of the NSPS.Therefore, it would be both interesting andimportant to examine the cross-cultural applicabil-ity of the NSPS in future research.Although replication of our results is clearly

required in both clinical samples of SAD patientsand community samples of healthy and anxious


control participants (an endeavor that is currentlyin progress in our laboratory), the preliminaryvalidation of the NSPS enables us to envision itsfuture use as a clinical tool to guide the assessmentand treatment of SAD. We hypothesize that clinicaloutcomes in SAD will be improved by practi-tioners’ assessment of the nature of their patients’self-portrayal concerns and the subsequent selec-tion of specific interventions to target thoseconcerns across the dimensions of the NSPS. Itmight be worthwhile to remain attuned to thepotentially useful distinction between concernsabout social skills/behaviors and concerns aboutpersonality in the context of SAD, even though thisdistinction is not captured in the factor structure ofthe NSPS. In our clinical experience, concernsabout social skills/behaviors that are endorsed bypatients with SAD tend to be most responsive tointerventions such as video feedback (e.g., Harvey,Clark, Ehlers, & Rapee, 2000; McManus et al.,2009) and similar types of behavioral experiments(Bennett-Levy et al., 2004) that are designed toimprove the self-other discrepancy in imaginedbehavioral deficits (e.g., Rodebaugh & Rapee,2006). Conversely, concerns about signs of anxietyor personality flaws in the context of SAD may bemore resistant to change with video feedback (seeOrr & Moscovitch, 2010b; Stopa, 2009b) and,therefore, might require therapeutic interventionwith a different set of intervention strategies.Finally, we are currently working on adapting thecurrent version of the NSPS to investigate in futurestudies not only how concerned socially anxiousindividuals are about publicly exposing fearednegative self-attributes, but also how much theymight overestimate the probability and social costsassociated with revealing each self-attribute tocritical others. We hope to track how suchprobability and cost estimates might differ acrossthe NSPS dimensions and change as a result oftreatment.In closing, the preliminary validation of the NSPS

enables us to renew our call for a paradigm shift intreating SAD (e.g., Moscovitch, 2009). Traditionalconceptualizations of exposure-based CBT for SAD(e.g., Heimberg, 2009) are guided by a therapeuticmindset that is characterized by the question:“What types of feared social situations must beconfronted in the service of anxiety reduction?”Confronting social situations is indeed necessary inthe treatment of social anxiety, but it is notsufficient. We hypothesize that the extinction oflearned fear in patients with SAD vis-à-vis theprinciples of emotional processing theory (Foa &Kozak, 1986; Foa & McNally, 1996; see alsoMoscovitch, Antony, & Swinson, 2009) will be

more effectively facilitated with both an a prioriexploration of patients’ self-portrayal concerns andan overarching therapeutic mindset that is guidedby the following question: “What about the selfdoes this patient perceive as being flawed, and howcan new learning about the self be promoted mosteffectively during therapy in the service of anxietyreduction?” Future research is required to supportthis testable prediction.

ReferencesAlden, L. E., & Taylor, C. T. (2004). Interpersonal processes in

social phobia. Clinical Psychology Review, 24, 857–882.Anderson, T. W. (1957). Maximum likelihood estimates for a

multivariate normal distribution when some observationsare missing. Journal of the American Statistical Association,52, 200–203.

Antony, M. M., Orsillo, S. M., & Roemer, L. (Eds.). (2001).Practitioner's guide to empirically based measures ofanxiety: AABT clinical assessment series. Dordrecht,Netherlands: Kluwer Academic Publishers.

Arbuckle, J. L., (2008) Amos 17.0 users guide. Chicago, IL.Baker, S. L., Heinrichs, N., Kim, H. J., & Hofmann, S. G.

(2002). The Liebowitz social anxiety scale as a self-reportinstrument: a preliminary psychometric analysis. BehaviourResearch and Therapy, 40, 701–715.

Bennett-Levy, J., Butler, G., Fennell, M., Hackmann, A.,Mueller, M., & Westbrook, D. (2004). Oxford guide tobehavioural experiments in cognitive therapy. New York:Oxford University Press.

Bögels, S. M., & Reith, W. (1999). Validity of two ques-tionnaires to assess social fears: The Dutch Social Phobiaand Anxiety Inventory and the Blushing, Trembling, andSweating Questionnaire. Journal of Psychopathology andBehavioral Assessment, 21, 51–66.

Brown, T. A. (2006). Confirmatory factor analysis for appliedresearch. New York: The Guilford Press.

Clark, D. M., Butler, G., Fennell, M., Hackmann, A.,McManus, F., & Wells, A. (1995). Social behaviourquestionnaire. Unpublished manuscript.

Clark, D. M., & Wells, A. (1995). A cognitive model of socialphobia. 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 Press.

Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003).Applied multiple regression/correlation analysis for thebehavioral sciences (3rd ed.). Mahwah: Lawrence ErlbaumAssociates.

Comrey, A. L., & Lee, H. B. (1992). A first course in factoranalysis (2nd ed.). Hillsdale, NJ: Lawrence ErlbaumAssociates.

Cuming, S., Rapee, R. M., Kemp, N., Abbott, M. J., Peters, L.,& Gaston, J. E. (2009). A self-report measure of subtleavoidance and safety behaviors relevant to social anxiety:Development and psychometric properties. Journal ofAnxiety Disorders, 23, 879–883.

Dozois, D. J. A., & Frewen, P. A. (2006). Specificity of cognitivestructure in depression and social phobia: A comparison ofinterpersonal and achievement content. Journal of AffectiveDisorders, 90, 101–109.

Gerlach, A. L., Wilhelm, F. H., Gruber, K., & Roth, W. T.(2001). Blushing and physiological arousability in socialphobia. Journal of Abnormal Psychology, 110, 247–258.


Foa, E. B., Huppert, J. D., Leiberg, S., Langer, R., Kichic, R.,Hajcak, G., & Salkovskis, P. M. (2002). The obsessive-compulsive inventory: Development and validation of ashort version. Psychological Assessment, 14, 485–496.

Foa, E. B., &Kozak,M. J. (1986). Emotional processing of fear:Exposure to corrective information. Psychological Bulletin,99, 20–35.

Foa, E. B., Kozak, M. J., Salkovskis, P., Coles, M. E., &Amir, N. (1998). The validation of a new obsessive-compulsive disorder scale: The Obsessive-CompulsiveInventory. Psychological Assessment, 10, 206–214.

Foa, E. B., & McNally, R. J. (1996). Mechanisms of change inexposure therapy. In R. M. Rapee (Ed.), Current contro-versies in the anxiety disorders (pp. 329–343). New York:The Guilford Press.

Fresco, D.M., Coles, M. E., Heimberg, R. G., Liebowitz, M. R.,Hami, S., Stein, M. B., & Goetz, D. (2001). The LiebowitzSocial Anxiety Scale: A comparison of the psychometricproperties of self-report and clinician-administered formats.Psychological Medicine, 31, 1025–1035.

Hackmann, A., Clark, D. M., & McManus, F. (2000).Recurrent images and early memories in social phobia.Behaviour Research and Therapy, 38, 601–610.

Hackmann, A., Surawy, C., & Clark, D. M. (1998). Seeingyourself through others' eyes: A study of spontaneouslyoccurring images in social phobia. Behavioral and CognitivePsychotherapy, 26, 3–12.

Hart, T. A., Flora, D. B., Palyo, S. A., Fresco, D. M., Holle, C.,& Heimberg, R. G. (2008). Development and examinationof the Social Appearance Anxiety Scale. Assessment, 15,48–59.

Harvey, A. G., Clark, D. M., Ehlers, A., & Rapee, R. (2000).Social anxiety and self-impression: cognitive preparationenhances the effects of video feedback following astressful social task. Behaviour Research and Therapy,38, 1183–1192.

Heimberg, R. G. (2009). A new model to facilitate individual-ized case conceptualization and treatment of social phobia:An examination and reaction to Moscovitch's model.Cognitive and Behavioral Practice, 16, 135–141.

Heimberg, R. G., Horner, K. J., Juster, H. R., Safren, S. A.,Brown, E. J., Schneier, F. R., & Liebowitz, M. R. (1999).Psychometric properties of the Liebowitz Social AnxietyScale. Psychological Medicine, 29, 199–212.

Hofmann, S. G. (2007). Cognitive factors that maintainsocial anxiety disorder: A comprehensive model and itstreatment implications. Cognitive Behaviour Therapy, 36,193–209.

Hofmann, S. G., Heinrichs, N., & Moscovitch, D. A. (2004).The nature and expression of social phobia: Toward a newclassification. Clinical Psychology Review Special Issue:Social Phobia and Social Anxiety, 24, 769–797.

Hurley, A. E., Scandura, T. A., Schriesheim, C. A., Brannick,M. T., Seers, A., Vandenberg, R. J., &Williams, L. J. (1997).Exploratory and confirmatory factor analysis: Guidelines,issues, and alternatives. Journal of Organizational Behaivor,18, 667–683.

Ingram, R. E., Ramel, W., Chavira, D., & Scher, C. (2005).Social anxiety and depression. In R. W. Crozier, & L. E.Alden (Eds.), The essential handbook of social anxiety forclinicians (pp. 241–264). New York: John Wiley & Sons.

Izgiç, F., Akyüz, G., Dogan, O., & Kugu, N. (2004). Socialphobia among university students and its relation to self-esteem and body image.Canadian Journal of Psychiatry, 49,630–634.

Kline, R. B. (1998). Principles and practice of structuralequation modeling. New York: Guilford Press.

Leary, M. R., & Kowalski, R. M. (1995). The self-presentationmodel of social phobia. In R. G. Heimberg, M. R. Liebowitz,D. A. Hope, & F. R. Schneier (Eds.), Social phobia:Diagnosis, assessment, and treatment (pp. 94–112). NewYork: Guilford Press.

Ledley, D. R., & Heimberg, R. G. (2005). Social anxietydisorder. In M.M. Antony, D. R. Ledley, & R. G. Heimberg(Eds.), Improving outcomes and preventing relapse incognitive-behavioral therapy (pp. 38–76). New York: TheGuilford Press.

Luce, K. H., Winzelberg, A. J., Das, S., Osborne, M. I., Bryson,S. W., & Taylor, B. C. (2007). Reliability of self-report:Paper versus online administration. Computers in HumanBehavior, 23, 1384–1389.

Mansell, W., & Clark, D. M. (1999). How do I appear toothers? Social anxiety and processing of the observable self.Behaviour Research and Therapy, 37, 419–434.

Markus, H. R., & Kitayama, S. (1991). Culture and the self:Implications for cognition, emotion, and motivation.Psychological Review, 98, 224–253.

Mattick, R. P., & Clarke, J. C. (1998). Development andvalidation of measures of social phobia scrutiny fear andsocial interaction anxiety. Behaviour Research and Therapy,36, 455–470.

McManus, F., Clark, D. M., Grey, N., Wild, J., Hirsch, C.,Fennell, M., Hackmann, A., Waddington, L., Liness, S., &Manley, J. (2009). A demonstration of the efficacy of two ofthe components of cognitive therapy for social phobia.Journal of Anxiety Disorders, 23, 496–503.

McManus, F., Sacadura, C., & Clark, D. M. (2008). Whysocial anxiety persists: An experimental investigation ofthe role of safety behaviours as a maintaining factor.Journal of Behavior Therapy and Experimental Psychiatry,39, 147–161.

Moscovitch, D. A. (2009). What is the core fear in socialphobia?: A new model to facilitate individualized caseconceptualization and treatment. Cognitive and BehavioralPractice, 16, 123–134.

Moscovitch, D. A., Antony, M. M., & Swinson, R. P. (2009).Exposure-based treatments for anxiety disorders: Theoryand process. In M. M. Antony, & M. B. Stein (Eds.),Handbook of anxiety and anxiety disorders (pp. 461–475).New York: Oxford University Press.

Moscovitch, D. A., & Hofmann, S. G. (2007). When ambiguityhurts: social standards moderate self-appraisals in general-ized social phobia. Behaviour Research and Therapy, 45,1039–1052.

Moscovitch, D. A., Hofmann, S. G., Suvak, M. K., & In-Albon,T. (2005). Mediation of changes in anxiety and depressionduring treatment of social phobia. Journal of Consultingand Clinical Psychology, 73, 945–952.

Moscovitch, D. A., Orr, E., Rowa, K., Gehring Reimer, S., &Antony, M. M. (2009). In the absence of rose-coloredglasses: Ratings of self-attributes and their differentialcertainty and importance across multiple dimensions insocial phobia. Behaviour Research and Therapy, 47, 66–70.

Orr, E., & Moscovitch, D. A. (2010a). Learning to re-appraisethe self during video feedback for social anxiety: Does depthof processing matter? Behaviour Research and Therapy, 48,728–737.

Orr, E., &Moscovitch, D. A. (2010b). Concerns about physicalappearance in social anxiety impede the therapeutic effectsof video feedback. Submitted for publication.

Osman, A., Barrios, F. X., Gutierrez, P. M., Williams, J. E., &Bailey, J. (2008). Psychometric properties of the BeckDepression Inventory-II in nonclinical adolescent samples.Journal of Clinical Psychology, 64, 83–102.


Osman, A., Gutierrez, P. M., Barrios, F. X., Kopper, B. A., &Chiros, C. E. (1998). The Social Phobia and SocialInteraction Anxiety Scales: Evaluation of psychometricproperties. Journal of Psychopathology and BehavioralAssessment, 20, 249–264.

Peters, L. (2000). Discriminant validity of the Social Phobia andAnxiety Inventory (SPAI), the Social Phobia Scale (SPS) andthe Social Interaction Anxiety Scale (SIAS). BehaviourResearch and Therapy, 38, 943–950.

Rapee, R.M., &Abbott, M. J. (2006). Mental representation ofobservable attributes in people with social phobia. Journalof Behavior Therapy and Experimental Psychiatry, 37,113–126.

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

Rodebaugh, T. L., & Rapee, R. M. (2006). Those who thinkthey look worst respond best: Self-observer discrepancypredicts response to video feedback following a speech task.Cognitive Therapy and Research, 29, 705–715.

Schlenker, B. R., & Leary, M. R. (1985). Social anxiety andcommunication about the self. Journal of Language andSocial Psychology, 4, 171–192.

Schultz, L. T., Heimberg, R. G., Rodebaugh, T. L., Schneier,F. R., Liebowitz, M. R., & Telch, M. J. (2006). TheAppraisal of Social Concerns Scale: Psychometric valida-

tion with a clinical sample of patients with social anxietydisorder. Behavior Therapy, 37, 393–405.

Spurr, J. M., & Stopa, L. (2002). Self-focused attention in socialphobia and social anxiety. Clinical Psychology Review, 22,947–975.

Stopa, L. (2009a). Why is the self important in understandingand treating social phobia? Cognitive Behaviour Therapy,38, 48–54.

Stopa, L. (2009b). Reconceptualizing the self. Cognitive andBehavioral Practice, 16, 142–148.

West, S. G., Finch, J. F., & Curran, P. J. (1995). Structuralequation models with nonnormal variables: Problems andremedies. In R. H. Hoyle (Ed.), Structural equationmodeling: Concepts, issues, and applications (pp. 56–75).Newbury Park, CA: Sage.

Wicherts, J. M., & Dolan, C. V. (2004). A cautionary note onthe use of information fit indexes in covariance structuremodeling with means. Structural Equation Modeling, 11,45–50.

Wilson, J. K., & Rapee, R. M. (2006). Self-concept certainty insocial phobia. Behaviour Research and Therapy, 44,113–136.

RECEIVED: August 31, 2009ACCEPTED: April 27, 2010Available online 16 December 2010

Documents

The Negative Self-Portrayal Scale- Development, Validation, And Application to Social Anxiety