Social Anxiety Disorder is characterised by anxiety about social or performance situations, and involves a fear of being embarrassed or negatively evaluated when under scrutiny by others (American Psychiatric Association 2013). Epidemiological research has demonstrated that social anxiety is a common disorder, characterised by early onset (Kessler et al. 2005), comorbid psychopathology (Beesdo et al. 2007), and functional impairment (Stein and Kean 2000; Wittchen and Fehm 1999).

Cognitive models of social anxiety identify a number of underlying beliefs, assumptions, and unhelpful thinking patterns hypothesised to contribute to the development and maintenance of anxiety before, during and after social or performance situations (Clark and Wells 1995; Hofmann 2007; Rapee and Heimberg 1997). The cognitive model assumes that individuals with social anxiety tend to interpret social situations as dangerous, due to dysfunctional beliefs and assumptions that they hold about themselves and others (Clark and Wells 1995; Hofmann 2007). According to Clark and Wells (1995), individuals focus on the somatic and behavioural symptoms of anxiety in social situations and these symptoms become further sources of perceived danger and contribute to the maintenance of dysfunctional cognitions held by the individual. For example, noticing symptoms of blushing or tremor are interpreted as evidence that one looks foolish to others, and will be evaluated as such. Similarly, Rapee and Heimberg (1997) argue that attention during social situations focuses on internal sensations and non-overtly positive external information, such as neutral facial expressions, which are interpreted negatively. Individuals become preoccupied with these internal sensations and negative thoughts, which further maintain symptoms of anxiety. In addition, individuals assume that their view of themselves reflects how others see them (Clark and Wells 1995). Consequently, a processing bias is established, such that negative beliefs about how one appears to others develops from self-perceptions, rather than from objective feedback. As a result, negative cognitions and anxious symptoms are further reinforced.

Clark and Wells (1995) established three categories of dysfunctional beliefs, which they proposed to be held by individuals with social anxiety. First, they argue that socially anxious individuals hold excessively high standards for their social performance. These standards lead to unhelpful thinking patterns, as individuals set standards too high to possibly achieve, resulting in concern over failure. Similarly, Rapee and Heimberg (1997) describe the discrepancy between the high standards held for social performance and an individual’s belief that they are not able to attain such standards as problematic in social anxiety. Second, individuals with social anxiety have conditional beliefs concerning social evaluation. For example, they are hypothesised to assume that if someone does not like them, they are inherently flawed in some way. Finally, unconditional beliefs about the self are held as a consequence of negative beliefs about their own worth. These unconditional beliefs involve ‘I am’ statements, such as ‘I am unworthy’ or ‘I am boring’. Clark and Wells (1995) suggest that these three categories of dysfunctional beliefs maintain anxiety and associated avoidance and safety strategy use.

Anticipatory processing is also evident in individuals with social anxiety, with thoughts prior to a socially threatening situation revolving around recollections of past perceived social failures, negative predictions about future social performance and the perceived consequences of such failure (Hofmann 2007). The anticipation of failure in future social situations results in increased negative perceptions of self, symptoms of anxiety and poorer performance in social settings (Clark and Wells 1995; Rapee and Heimberg 1997). As a consequence of negative self focused attention, individuals encode more threatening cues in social situations, thereby leading to anxious symptomatology and behaviours, such as avoidance, which further reinforce their negative view of self. In support of this theory, research has found that individuals who completed a modified dot-probe task and engaged in anticipatory processing were found to turn their attention inward toward physiological symptoms, which were perceived as threatening (Mills et al. 2014). Similarly, Hinrichsen and Clark (2003) found that participants high in social anxiety were more likely to engage in anticipatory anxiety and that these processes led to heightened levels of anxiety during a speech task, compared to controls. Moreover, compared to distraction, anticipatory processing in socially anxious individuals has been associated with greater levels of self-reported anxiety, increased skin conductance, stronger conditional and high standard beliefs and a greater negative self-image (Vassilopoulos, 2005; Wong and Moulds 2011). These results distinguish the mental processes socially anxious individuals engage in compared to controls, and suggest that these processes maintain anticipatory processing in future social situations.

Similarly, following a social situation, individuals engage in post-event processing where they ruminate about their perceived failure and focus on negative images of themselves (Clark and Wells 1995). Post-event rumination has been found to maintain state social anxiety following a speech task in both clinical (Abbott and Rapee 2004) and student populations (Wong and Moulds 2009). Socially anxious individuals have been found to engage in more post-event processing than controls and recalled more negative information about the event and oneself (Mellings and Alden 2000; Rachman et al. 2000). Furthermore, high socially anxious individuals tend to engage in greater negative rumination as a maladaptive coping response to negative social events compared to low socially anxious individuals (Kocovski et al. 2005). Consistent with these findings, studies have reported that socially anxious individuals engage in greater rumination and recall more negative information about oneself the day after a social event (Mellings and Alden 2000; Dannahy and Stopa 2007) and in the week following a speech task (Abbott and Rapee 2004; Penney and Abbott 2015; Perini et al. 2006). This tendency to engage in negative post-event rumination maintains individuals’ negative mental representation as a social being and leads to negative predictions about future social situations (Brozovich and Heimberg 2008), essentially maintaining the vicious cycle of SAD.

Experimental paradigms have been used to demonstrate the association between social anxiety and cognitive biases. Stopa and Clark (2000) found that individuals with social anxiety were more likely than controls to make negative assumptions and interpretations of ambiguous social events. These individuals engaged in dysfunctional thinking by developing catastrophic consequences for these events. Mellings and Alden (2000) found that highly socially anxious individuals overestimated the visibility of anxious behaviours during a conversation with a confederate, compared to objective ratings by an independent assessor. Cognitive biases are further evident in an experiment conducted by Rapee and Lim (1992), where the discrepancy between socially anxious participants’ ratings of their performance during a public speaking task, compared to observer ratings, were significantly greater for socially anxious participants, than controls. Studies have also demonstrated self-focused attention in socially anxious participants when giving a speech, as individuals directed their attention away from an external cue toward internal cues. Tanner et al. (2006) report that individuals with social anxiety had a greater number of negative thoughts about giving a speech compared to controls. Moreover, participants with social anxiety have been found to make negative appraisals immediately after a speech task, and maintain these appraisals by engaging in negative rumination a week later (Abbott and Rapee 2004). In addition, negative biases in attention and cognition have been shown across a range of tasks such as the emotional stroop test and tasks involving detection of angry faces versus neutral or happy faces (Becker et al. 2001; Gilboa-Schechtman et al. 1999; Mogg et al. 2004). It is therefore evident that individuals with social anxiety have greater cognitive biases than controls.

Self-report measures of dysfunctional thinking styles in socially anxious individuals compliment experimental research aimed at assessing such biases. Assessment of cognitive factors in research and clinical settings may establish a deeper understanding of the important role that beliefs, assumptions and unhelpful thinking styles play in maintaining anxious symptomatology. It is vital that only measures with good psychometric properties are used in order to interpret assessment scores accurately, and to make valid and reliable conclusions.

Over the past decade there has been an increase in the development and evaluation of trait self-report measures aimed at assessing the beliefs and thinking styles associated with social anxiety, including perceived standards, rumination, and conditional and unconditional beliefs about oneself and others. While the psychometric properties of self-report measures of trait social anxiety have been previously analysed and evaluated (Modini et al. 2015), research to date has yet to systematically assess the psychometric properties of cognitive trait self-report measures in the domain of social anxiety. Consequently, the primary aim of this systematic review is to comprehensively identify trait cognitive self-report measures, collate and evaluate the findings of identified studies that have investigated the psychometric properties of these measures, and finally to make recommendations regarding the utility of the identified measures.

The psychometric properties of cognitive trait self-report measures of social anxiety will be evaluated using a standardised and published tool developed by Terwee et al. (2007). This is an appraisal of adequacy tool which rates the psychometric properties of measures using nine quality criteria including content validity, internal consistency, criterion validity, construct validity, reproducibility (agreement), reproducibility (reliability), responsiveness, floor or ceiling effects, and interpretability. Terwee et al. (2007) refined previous criteria available for health status questionnaires (e.g., the Scientific Advisory Committee), in order for the reader to understand what constitutes good measurement properties. While this tool may appear overly stringent, it is vital that clear criteria are evaluated in order to select a high quality questionnaire for a specific purpose. In addition, the tool has been widely and effectively used in the social anxiety literature (Modini et al. 2015) and various other domains (Burton et al. 2015; Furlan et al. 2011; Kaur et al. 2016). The aim of this paper is to evaluate trait cognitive factors of social anxiety and compare and assess the qualities of a range of cognitive measures available in order to aid researchers and clinicians in selecting measures that are interpretable and psychometrically sound.

Method

Search Strategy

The search strategy followed guidelines outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (Moher et al. 2009). Electronic databases including PsycINFO, MEDLINE and EMBASE were searched. A range of keywords relating to social anxiety, measurement, and psychometric terms including “Validation Studies” and “Factor Analysis” were used. The search strategies created for the three databases are available on request to the authors. The final search was conducted for all databases on 22/12/2015. In addition, the reference lists of all included studies were scanned to identify any additional relevant publications.

Inclusion and Exclusion Criteria

The criteria used for inclusion and exclusion was as follows:

  1. a)

    The measure had to be specifically developed in relation to cognitions, that is, beliefs, expectations, assumptions and thoughts related to social anxiety (excluding public speaking, state or trait-bases measures of social anxiety)

  2. b)

    Describe the development, validation and/or the psychometric investigation of a self-report cognitive measure of social anxiety (excluding non-psychometric studies such as a literature review or meta-analysis)

  3. c)

    Utilise an adult population (clinical or general) for development and validation purposes (excluding child and adolescent studies); and be

  4. d)

    Published in the English language (excluding non-English speaking publications) and in a peer-reviewed journal (excluding book chapters, non-peer reviewed publications and doctoral theses).

Selection Process

The selection process involved two independent reviewers (JS and MA) identifying ineligible papers by screening titles and abstracts against the inclusion and exclusion criteria, and identifying any duplicates. Of the remaining papers, the full text was obtained and analysed independently by the same two reviewers. Any discrepancies on the inclusion of papers at either stage were to be resolved by a discussion between the two reviewers. The agreement between the two reviewers was 93 %, which equates to an inter-rater agreement, κ = .84.

Appraisal of Quality

The psychometric properties of included studies were analysed by applying the criteria of adequacy for measurement properties according to Terwee et al. (2007). This appraisal tool was developed to determine the methodological quality of studies (Terwee et al. 2007). The tool refines available quality criteria for studies on the development and evaluation of health status questionnaires, and has been successfully used in previous systematic reviews (Burton et al. 2015; Modini et al. 2015).

The quality appraisal tool used in the present study assesses nine measurement properties that include: (1) content validity, (2) internal consistency, (3) criterion validity, (4) construct validity, (5) reproducibility: agreement, (6) reproducibility: reliability, (7) responsiveness, (8) floor and ceiling effects, and (9) interpretability. Table 1 outlines the definition and adequacy criteria for each measurement property, which were rated for each measure as ‘+’ for being evaluated as good, ‘?’ for being intermediately evaluated, ‘-‘ for being a evaluated as poor and ‘0’ was assigned if no information on that criterion was identified. The criteria for internal consistency was modified so that studies could use the results of past factor analyses, rather than complete a new factor analysis, or alternatively use item response theory (IRT) analyses, when calculating Cronbach alphas. Terwee et al. (2007) indicate that all measurement properties are equally important, and as such, we followed the recommendation not to provide a summary or overall score.

Table 1 Criteria of adequacy of psychometric properties (Terwee et al. 2007)
Full size table

The appraisal tool emphasises the importance of reviewing the methodological quality of a paper when assigning quality criteria ratings. If a doubtful design or method is present, for example if there is no clear description of the study design, then this paper should be given an intermediate rating (‘?’) when assigning quality criteria. This is important to take into account when assigning ratings, as it is more likely that papers with low methodological quality will report biased results.

Results

Results of Search Strategy

The initial search identified 3091 potential studies. There were 1289 duplicates, resulting in 1802 studies. Of these, 40 met the inclusion criteria. Ten additional studies were included after cross checking reference lists for articles of interest. Consequently, there were a total of 50 studies that were included in this systematic review. The selection process is summarised in Fig. 1 and studies included in this review are described in Table 3. The results of a Factor Analysis or IRT analysis carried out by the included studies are reported in Tables 2 and 3. Twenty-one measures of cognitive factors were identified. The measures covered three broad categories of cognitive constructs, including anticipatory and post-event (ASBQ, PCQ, PEP, PEP-Revised, PEPQ-R, PEP-extended, Subjective Probability (Social) Scale), threat appraisals (ASC, BFNE, FNE, FPES), and beliefs about the self or others in a social context (CONSE-Q, DPSOS, IPSM, IEQ, NSPS, SAT, SISST, SBSA, SSPS, STABS).

Fig. 1
figure 1

Flow diagram of study selection

Full size image
Table 2 Included Study Descriptions
Full size table
Table 3 Description of Questionnaires
Full size table

Assessment of Psychometric Properties

Each study was assessed for its psychometric properties in line with the criteria set out by Terwee et al. (2007). This assessment was conducted by two reviewers independently (JS and AB), with agreement in regard to criteria of adequacy being 97 %, which equates to an inter-rater agreement, κ = 0.95. Disagreements were discussed between the two raters and consensus was reached, hence, a third rater was deemed unnecessary. Each included study was rated on all nine criteria and their individual ratings are available on request to the authors. Summary ratings of all measures are displayed in Table 4.

Table 4 Overview of Ratings on Psychometric Properties
Full size table

Content Validity

Content validity refers to how accurately the construct of interest is reflected by the items in the measure. This is essential to ensure the questions are relevant to the construct (Terwee et al. 2007). Two studies presented evidence that the items were formulated after a literature study, a clear description was provided of the measurement aim and that the target population and experts were involved in item selection. These studies indicated that the SISST and the STABS possess adequate content validity.

Internal Consistency

Internal consistency refers to the extent to which items in a measure are correlated. This is essential to ensure that they are measuring the same concept (Terwee et al. 2007). Twenty-seven studies demonstrated internal consistency. These studies applied factor analysis to determine whether the items formed one or more overall scale, with a Cronbach’s alpha between 0.70 and 0.95. These twenty-seven studies indicated that the ASBQ, ASC, BFNE, CONSE-Q, DPSOS, FNE, FPES, IPSM, PEPQ, PEPQ-R, PEPQ-Revised, PEPQ-Extended, SAT, SISST, SBSA and STABS all possess adequate internal consistency. The NSPS reported a Cronbach’s alpha outside of the satisfactory margin and was therefore given a ‘poor’ rating.

Criterion Validity

Criterion validity refers to the extent to which scores on a measure relate to a gold standard. This ensures that the new measure is theoretically related to a well established measure (Terwee et al. 2007). There is no ‘gold’ standard cognitive social anxiety self-report measure with which to compare other measures. However, when an altered version of a measure is created, the original version can be said to be the gold standard and used as such for comparison purposes (Mokkink et al. 2010). One study demonstrated that an altered version is highly correlated with the original version, concluding that that the BFNE possessed adequate criterion validity.

Construct Validity

Construct validity refers to the extent to which scores on a measure relate to others measures and are in correspondence with theoretically derived hypotheses related to the construct of interest (Terwee et al. 2007). Twenty-eight studies described evidence of construct validity. These studies indicate that the ASBQ, ASC, BFNE, CONSE-Q, DPSOS, FNE, FPES, IPSM, NSPS, PEPQ, PEPQ-R, PEPQ-Revised, PEPQ-Extended, SAT, SISST, SBSA, STABS and SUBJECTIVE PROBABILITY (SOCIAL) SCALE all possess construct validity.

Reproducibility – Agreement

Agreement refers to the extent to which results on a measure remain stable over time. This ensures consistency of results (Terwee et al. 2007). Eleven studies assessed agreement, however, most reported a Pearson Correlation to demonstrate the relationship between repeated administrations of a measure. Terwee et al. (2007) consider this an inadequate method, on its own. In order to receive a positive rating according to Terwee et al. (2007), the absolute measurement error should be smaller than the minimal amount of change, hence the tool requires that the minimal important change factor (MIC) needs to be defined. However, no study defined the MIC or took systematic differences into account.

Reproducibility – Reliability

Reliability refers to the extent to which patients can be distinguished from each other. For example, in order to discriminate between high or low socially anxious participants (Terwee et al. 2007). Three studies provided an intraclass correlation coefficient (ICC). These study indicated that the FPES, SISST, and STABS possess adequate reliability.

Responsiveness

Responsiveness refers to the degree to which a measure is able to detect clinically important changes over time (Terwee et al. 2007). This is essential for example in clinical settings, in order to examine whether socially anxious symptoms are improving during or after therapy. Tweleve studies were found to be responsive, demonstrating that scores on these measures decreased significantly following treatment. The intervention in the majority of studies was cognitive behavioural therapy. Results of these studies indicated that the ASC, BFNE, CONSE-Q, FPES, IPSM, NSPS, SISST, SSPS and STABS possess responsiveness.

Floor and Ceiling Effects

Floor and ceiling effects occur if more than 15 % of respondents receive the lowest or highest score on a measure. This is essential to assess as both floor and ceiling effects limit the range of data reported by a measure (Terwee et al. 2007). One study explicitly reported that less than 15 % of the respondents achieved the highest or lowest possible scores. This study indicated that the FNE possesses floor and ceiling effects.

Interpretability

Interpretability refers to the extent to which qualitative meaning can be given to quantitative scores (Terwee et al. 2007). This is essential in order to guide the reader in interpreting and understanding scores on a measure. Five studies reported the mean scores and standard deviation for at least four relevant subgroups with an adequate sample. These relevant subgroups were typically either a community sample or participants with a different principal anxiety disorder diagnosis (for example, Generalized Anxiety Disorder). These studies indicated that the ASC, FNE, NSPS, and SISST possess adequate interpretability. This criterion in Terwee et al. is only said to possess sound interpretability if the study in isolation reported data for four individual subgroups, however, it is also possible that a measure may assess four subgroups across different studies.

The SISST received the greatest number of positive ratings, being six out of nine. The STABS received five positive ratings. Four other measures, the ASC, BFNE, FNE and FPES received four positive ratings. The PCQ, SUBJECTIVE PROBABILITY (SOCIAL) SCALE and the SSPS received the least number of positive ratings, being one out of nine. Refer to Table 4 for an overview of ratings on psychometric properties according to Terwee et al. (2007) for each measure.

Discussion

Before a cognitive self-report measure is used in research or clinical settings, it is essential that its methodological qualities are evaluated and considered adequate. This ensures that an appropriate measure, with good psychometric properties is selected when assessing cognitive aspects of social anxiety. Thus, the aim of this systematic review was to comprehensively identify cognitive self-report measures, evaluate their findings, and to make recommendations regarding their clinical usefulness. Of the twenty-one measures identified, the cognitive constructs assessed included anticipatory and post-event processing (ASBQ, PCQ, PEP, PEP-Revised, PEPQ-R, PEP-extended, Subjective Probability (Social) Scale), threat appraisals (ASC, BFNE, FNE, FPES), and beliefs about the self or others in a social context (CONSE-Q, DPSOS, IPSM, IEQ, NSPS, SAT, SISST, SBSA, SSPS, STABS).

The included studies were analysed on reported psychometric properties by applying the appraisal tool developed by Terwee et al. (2007). Results demonstrated that no cognitive self-report measure was given a positive rating across all nine domains of psychometric properties. The SISST received the greatest number of positive ratings, being six out of nine. The STABS received five positive ratings, while the ASC, BFNE, FNE and FPES received four positive ratings. It is possible to conclude that the SISST has the highest methodological quality of cognitive self-report measures of social anxiety that have been assessed to date. However, the measure lacks criterion validity, reproducibility (agreement), and no information was found on floor and ceiling effects. It is therefore important that no measure is given more support than another, until all nine areas have been more thoroughly assessed. That is, the ultimate quality for each measure cannot be truly known until research has assessed each measure against each psychometric property. Such data may lead to necessary modifications to the measures. In the instance that a number of measures have perfect ratings across all 9 criteria, it will still be useful to compare across measures in order to select an appropriate measure for specific purposes in terms of the symptom or process being measured.

The measures that received the greatest number of positive ratings addressed a range of cognitive constructs. The ASC, FPES, FNE and BFNE measure social-evaluative threat and threat appraisals. Socially anxious individuals have been shown to be hyper-vigilant to social threat and encode more threatening cues in social situations, thereby leading to anxious symptomatology and behaviours (Becker et al. 2001). The SISST and STABS measure self-statements and cognitions directly related to social situations. Previous literature has highlighted the role of dysfunctional thoughts and self-focused attention in maintaining symptoms of social anxiety (Clark and Wells 1995; Hofmann 2007).

The present review utilised a standardised and systematic approach. Original measures were included, as well as short and revised forms of measures in order to evaluate a comprehensive and wide array of trait cognitive measures of social anxiety. The review demonstrates evidence of good inter-rater reliability in regard to the included studies in the review. Additionally, measures were evaluated on their methodological quality using a standardised and previously validated quality assessment tool, which has been successfully used in previous social anxiety and health research (Burton et al. 2015; Modini et al. 2015).

When reviewing the results of the present review, it is important to take into account the limitations of the appraisal tool utilised (Terwee et al. 2007). First, the appraisal tool may be viewed as overly stringent, resulting in the low number of positive ratings for each study. For example, although many studies reported a Pearson Correlation for reproducibility (agreement), they were given an intermediate rating, as they did not report whether the measurement error was smaller than the minimal amount of change. Some may argue that this is an unnecessary stringency, as Pearson Correlation has been widely and successfully used in the past (Sainfort and Booske 2000). Less stringent criteria may allow for an overall higher number of positive ratings for each measure, resulting in stronger support of their future use. The authors of this tool themselves note that there is limited empirical data to support their quality criteria, which will likely receive further clarification with increased empirical data. It should also be noted that the use of this tool in past psychometric papers (Furlan et al. 2011; Kaur et al. 2016) is perhaps evidence that their criteria is generally well received and accepted. Moreover, the appraisal tool utilised allows the selection of high quality questionnaires and is a positive step towards identifying reliable and valid measures for a specific purpose. It further enables clear and unambiguous reporting of relevant psychometric properties. Second, many studies in this review did not report specific pre-defined hypotheses for testing construct validity, and were therefore given an intermediate rating. However, there was no clear criterion provided about how specific these hypotheses should be. Thus, decisions made about construct validity may be considered somewhat subjective. Future studies should refine the criteria by providing a clear explanation about how specific the hypotheses for testing construct validity should be. Third, there may be psychometric properties that are more or less important for a measure across the different domains, given both the setting and purpose for administering the measure. For example, for the IPSM, which measures anticipation of social situations, criterion validity may not be as important as responsiveness, as a clinician would perhaps be more interested in whether there are clinically important changes over time in a patient’s anticipation levels, as opposed to ensuring that the measure was related to a gold standard. Similarly, for the SISST, any change in patient’s positive and negative self-statements in regard to social events should be able to be detected over time in order to assess treatment outcomes. It could be argued that this is a more important criteria than floor and ceiling effects, as clinical changes are more beneficial in therapy than knowing how many respondents achieved the lowest and highest score on a measure. Similarly, for the ASBQ, construct validity may be more important in order to ensure that the assessed anticipatory processing relates to other measures and is consistent with theoretically derived hypotheses, as opposed to reproducibility (reliability), which assesses the extend to which patients can be distinguished from each other. Finally, the tool is limited, as it does not include criteria for the method and results of studies utilising Item Response Theory (IRT). Future studies can extend on the current tool by including criteria specific to IRT. Another evaluation tool that focuses on methodological quality and includes IRT analyses is being developed (Mokkink et al. 2010). When this tool is completed it would be beneficial to apply their criteria to the included studies in this review.

The results of this review suggest opportunities for future research. First, the findings highlight the lack of assessed psychometric properties for cognitive self-report measures of social anxiety. It is important to note that this does not mean that the included measures in this review are not reliable or valid tools. The results of the present review highlight the need for further investigation of the psychometric properties of the included measures to establish whether they failed to receive more positive ratings due to lacking adequacy in those domains, or whether the measure is adequate but that there is a lack of evidence providing information on these domains. Future research is needed to provide more data on the different domains of Terwee et al. (2007) identified psychometric properties, in order to make recommendations regarding their clinical usefulness. Specifically, studies should provide data on construct validity, reproducibility, and floor and ceiling effects. Second, the review is limited to an adult population. As social anxiety occurs throughout the lifespan, future studies can expand on this review by evaluating self-report measures in child and adolescent populations. This will enable the selection of psychometrically adequate cognitive measures to be used in different age groups. Third, this review is further limited as it only includes cognitive self-report measures. However, the use of state based cognitive and symptom measures, such as the Performance Questionnaire (Rapee and Lim 1992) and the State Anxiety Rating (SAR) (Rapee and Abbott 2007), as well as measures of avoidance and safety behaviours, have traditionally been used as part of a broader clinical assessment of social anxiety and it would be beneficial to assess these categories of measures in future reviews. Fourth, this review does not take into account the purpose of researchers and clinicians in selecting a particular measure. While only a psychometrically appropriate and adequate measure should be selected, a measure should also only be chosen if it best fulfils the purpose of the researcher or clinician. Finally, while a plethora of empirical studies exist that may inadvertently provide some evidence of convergent validity and internal consistency data for the measures reviewed, these papers were not included as they are not specifically aimed at assessing psychometric properties, and rather present data that may be relevant as an adjunct to their specific aims. While it was not feasible to include such papers in this review, their exclusion it is a notable limitation. However, this limitation is not a major concern given that the measures reviewed typically performed well on convergent validity and internal consistency, which are standardly reported in empirical studies that include trait cognitive self-report measures in the domain of social anxiety. The limitations of the present review mean that the results should be interpreted with caution; however, the evaluation utilised is in line with the majority of reviews evaluating psychometric properties.

The present review has attempted to provide a clear and thorough assessment of the psychometric properties of identified trait cognitive self-report measures of social anxiety. It is hopeful that the results of this review will encourage researchers and clinicians to select an evidence-based trait cognitive self-report measure. Despite no measure fulfilling all nine domains of the psychometric properties, researchers and clinicians will be able to quickly compare and evaluate the usefulness of different measures based on the identified psychometric domains that best compliment their specific aims. Future research should provide more data on the different domains to more fully determine the psychometric quality of each measure. While some measures demonstrate adequate psychometric properties, for now, there is no trait cognitive self-report measure thsat can be recommended over and above another.