Psychometric properties of the Self-Beliefs related to Social Anxiety (SBSA) scale in a sample of individuals with social anxiety disorder

Highlights

• Psychometrically evaluated Self-Beliefs related to Social Anxiety (SBSA) scale.

• In a sample with social anxiety disorder (SAD), the SBSA exhibited three factors.

• The SBSA had good internal consistency.

• The SBSA had good construct validity and discriminative validity.

• SBSA can be used with individuals with SAD in research and clinical contexts.

Abstract

The Self-Beliefs related to Social Anxiety (SBSA) scale assesses maladaptive social-evaluative beliefs, a key aspect in models of social anxiety disorder (SAD) that is frequently measured in research and clinical contexts. The SBSA has been evaluated psychometrically in student samples, but not in a large sample of individuals diagnosed with SAD. The current study tested the psychometric properties of the SBSA in a sample of individuals with SAD pooled from several studies (total N = 284). Results showed that the optimal factor structure for the SBSA was a correlated three-factor model (high standard beliefs factor, conditional beliefs factor, unconditional beliefs factor). The SBSA total and its subscales (formed based on the factors) exhibited good internal consistency. In terms of construct validity, the SBSA total, the high standard beliefs subscale, and conditional beliefs subscale had stronger associations with a measure of social anxiety than with a measure of depression, although the unconditional beliefs subscale was similarly related to both measures of social anxiety and depression. In terms of discriminative validity, the sample of individuals with SAD had higher SBSA total and subscale scores compared with a sample of individuals without SAD (N = 32). These findings provide a psychometric evidence base justifying the use of the SBSA for the assessment of maladaptive social-evaluative beliefs.

Introduction

Social anxiety disorder (SAD) is a debilitating mental disorder characterised by a severe and persistent anxiety response to potential evaluation from other people (American Psychiatric Association, 2013). Central to aetiological and maintenance models of SAD are maladaptive social-evaluative beliefs (e.g., Clark & Wells, 1995; Rapee & Heimberg, 1997; Wong & Rapee, 2016). From an aetiological perspective, several factors have been proposed (e.g., inherited tendencies, parent behaviours, peer experiences, life events, and culture) that increase the threat value assigned to social-evaluative stimuli within an individual (Wong & Rapee, 2016). Increases in the threat value for social-evaluative stimuli are proposed to be reflected through changes in an individual at the neurobiological and cognitive levels. For example, after the threat value is increased for social-evaluative stimuli within an individual, any exposure then to social-evaluative stimuli will lead to greater amygdala activity at the neurobiological level and stronger maladaptive social-evaluative beliefs at the cognitive level (Wong & Rapee, 2016). From a maintenance perspective, the activation of maladaptive social-evaluative beliefs whenever social-evaluative stimuli are encountered essentially results in a persistent anxiety response. In addition, these maladaptive social-evaluative beliefs can trigger other unhelpful cognitive and behavioural processes (e.g., anticipatory processing, attentional biases, post-event processing, safety behaviours; e.g., Clark & Wells, 1995) which ultimately reinforce the threat value assigned to social-evaluative stimuli and thus reinforce maladaptive social-evaluative beliefs.

Given the theoretical importance of maladaptive social-evaluative beliefs in the development and maintenance of SAD, currently in the literature, there are several self-report questionnaires that capture maladaptive social-evaluative beliefs characteristic of SAD in adults (see Table 1; for a review, see Wong, Gregory, & McLellan, 2016). Although these self-report questionnaires have been psychometrically evaluated, not all of these measures have had extensive psychometric evaluation specifically in samples of individuals with SAD. Given the utility of measuring the maladaptive social-evaluative beliefs of individuals with SAD in research (e.g., to further evaluate models of SAD) and clinical contexts (e.g., to determine strength of beliefs at initial assessment), it is important that instruments used to assess these beliefs have evidence of satisfactory psychometric properties when used with individuals with SAD, in line with evidence-based assessment principles (Hunsley & Mash, 2007). The 15-item Self-Beliefs related to Social Anxiety (SBSA) Scale (Wong, Moulds, & Rapee, 2014) is one such measure requiring further examination when used with individuals with SAD. The items of the SBSA were developed to assess the three types of maladaptive social-evaluative beliefs proposed in the Clark and Wells (1995) model: (a) high standard beliefs (e.g., “I have to convey a favourable impression”), (b) conditional beliefs (e.g., “If I make mistakes others will reject me”), and (c) unconditional beliefs (e.g., “People think I’m boring”).

Previous research on the SBSA has examined its psychometric properties mainly in samples of unselected undergraduates (N = 600, Wong & Moulds, 2011; N = 235, Wong et al., 2014) and only preliminary psychometric properties have been provided for a relatively small sample of individuals with SAD (N = 33, Wong et al., 2014). In terms of factor structure, the optimal model for the SBSA has been shown to be a correlated three-factor model in samples of undergraduates, with the three factors corresponding to high standard, conditional, and unconditional beliefs (Wong & Moulds, 2011; Wong et al., 2014). This factor model is therefore consistent with the conceptualisation of maladaptive social-evaluative beliefs proposed in the Clark and Wells (1995) model, and has been used to justify formation of SBSA subscales corresponding to the three belief types.

In terms of reliability, the SBSA total score and three subscale scores have all demonstrated good internal consistency in undergraduate samples (Cronbach’s alphas ranged from .80 to .94; Wong & Moulds, 2011; Wong et al., 2014). Similar internal consistencies for the SBSA total score and subscale scores have also been obtained in a small sample of individuals with SAD (Cronbach’s alphas ranged from .78 to .85; Wong et al., 2014). In an undergraduate sample, the SBSA total score and subscale scores have demonstrated good test-retest reliability over nine days (rs ranged from .72 to .82; Wong & Moulds, 2011).

With regard to construct validity, the SBSA total has been shown to have positive associations with measures of social anxiety, and these associations have been larger than associations with measures of depression (Wong & Moulds, 2011; Wong et al., 2014). Interestingly, the SBSA subscales corresponding to high standard beliefs and conditional beliefs have been shown to be more strongly related to measures of social anxiety than to measures of depression, whereas the SBSA subscale corresponding to unconditional beliefs has been shown to be similarly related to measures of social anxiety and depression (Wong & Moulds, 2011; Wong et al., 2014). In terms of discriminative validity, a small sample of individuals with SAD has been shown to score significantly higher on the SBSA and its subscales compared to an undiagnosed sample of undergraduates (Wong et al., 2014). It is important to consider these validity results from a theoretical perspective. Given the role of maladaptive social-evaluative beliefs in driving the anxiety response in social situations in models of SAD (e.g., Clark & Wells, 1995), it would be expected that higher scores on the SBSA and its subscales would be related to greater social anxiety severity, and this is reflected in the previous construct validity and discriminative validity results of the SBSA. In addition, given the emphasis on social-evaluative concerns in the content of SBSA items (e.g., negative evaluation from other people; need to act in accordance with high social standards), it would be expected that scores on the SBSA and its subscales would be positively related with social anxiety severity, more so than with other negative emotional states where social-evaluative concerns may not be as salient, such as depression. Although this expectation is reflected in the majority of the previous construct validity results, it is interesting that scores on the unconditional beliefs subscale of the SBSA have been shown to be similarly related to social anxiety and depression. To explain this result, it may be the case that unconditional beliefs, by emphasising social-evaluative concerns that are absolute and global in nature (e.g., “People think I’m boring”), actually possess characteristics of depressive thinking (Dent & Teasdale, 1988; Kovacs & Beck, 1978), resulting in their similar relationship with social anxiety and depression.

Without a psychometric evaluation of the SBSA in a large sample of individuals with SAD, there is currently only preliminary evidence to evaluate whether it is appropriate to use this measure with individuals with SAD, derived from undergraduate samples and a small sample of individuals with SAD (e.g., Wong et al., 2014). Thus, the current study aimed to evaluate the psychometric properties of the SBSA in a large sample of individuals diagnosed with SAD, including factor structure, internal consistency, construct validity, and discriminative validity. Factor structure of the SBSA was examined to determine whether the factors underlying the measure in a sample of individuals with SAD are consistent with the Clark and Wells (1995) model of SAD and their proposal of three types of maladaptive social-evaluative beliefs (high standard beliefs, conditional beliefs, unconditional beliefs). Evidence of such a factor structure would also justify the formation of SBSA subscales corresponding to these belief types when using the measure with individuals with SAD. Despite the potential of subscales, we note that researchers at times calculate an SBSA total score as an overall index of the severity of maladaptive social-evaluative beliefs (e.g., Gregory, Wong, Marker, & Peters, 2018; Shikatani, Antony, Kuo, & Cassin, 2014). As such, evaluation of other psychometric properties of the SBSA in the current study includes consideration of the SBSA total score and SBSA subscales formed. Internal consistency of the SBSA was examined to determine in a sample of individuals with SAD the extent to which SBSA item scores agree with each other, and whether SBSA item scores on the same subscale also agree with each other. Construct validity of the SBSA was examined to determine: (a) whether the SBSA and its subscales would be associated with social anxiety more so than with depression in a sample of individuals with SAD, given the emphasis of SBSA items on social-evaluative concerns and the conceptualisation of maladaptive social-evaluative beliefs in models of SAD (e.g., Clark & Wells, 1995), and (b) whether an exception to this is the unconditional beliefs subscale of the SBSA, which should be similarly associated with social anxiety and depression in a sample of individuals with SAD given the emphasis of unconditional beliefs on social-evaluative concerns that are absolute and global in nature. Discriminative validity was examined to determine whether individuals with SAD would score higher on the SBSA and its subscales compared to individuals without SAD, again in line with the emphasis of SBSA items on social-evaluative concerns and the conceptualisation of maladaptive social-evaluative beliefs in models of SAD (e.g., Clark & Wells, 1995).

Based on previous theory and research (e.g., Clark & Wells, 1995; Wong et al., 2014), we had the following hypotheses. First, we hypothesised that within a confirmatory factor analysis (CFA) framework, a correlated three-factor model (three factors corresponding to high standard beliefs, conditional beliefs, and unconditional beliefs) would emerge as the best-fitting model relative to competing models (i.e., a one-factor model with all items loading on one-factor, and a correlated two-factor model found in previous research with high standard beliefs loading on one factor and conditional and unconditional beliefs loading on another factor; see Wong & Moulds, 2011). Second, we hypothesised that the SBSA and its subscales would exhibit good internal consistency. Third, in terms of construct validity, we hypothesised that the SBSA and its subscales would have stronger correlations with a measure of social anxiety than with a measure of depression, with the exception of the unconditional beliefs subscale which is expected to exhibit similar correlations with a measure of social anxiety and a measure of depression. Finally, in terms of discriminative validity, we hypothesised that individuals with SAD would score higher on the SBSA and its subscales relative to individuals without SAD.