Relationships between core-beliefs, bivalent fear of evaluation, and social anxiety symptoms: a structural equation model

ABSTRACT

Objective

It is increasingly recognised that social anxiety is underpinned by a bivalent fear of evaluation. The present study tested how well a core-beliefs measure developed for social anxiety was accounted for within this bivalent model. It was hypothesised that bivalent fear of evaluation would predict more variance in social anxiety symptoms than negative alone, and that known core beliefs would predict fear of negative more than fear of positive evaluation.

Method

An online survey was completed by 346 undergraduates who were recruited in exchange for course credit, through a targeted strategy that ensured that they experienced social anxiety symptoms. The survey was completed on Qualtrics, with all questions appearing in a standardised order.

Results

The hypothesised model explained 45.4% of social anxiety symptom variance. Core-beliefs explained unique variance in fear of positive and negative evaluation but had a larger relationship with fear of negative evaluation.

Conclusions

These results suggest that different core-beliefs underpin fear of positive and negative evaluation. Better knowledge of the core-beliefs underpinning fear of positive evaluation may inform the development of more effective psychological treatments

KEY POINTS

What is already known about this topic:

1. Bivalent fear of evaluation is increasingly recognised in social anxiety.

2. Fear of positive evaluation uniquely contributes beyond negative evaluation.

3. Core belief measures exist based on fear of negative evaluation theories.

What this topic adds:

1. Established core beliefs do not well explain variance in positive evaluation.

2. Only conditional beliefs explain variance in both fear of negative and positive evaluation.

3. Further research is needed on core beliefs related to fear of positive evaluation.

KEYWORDS:

• Bivalent fear of evaluation
• social anxiety
• fear of positive evaluation
• core beliefs
• structural equations modelling
• fear of negative evaluation

Social anxiety disorder (SAD) is characterised by significant discomfort with, and avoidance of, socially-evaluative situations (American Psychiatric Association [APA], 2013). Around 12% of the population will experience SAD in their lifetime (Kessler et al., 2005) and around 4.7% struggle with SAD each year (Australian Bureau of Statistics, 2007). SAD is often comorbid with other anxiety disorders and depression (Koyuncu et al., 2019). Social anxiety is associated with lower self-esteem (Iancu et al., 2015), and also greater social withdrawal and lower peer acceptance (Barzeva et al., 2020). Of all anxiety disorders, SAD is associated with the highest rate of unemployment (Moitra et al., 2011). Fear of positive and negative evaluation also predicts lower quality of life in individuals with SAD (Dryman et al., 2016). Given the high prevalence, functional impairment, and chronicity of this disorder, it is important to understand the underlying processes in SAD to further develop targeted treatments.

The bivalent fear of evaluation (BFOE) theory of SAD suggests that social anxiety symptoms are underpinned by a co-occurring fear of positive and negative evaluation (Weeks & Howell, 2012, 2014). In this perspective, the long-recognised fear of negative evaluation (FNE) is integrated with a simultaneous fear of positive evaluation (FPE). An evolved threat-detection sensitivity towards both leaving unfavourable social impressions to avoid risking group exclusion (Clark & Wells, 1995; Gilbert, 2001; Rapee & Heimberg, 1997; Watson & Friend, 1969) and also leaving overly favourable social impressions to avoid risking unnecessary social conflict and competition, is thought to be a product of evolution (Trower & Gilbert, 1989; Weeks & Howell, 2014). This evolved threat system then becomes entangled by patterns of thinking. In the first iteration of their cognitive theory of social anxiety, Rapee and Heimberg (1997) proposed that social situations activate social core-beliefs, which in turn set off thoughts related to negative fear of evaluation that then trigger social anxiety symptoms. However, Weeks and Howell (2012) argue that fear of evaluation is best understood as a co-occurring fear of both positive and negative evaluation, to which Heimberg et al. (2010) updated the Rapee and Heimberg (1997) to more broadly encapsulate evaluation in general, instead of negative evaluation specifically.

The evidence-base for the BFOE is still emerging. With regard to cross-sectional data, a recent meta-analysis found consistency across studies that BFOE accounts for more variance in social anxiety symptoms than FNE alone, confirming FPE as a distinct construct that explains an additional 9% of unique variance in social-observational anxiety symptoms on the Social Phobia Scale (SPS) and 6% with social-interaction anxiety on the Social Interaction Anxiety Scale (SIAS; Mattick & Clarke, 1998) when considered with FNE, beyond the 33% (SPS) and 41% (SIAS) of variance accounted for by FNE alone, showing FPE to have a greater role with SPS compared to SIAS (Cook et al., 2022). The evidence for longitudinal studies is in its infancy and mixed. Reichenberger and Blechert (2018) suggests that FPE is independent, and not delayed, FNE, based on the longitudinal findings of Rodebaugh et al. (2012), which conclude FPE is an independent construct. More recently, Johnson et al. (2020) found in the context of an FNE-focused intervention RCT that FNE better accounted for the changes in SIAS individually than with FPE. However, there are limitations to this study; the SPS was not used, which has a stronger relationship with FPE than SIAS (Cook et al., 2022), and it needs to be replicated with an FPE-focused intervention (e.g., Weeks et al., 2020) to control for the effects of the FNE-focused intervention. To date, no research has focused on core-beliefs with regard to FPE vs FNE.

SAD is also underpinned by a range of maladaptive core-beliefs, specifically in regard to social-evaluative situations (Wong et al., 2014). Beck and Clark (1988) defined core beliefs for anxiety and depression as the enduring and underlying assumptions of the world within the cognitive system, that guide how information and emotions are processed. Core beliefs are important to consider, as the way information is processed argued to maintain symptoms of social anxiety in the cognitive model of SAD (Heimberg et al., 2010; Rapee & Heimberg, 1997; Wong et al., 2014). For SAD specifically, there are several major theoretical perspectives that describe how information is processed in a way that emotions related to fear of evaluation and avoidance. These include by Hofmann (2007) (i.e., perceiving being unable to meet unrealistically high standards for oneself), Rapee and Heimberg (1997) (i.e., the overestimated likelihood and cost of negative evaluation), and Clark and Wells (1995) (i.e., maladaptive beliefs that become activated in a socially-evaluative situation and make the situation appear more dangerous than it actually is). The application of these FNE models in cognitive behavioural therapy (CBT) were found to be efficacious in the treatment of SAD in a recent meta-analysis, with improvements maintained for over 12 months, and also with benefits in secondary outcomes such as self-esteem and quality of life (Kindred et al., 2022). FPE-focused CBT also looks promising, but it is still in its infancy (Weeks et al., 2020).

Wong et al. (2014) quantified the theories by Clark and Wells (1995), Hofmann (2007), and Rapee and Heimberg (1997) and developed them into a validated scale: The Self-Beliefs Related to Social Anxiety scale (SBSA). The SBSA contains three subscales that represent core beliefs: High Standards where a perceived standard of behaviour must remain high enough to avoid negative evaluation, Conditional Beliefs where negative evaluation will occur unless an ideal is met, and Unconditional Beliefs which involve assumed negative evaluation over a characteristic they have (Wong et al., 2014). However, as this measure was designed to capture self-beliefs primarily related to FNE theories, it is unclear how relevant these core beliefs are for symptoms related to FPE. Cognitions developed for FPE includes concerns for social reprisal (Weeks et al., 2015) and disqualifications of positive social outcomes (Weeks, 2010), which are quite different to the abovementioned FNE core beliefs. Therefore, it is likely that the core beliefs associated with FNE and FPE would differ. A recent review also suggests that FPE may undermine cognitive behaviour therapy by positive feedback being potentially threatening without appropriately addressing the underlying cognitions of FPE (Reichenberger & Blechert, 2018). Thus, it is important to empirically test the relationship between underlying core beliefs for both FNE and FPE.

The present study aims to confirm a structural equation model (SEM), based upon the BFOE theory of social anxiety symptoms (Cook et al., 2019; Weeks & Howell, 2012; Weeks et al., 2008), and extend previous research to also account for the role of core-beliefs (Wong et al., 2014). Consistent with BFOE theory, it was hypothesised that BFOE (i.e., FPE and FNE) will account for the unique variance in social anxiety symptoms beyond the contribution of FNE alone. It was also hypothesised that social-anxiety-related core-beliefs would account for variance in BFOE, which would also account for variance in social anxiety symptoms, and that this model will fit the data in SEM (see Figure 1 for the operationalised conceptual model), consistent with the theory that core-beliefs account for a portion of the degree to which individuals fear of evaluation, and that fear of evaluation also accounts for a portion of the variance in social anxiety symptoms (Clark & Wells, 1995; Hofmann, 2007; Rapee & Heimberg, 1997; Wong et al., 2014). Finally, it was hypothesised that in line with the strong grounding in the FNE tradition literature, the core-beliefs construct (Wong et al., 2014) would have a larger association with FNE than FPE.

Figure 1. Conceptual bivalent fear of evaluation model of SAD.

Methods

Participants

This sample consisted of 346 psychology undergraduates, recruited between July and October 2018. As described by Connor et al. (2001) a cut-off score of six or greater on the Mini-SPIN was used to classify cases as “probable SAD” and thus only respondents whose responses exceeded this cut-off were included in the analysis from an original pool of 575 respondents. This was done to make a clinical-analogue sample and improve the relevance and validity to SAD. After the cut-off was applied, the mean Mini-SPIN score was 8.40 with a standard deviation of 1.96. The participants were 83.2% female (15.6% male; 0.9% non-binary), 74.6% single, 59.5% unemployed (38.2% part-time employed), 39% born in Australia (33.5% born in China), 69.7% heterosexual (14.7% homosexual), and had a mean age of 19.6 (SD 3.6) years. This study was approved by The University of Melbourne Human Research Ethics Committee (Ethics ID: 1851937).

Procedure

Students were recruited via the Research Experience Program and received credit towards their course following participation in the online survey. A targeted recruitment strategy was used, and students were encouraged to participate in the study if they experienced social anxiety symptoms. The study questionnaire was hosted on the University Qualtrics platform, and all measures were presented to participants in a standardised order. Being an online survey, participants were able to complete the survey at their convenience.

Measures

Mini-SPIN

Mini Social Phobia Inventory (Mini-SPIN; Connor et al., 2001) contains three items and is a brief screening tool for generalised SAD (e.g., “fear of embarrassment causes me to avoid doing things or speaking to people”). Items on Mini-SPIN were measured using a five-point rating scale (0–4; “not at all characteristic or true of me” to “extremely characteristic or true of me”), with greater scores indicating greater social anxiety. As described above, the Mini-SPIN was used to determine inclusion in the final study sample.

Social Phobia Scale

Social Phobia Scale (SPS; Mattick & Clarke, 1998) contains 20 items and measures social anxiety symptoms related to social evaluation (e.g., “I feel awkward and tense if I know people are watching me”). SPS items are responded to on a five-point rating scale (0–4; “not at all” to “extremely”), with greater total scores indicating more severe symptomatology. SPS focuses on the symptom domain of social anxiety related to public observation of behaviour rather than the interaction anxiety sister-scale. The SPS was included in the model as the dependent variable.

Brief Fear of Negative Evaluation Scale – Straightforward

Brief Fear of Negative Evaluation Scale – Straightforward (BFNE-S; Weeks et al., 2005) contains eight items and measures fears specific to negative evaluation (e.g., “I often worry that I will say or do wrong things”). Participants responded to BFNE-S items on a five-point rating scale (1–5; “not at all characteristic of me” to “extremely characteristic of me”), with greater total scores indicating more fear of negative evaluation. To be consistent with previous research (Cook et al., 2019; Weeks & Howell, 2012), the straightforward version was used and negatively worded items will be omitted. This measure was used to represent the FNE component of BFOE.

Fear of Positive Evaluation Scale

Fear of Positive Evaluation Scale (FPES; Weeks et al., 2008) contains eight items and measures fears specific to positive evaluation (e.g., “I generally feel uncomfortable when people give me compliments”). Participants responded to FPES items on a 10-point rating scale (0– 9; “not at all true” to “very true”), with greater total scores indicating more fear of positive evaluation. This measure was used to represent the FPE component of BFOE.

Self-Beliefs Related to Social Anxiety Scale

Self-Beliefs Related to Social Anxiety Scale (SBSA; Wong et al., 2014) contains 15 items and measures core-beliefs related to social anxiety. Wong and colleagues indicated that the SBSA potentially has a three-factor structure, including subscales of high-standard beliefs (e.g., “I have to convey a favourable impression”), unconditional beliefs (e.g., “People think badly of me”), and conditional beliefs (e.g., “If I make mistakes others will reject me”). Participants responded to the SBSA on an 11-point rating scale (0–10; “do not agree at all” to “strongly agree”), with greater total scores indicating stronger socially anxious core-beliefs. This scale was used to represent core self-beliefs related to social anxiety.

Data analysis strategy

The analysis was undertaken using SPSS v25 and AMOS v25. The SEM model contained nine free parameters. Using the upper estimate of Bentler and Chou’s (1987) 5–10 cases per free parameter recommendation as a guide, a sample of at least 90 is needed. SEM assumptions regarding sample size, multivariate normality, linearity, multicollinearity (Fornell & Larcker, 1981; Tabachnick & Fidell, 2014) and directionality (Kline, 2015) were checked after screening for missing or miscoded values. A maximum likelihood SEM was used to analyse the model, with Muncks’ (1979) method used to account for measurement error using the alpha scores while summating the scales and reducing model complexity and reduce estimation bias. Model complexity was reduced in order to be able to test the SEM model and not be substantially underpowered, as a non-simplified model would require over 1,000 participants to comfortably test due to a substantially higher number of free parameters. The co-occurring BFOE theoretical correlation between FPE and FNE could not be achieved directly in SEM with FPE and FNE as they are also being dependent variables. As such, BFOE was implied and represented by a correlation between the FPE and FNE residuals. Testing the FNE-only model was done by removing the FPE construct and all FPE relationships from the analysis. Two-thousand bootstrap samples were used to calculate effect sizes and 95% confidence intervals (Kline, 2016). Directionality (Kline, 2015) was assumed in this model in-line with cognitive-behavioural theory: core-beliefs are mental representations an individual holds that account for expectations and feelings of fear of evaluation, which account for distressing social anxiety symptoms and maladaptive avoidance behaviours (Rapee & Heimberg, 1997). The significance of the difference in correlations was calculated using the tool developed by Soper (2021).

A psychometric question is raised regarding the SBSA. According to Wong et al. (2014), the items are best divided into a three-correlated-factor solution due to superior fit indices during the scale development. However, the authors also report very high correlation (r = .82) between conditional and high standards, and an even higher correlation (r = .86) between conditional and unconditional beliefs (with a r = .55 correlation between unconditional and high-standards). Given these very strong correlations, the possibility for multicollinearity exists that the constructs are too similar to be considered distinct. Thus, we see it as prudent to test both models and evaluate the psychometric merits of both conceptualisations in the present analysis.

Using Kline’s (2016) recommendation, model fit was evaluated using a Chi-square test accompanied by three fit indices, such as Comparative Fit Index (CFI; Bentler, 1990), Steiger-Lind Root Mean Square Error of Approximation (RMSEA; Steiger, 1990), and Standardised Root Mean Square Residual (SRMR; Kline, 2016). Consistent with Cook et al. (2019), Goodness of Fit Index (GFI; Byrne, 2010) will be used instead of RMSEA due to low degrees of freedom in the model (Kenny et al., 2015), and Hu and Bentler (1999) acceptable cut-off values of CFI > .95, SRMR < .08, and GFI > .90 will be used to evaluate the model.

Results

Descriptive statistics

Only the demographic variables of gender and sexual orientation were significantly correlated to the outcome variable (SPS), explaining only a potential 7.7% of variance from extraneous demographic variables. Women (33.3) scored higher on the SPS than men (24.5); F(1,340) = 16.03, p < .001, η_p² = .045 on average, and participants who identified as bisexual, asexual, or unsure (38.2) scored higher on average than heterosexual participants (30.5); F(3,342) = 3.79, p = .011, η_p² = .032. Correlations, means (M), standard deviations (SD), and internal consistencies (Cronbach alpha) are summarised in Table 1. None of the study scales were significantly skewed.

Table 1. Correlations, internal consistencies, means and standard deviations of all variables.

	Correlations
	SBSA	FPES	BFNE-S	α	M (SD)
SBSA	–	–	–	.90	71.86(28.52)
FPES	.359***	–	–	.81	34.44(13.69)
BFNE-S	.695***	.364***	–	.89	28.09(6.43)
SPS	.547***	.497***	.604***	.92	32.07(15.02)

***p < .001; SBSA = Self-Beliefs related to Social Anxiety scale; FPES = Fear of Positive Evaluation Scale; BFNE-S = Brief Fear of Negative Evaluation Scale – Straightforward; SPS = Social Phobia Scale.

Structural equation modelling

To test the conceptual model (shown in Figure 1), each latent variable had a corresponding summated scale with Munck’s method to summate scales, simplify model complexity, and a Fornell and Larcker (1981) test verified all included variables were distinct. In all cases, the constructs were considered to have greater variance than their squared correlation, suggesting that each variable was distinct and unique from every other variable (see Table 2).

Table 2. Fornell and Larckers test of distinctness to satisfy the assumption of non-multicollinearity.

	SBSA	FPES	BFNE-S
FPES
Average variance extracted	.522
Correlation squared	.185
BFNE-S
Average variance extracted	.728	.708
Correlation squared	.588	.187
SPS
Average variance extracted	.698	.677	.884
Correlation squared	.359	.310	.440

The hypothesised model was a reasonably good fit to the data (χ²(1) = 10.930, p = .001, CFI = 0.980, GFI = .985, SRMR = .029; see Figure 2) explaining 45.4% of the variance in social anxiety symptoms (95% CI: 37.6%, 52.9%). This suggests that this model is a reasonable representation of the data and explains a reasonably tight, moderate amount of variance in social anxiety. As hypothesised, considering FPE and FNE together explained more variance in social anxiety than FNE alone (45.4% vs 36.5%; an 8.9% increase) and the FNE-only model was a poorer fit to the data (χ²(1) = 17.234, p < .001, CFI = 0.959, GFI = .969, SRMR=.052). Core-beliefs had a moderate positive relationship with FNE and a weak positive relationship with FPE, which in turn FNE had a moderate positive relationship and FPE had a weak positive relationship with social anxiety symptoms, respectively, also as predicted. There was an extremely weak bivariant correlation between FPE and FNE residuals (r=.19), suggesting that FPE and FNE co-occur and both simultaneously contribute to social anxiety, although of a different relative strength between positive and negative due to the low strength of the relationship. The hypothesised core-beliefs explaining more variance in FNE (48.4%; 95% CI: 39.2%, 56.4%, p = .001), than FPE (12.9%; 95% CI: 6.5%, 21.1%, p = .001) was also supported (see Table 3 for a summary of effect sizes), and this difference in the strength of correlation was significant (Δr = .34 [.36, .70], z = 6.42, p < .01). This suggests that Wong et al. (2014) core self-belief construct had a much stronger relationship with FNE than FPE. The alternate three-correlated-factor model also was a marginally poorer fit to the data (χ²(3) = 31.873, p < .001, CFI = 0.971, GFI = .971, SRMR = .036; see Figure 3), suggesting that the SBSA scale has stronger psychometric properties in the context of BFOE social anxiety variables when used as a unidimensional scale. When comparing the correlations at the level of individual core beliefs, only the difference in strength for High Standards was significant (Δr = .50 [−0.9, .41], z = 6.89, p < .01).

Figure 2. Hypothesised SEM model with beta weights and squared multiple correlations; ***p < .001.

Figure 3. Hypothesised alternate SEM 3-factor model with beta weights and squared multiple correlations; ***p < .001.

Table 3. Effect sizes for path analysis within the structural model.

β Pathway to SPS	Effect size	95% CI Effect Size	% SPS Variance Explained
SBSA
(SBSA>FPES) x (FPES>SPS)
.360 × .319	.115
(SBSA>BFNE-S) x (BFNE-S>SPS)
.696 × .488	.340
Total effect	.454	[.387, .520]	20.6%
FPES
(FPES>SPS)
.319	.319
Total effect	.319	[.229, .416]	10.2%
BFNE-S
(BFNE-S>SPS)
.488	.488
Total effect	.488	[.403, .562]	23.8%

Values in bold denote significant effect (p < .001). Effect sizes are considered very small (.01), small (0.20), medium (0.50), and large (0.80) (Cohen, 1988; Sawilowsky, 2009). SBSA = Self-Beliefs related to Social Anxiety scale; FPES = Fear of Positive Evaluation Scale; BFNE-S = Brief Fear of Negative Evaluation Scale – Straightforward; SPS = Social Phobia Scale.

Discussion

This novel study confirmed the role of core-beliefs in a BFOE model of social anxiety using SEM. This study bridges two theoretical approaches that underpin social anxiety, with BFOE theory (Cook et al., 2019; Weeks & Howell, 2012; Weeks et al., 2008), and core-beliefs (Wong et al., 2014), extending our current understanding of social anxiety in an integrated way. The model fit the data reasonably well, finding that the relationship between core beliefs and social anxiety was indirect through bivalent fear of evaluation, and that core-beliefs as currently understood by the FNE literature are a poor predictor of FPE. Thus, the study hypotheses were supported.

The first hypothesis predicted that BFOE (i.e., FPE and FNE) would predict unique variance in social anxiety symptoms beyond the contribution of FNE alone. This hypothesis was supported as the BFOE model was a better fit to the data, and the presence of FPE accounted for an additional 8.9% of social anxiety symptoms. This finding is consistent with the meta-analysis by Cook et al. (2022), which found a 9% increase across studies. According to BFOE theory (Weeks & Howell, 2012, 2014), FPE and FNE can co-occur: not only can individuals fear negative evaluation and possible group rejection that can threaten survival (Gilbert, 2001), but they also can fear positive evaluation, possible competition and being able to maintain the more valuable status (Trower & Gilbert, 1989). Weeks and Howell (2014) suggest that socially anxious individuals simultaneously believe that they are of low value, easily ostracised and unlikely to become valuable due to resistance and reprisal. The present findings add further support to this interpretation of social anxiety and fear of evaluation, specifically regarding symptoms related to being publicly observed undertaking a task.

The second hypothesis predicted that modelling social-anxiety-related core-beliefs to predict variance in BFOE, in turn explaining social anxiety symptoms variance (see Figure 1), would fit the data. This hypothesis was supported as the model was a reasonably good fit for the data according to the fit indices. This finding is consistent with the cognitive-behavioural theory tradition, in terms of the FNE relationship. Wong et al. (2014) suggest that based upon Rapee and Heimberg’s cognitive-behavioural model (1997), core-beliefs in SAD are maladaptive beliefs that potentially inform how an individual makes an estimation of the probability and consequences of being negatively evaluated in a socially-evaluative situation. For example, “If I don’t get everything right, I’ll be rejected” (p.303). The greater the discordance between the mental representation and the perceived demands of the situation, the greater the experience of fear of evaluation (Wong et al., 2014). This was evident in the findings of our model. Core beliefs explained almost half of the variance in FNE. When individuals have stronger beliefs, for instance that they are easily rejected if they make a mistake, it explains a portion of feeling sensitised to negative evaluations in others, and these same feelings towards negative evaluation also further explain a portion of socially anxious feelings, regarding how they appear when their behaviours are observed by others.

The third hypothesis predicted that core-beliefs would have a greater effect on FNE than FPE. This hypothesis was also supported, extending our existing understanding of the relationships between core beliefs and fear of evaluation broadly, and empirically bridges the traditional CBT literature with the BFOE literature. The core-belief construct developed by Wong et al. (2014) was grounded in the FNE-only literature perspective, incorporating Hofmann (2007) model of perceiving oneself as unable to meet unrealistically high standards set by oneself, Rapee and Heimberg’s (1997) model of the overestimated likelihood and cost of negative evaluation (although it must be noted that fear of evaluation was broadened in the later model update; Heimberg et al., 2010), and Clark and Wells (1995) model of maladaptive beliefs that become activated in a socially-evaluative situation and make the situation appear more dangerous than it actually is. A theme across these models is that someone aiming higher than the perception of their level of ability, expects negative evaluation from others and this evaluation is very threatening. This theoretical basis makes sense to why so much variance was explained in FNE. However, its ability to predict variance in FPE was relatively weak by comparison. For instance, Wong et al. (2014, p. 300) describes Hofmann (2007) framework with socially anxious individuals as attempting to impress others by meeting a perceived high standard. This view would suggest that positive evaluation is desirable by socially anxious individuals. However, the BFOE theory of social anxiety suggests that socially anxious individuals feel discomfort with positive evaluation, may disqualify it and also think that it may lead to social reprisal (Cook et al., 2019; Weeks & Howell, 2012, 2014). This is consistent with the concern raised by Reichenberger and Blechert (2018) that FPE can potentially render cognitive behavioural therapy ineffective if receiving positive feedback leaves an individual feeling worse. The finding that less variance was explained in FPE relative to FNE, on the one hand, may be interpreted to reflect that FPE is less affected by core-beliefs. However, given the relative lack of empirical research on FPE, it is premature to draw this conclusion. More likely, there is a gap in our understanding of core beliefs that may specifically underlie FPE. Further research is required to identify core beliefs that sensitise individuals to fear positive evaluation.

Regarding the alternate model with a three-factor version of the SBSA employed, similar, but more nuanced considerations can be taken away, although with caution as the model had slightly weaker psychometric properties than when SBSA was used as a unidimensional-factor. Wong et al. (2014) suggest that the SBSA is best interpreted as consisting of three-constructs; high-standard beliefs, conditional beliefs, and unconditional beliefs. The authors’ rationale included that the different factors were differently related to depression compared with social anxiety. The findings of the current study are similar as the three SBSA constructs related differently to both FNE and FPE in the three-factor model. High-standard beliefs moderately predicted variance in FNE, however, did not significantly predict variance in FPE at all. Individuals fearing negative evaluation may feel they would not be able to live up to the high standards set by themselves, however, setting a high standard for oneself had no relationship with FPE. That this different relationship was found between the fears of evaluation and high standards poses new questions for further research. Conditional beliefs predicted variance in both FNE and FPE roughly equally, suggesting it could potentially contribute to both fears of evaluation and is the strongest of the three-factors. Finally, unconditional beliefs had no relationship with either FNE or FPE, rendering it irrelevant as a predictor of fear of evaluation. In all, some light is shed on some types of core beliefs predicting variance, particularly in terms of FNE, but it also raises more questions about not understanding how core-beliefs relate to bivalent fear of evaluation, supporting the overarching argument of this study that the core beliefs that potentially explain variance in FPE are poorly understood at present. However, it must be remembered that the three-factor model was a poorer fit to the data than the one-factor model, so this is potentially a less compelling conclusion at the current time.

Limitations of the present study include sampling, method and scale selection. The sampling in this study was of convenience to improve sample size and meet the power requirements, which was a great asset to the SEM analysis as clinical samples with confirmed diagnoses in this field are usually much smaller and would be comparatively underpowered to address the research question. However, the sample were not confirmed to have a SAD diagnosis by a clinician. To mitigate this, we used a validated screening tool to improve sample representativeness to as best we could make the study more relevant to clinical SAD. The methodological design was cross-sectional, and directionality was assumed. This gives us the potential to learn that constructs potentially explain variance in others due to the unique relationship between each construct in the context of considering all constructs simultaneously. This provides insight into which constructs may be useful selections in future research that could then more confidently explore the relationships causally with an experimental design. The core-beliefs scale used was selected as a validated and established measure. However, there are other constructs validated as social anxiety core-belief scales that could be analysed in a similar way. This research focused only on social anxiety symptoms related to being publicly observed (i.e., SPS scale) because this aspect of social anxiety is more relevant to FPE, according to a recent meta-analysis (Cook et al., 2022). Interaction anxiety (as measured with the SIAS for example) could also be investigated in a similar model to address the relationships with interaction anxiety symptoms, as a recent longitudinal study suggests focusing on FNE (Johnson et al., 2020), which differed from the earlier longitudinal findings that suggested both FPE and FNE (Reichenberger & Blechert, 2018; Rodebaugh et al., 2012). The temporal relationship between the development of FNE and FPE needs to be explored more thoroughly. Future research could repeat this approach with other validated social anxiety scales to confirm whether other measures predict greater variance in FPE and how they fit with different types of social anxiety symptoms. Furthermore, social anxiety and depression are often co-morbid (Koyuncu et al., 2019), depression has a positivity impairment (Dunn, 2012) similar to FPE (Weeks & Howell, 2012), and future studies should consider how the core beliefs associated with depression differentially predict variance in FNE compared with FPE.

This study has implications relevant to our current theoretical understanding of SAD. The extension to include core beliefs demonstrates how important these beliefs are to understand the over-sensitivity to negative evaluation found in socially anxious individuals. Furthermore, a gap in our understanding has become much more evident regarding FPE. That we do not presently well understand the core beliefs that sensitise individuals to positive evaluation means that these concerns are potentially unaddressed in psychotherapy. Further research is needed to address this gap in our knowledge, potentially by looking into the distinct theoretical literature associated with FPE (e.g., Weeks & Howell, 2014).

In conclusion, the present study aimed to test a model of BFOE in social anxiety that also accounted for core-beliefs. Perhaps unsurprisingly, the findings suggest that core beliefs developed from the FNE literature relate more to FNE than to FPE. While this confirms the FNE perspective, it leaves a gap in our understanding of core beliefs that are associated with FPE. FPE has a different theoretical framework associated with it and is more related to competition and conflict than to feeling socially inadequate. The finding of FNE core beliefs accounting for little variance in FPE, and specifically that the High Standards accounted for variance in FNE but not FNE, offers insight into how core beliefs as understood in the FNE frameworks have limitations in their ability to explain the belief systems associated with FPE. Therefore, there is a gap in our understanding of the core-beliefs that underpin fear of positive evaluation. Further research is needed to better understand this under-addressed aspect of SAD.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data not available – participant consent.

Additional information

Funding

This work was supported by the National Health and Medical Research Council [APP1073041]; Australian Commonwealth Government Research Training Program Scholarship.