A sequential mediation model based on an emotion-focused approach for social anxiety: Traumatic childhood experiences, shame, and self-criticism

Abstract

BACKGROUND:

Instead of cognitive distortions, emotion-focused therapy emphasizes the emotion schemes of shame/deficiency in the psychopathology of social anxiety. The aim of this research was to examine the mediating role of shame and self-criticism in the relationship between traumatic childhood experiences and social anxiety using a sequential model.

MATERIALS AND METHODS:

This research was conducted in a cross-sectional descriptive design. The study sample included 420 students (175 males and 245 females (who were selected as a convenience sample from the first semester (February 2022 to July 2023). Research instruments used in the study included the Childhood Trauma Questionnaire Short Form (CTQ-SF), Young’s Shame/Deficit Schema Scale, Gilbert’s Self-Criticism Forms, and Connor’s Social Anxiety Scale. Data analyses included descriptive statistics (mean and standard deviation) and inferential statistics, which also used the Pearson correlation coefficient and structural equation model using the partial least squares regression method.

RESULTS:

The entire model is significant and 20% of the variance of social anxiety is predicted in this way (R² = 0.20, P < .001). The fit indices also showed the goodness of the model. The results demonstrate both the significant direct impact of traumatic childhood experiences on social anxiety (β = 0.178, t = 2.97, P < .001, 95% CI) and the significant indirect effects of these experiences through shame schema and self-criticism (β = 0.25, t = 6.68, P < .001, 95% CI). The total effects of these childhood traumas (direct and indirect (on social anxiety were also statistically significant (β = 0.42, t = 8.49, P < .001, 95% CI).

CONCLUSIONS:

The results confirm emotion-focused therapy model for social anxiety, which showed that self-criticism and shame mediate the relationship between traumatic experiences and social anxiety. The idea that self-criticism is a mechanism for coping with shame to avoid appearing imperfect again was also supported in this research.

Introduction

The Diagnostic and Statistical Manual of Mental Disorders − Text Revision (DSM-5-TR), fifth edition, defines social anxiety disorder (SAD) as a person’s intense fear or agitation in one or more social situations in which or where he is observed they could be judged by others.[1] Every aspect of life, including relationships, and significant functional, academic, and professional achievements can be severely affected by SAD.[2,3,4,5]

The prevalence of SAD has been estimated to range from 7 to 12% in Western countries by 43 epidemiological studies.[6] In Iran, the rate of social anxiety among students is 5.5% in females and 4.7% in males.[7] The 2010 epidemiological study in Golestan province found a prevalence rate of 10.1% (8.6% in males and 11.6% in females).[8]

The Clark and Wells model, which is based on the cognitive-behavioral approach, highlights the role that biased information processing and ineffective beliefs play in the development and maintenance of this disorder.[9,10] Cognitive models also emphasize the important role of biases in information processing (e.g., biases in attention, memory, and interpretation of ambiguous stimuli), protective behaviors, avoidance, and postevent processing in the persistence of social anxiety symptoms.[11]

The psychopathology of social anxiety mostly focuses on the sustaining and maintaining factors within prevailing cognitive-behavioral approaches, neglecting the role of primary emotions.

Derived from emotion-focused therapy, Eliot and Shahar’s theoretical model for SAD[12,13] posits that early traumatic experiences, particularly those associated with social degradation, lead to a shame-based emotional scheme (primary emotional process). This social degradation encompasses experiences such as public humiliation, bullying, and different forms of abuse, including physical, sexual, and emotional. The absence of emotional support, validation, and family warmth further exacerbates the traumatic emotional pain, resulting in fear, sadness, and shame.

The scheme leads to the development of a shame-ridden defective self, characterized by a coach/critic/guardian (CCG) aspect of self and hypervigilance, aimed at preventing the reliving of shame and humiliation. Consequently, this self-organization leads them to develop a strong sense of protection, become anxious in social situations, and engage in strict critical self-monitoring (secondary emotional process).[12,13,14]

In Lazarus and Shahar’s (2018) study,[15] high levels of social anxiety were shown to be associated with high levels of shame and self-criticism. Additionally, shame is a predictor of later self-criticism, meaning self-criticism is actually a coping mechanism for reducing shame.[15]

In one study, the association between childhood maltreatment and social anxiety was shown to be mediated by proneness to shame and self-criticism, according to a mediation model based on psycho-evolutionary and emotion-focused therapy.[16]

Considering that these structures in the psychopathology of social anxiety have been neglected in the prevailing theories for this disorder, consequently, the aim of this study is to investigate the theoretical proposition of emotion-focused therapy by examining the potential mediating role of shame-ridden/defective and self-criticism in the association between traumatic childhood experiences and social anxiety.

Materials and Methods

Study design and setting

Cross-sectional descriptive design was used in this investigation. The study population comprised all students enrolled in public universities affiliated with the Ministry of Science in Tehran during the academic year 2022-2023. The sample was chosen based on convenience from students at Shahid Beheshti University who were enrolled from the first semester (February 2022 to July 2023).

Study participants and sampling

It is typically recommended to have at least 20 samples for each latent variable when using structural equation modeling.[17] A sample size exceeding 200 is preferable for robust structural equation modeling.[18] Using Cochran’s formula, the minimum required sample size was calculated to be 381 individuals. In the present study, a total of 420 participants successfully completed the research instruments and were subjected to statistical analysis.

The mean and standard deviation of the participants’ ages were 26.4 and 6.01, with the youngest participant being 18 years old and the oldest 47 years old. Data collection methods included a traditional pen-and-paper questionnaire (105 subjects) and an online survey accessible through a specific website (https://porseshnameonline.com), whichh was distributed to participants in Telegram, WhatsApp, and Instagram groups. The student assistants who cooperate with the counseling center of Shahid Beheshti University also provided links to students in faculty groups.

Questions were designed on this website in such a way that without completing a questionnaire, one could not enter the next question, so it was ensured that the questions were fully answered. The answers of the participants can be accessed in an Excel file.

Data collection tool and technique

Demographic information, such as age, gender, education level, and marital status, were collected using a self-administered questionnaire. In this study, four other scales were used to collect information. Social Phobia Inventory (SPIN), developed by Connor et al.,[19] is a 17-item scale that assesses social phobia through three subscales: fear, avoidance, and physiological symptoms. The items are rated on a 5-point Likert scale, ranging from zero (“not at all”) to four (“extremely”). The reliability and validity of this scale are well-established in Iran, as several studies have reported.[20,21] The internal consistency of the total scale in this study with Cronbach’s alpha was 0.94.

Childhood Trauma Questionnaire-Short Form, developed by Bernstein et al.,[22] is a 28-item and 5-point Likert scale. This instrument is used to evaluate abuse across five subscales: sexual, physical, and emotional abuse, as well as physical and emotional neglect, culminating in an overall abuse score (general maltreatment).[23] The Persian version of the childhood trauma questionnaire showed acceptable psychometric properties.[24] Ebrahimi and Dejkam’s research[25] yielded Cronbach’s alpha coefficients ranging from 0.81 to 0.97 for the CTQ-SF subscales. The internal consistency of the total scale (α = 0.90) was obtained in the current study.

The 5-item shame/deficiency subscale assessing shame (questions 21, 22, 23, 24, 25) from Young Schema Questionnaire-Short Form (YSQ-SF)[26] was employed to assess levels of shame. A 6-point Likert scale is used to score the 75 items on the YSQ-SF. The scale goes from one (“Completely untrue for me”) to six (Describes me perfectly”). The Persian version of this scale was developed by Khosravani et al.,[27] and acceptable reliability and validity were reported. The internal consistency of this scale was (α = 0.94) in the current study.

Forms of Self-Criticizing/Attacking and Self-Reassuring Scale, developed by Gilbert, Clark, Hempel, Miles, and Eisener,[28,29] comprises 22 items using a 5-point Likert scale, ranging from 0, indicating “not at all like me” or “I disagree,” to 4, indicating “completely like me” and “agree” and distributed across three dimensions: Inadequate Self, Hated Self, and Reassure Self.[28] Cronbach’s alpha was 0.90 for self-inadequacy and 0.86 for self-hatred and self-assurance.[28]

The Iranian version of this scale was translated by Ghahramani et al.[30] and standardized on students and the SCRS factors’ Cronbach’s alpha coefficient fell within the desired range: self-hatred (0.73), self-reassurance (0.75), and self-inadequacy (0.89).

Ethical consideration: This research was approved by the ethics committee of the University of Rehabilitation Sciences and Social Health (code of ethics: IR.USWR.REC.1399.085). The participants in the research study gave their informed consent.

In this study, the participants were granted full autonomy to engage in the research process. There was informed consent at the beginning of the designed questionnaires, where they ticked the “I agree” option and then entered the questions. All of them were assured that the obtained information will only be used in a research and confidentiality will be respected.

Data analysis

The initial phase of the study involved the utilization of statistical measures such as mean, standard deviation, frequency, and frequency percentage, facilitated by SPSS-24 software. To examine the mediating role of shame schema and self-criticism in the association between traumatic childhood experiences with SAD, Pearson correlation coefficient and structural equation modeling were used using the partial least squares regression method with SmartPLS-SEM v3.3.7.

Results

According to the data presented in Table 1, the research sample comprised 41.7% male and 58.3% female participants. The distribution of marital status among the participants was as follows: 71% were single, 28% were married, and 7% were divorced. Regarding educational attainment, 49% held a bachelor’s degree, 36% possessed a master’s degree, and 15% had obtained a doctorate.

Table 1.

Demographic characteristics of the sample

Variable		Frequency % of Total		SAD				SS				SC				CT
				M (SD)		P		M (SD)		P		M (SD)		P		M (SD)		P
Gender						0.84				0.71				0.55				0.91
Male		175 (% 41/7)		26.0 (15.5)				13.0 (6.3)				36.7 (19.4)				72.4 (23.01)
Female		245 (% 58/3)		26.1 (3.3)				13.3 (6.6)				35.4 (5.7)				72.6 (24.2)
Marital Status						0.92				0.72				0.51				0.16
Single		299 (%71/2)		26.2 (15.6)				13.1 (6.5)				36.0 (20.1)				72.6 (23.5)
Married		114 (%27/1)		25.8 (15.3)				13.2 (6.6)				36.4 (20.2)				73.3 (24.3)
Divorced		7 (%1/7)		28.1 (20.3)				11.8 (3.3)				27.4 (11.09)				55.7 (17.2)
Education						0.08				0.06				0.21				0.13
BSc		207 (%49/3)		27.7 (16.01)				13.7 (6.1)				36. 6 (19.2)				74.5 (24.2)
MSc		150 (%35/7)		24.01 (15.1)				12.1 (6.8)				33.9 (20.4)				69.4 (23.3)
PhD		63 (% 15)		26.2 (14.9)				13.6 (6.5)				38.9 (20.1)				73.4 (23.1)

CT=Childhood Trauma; SC=Self-Criticism; S=Shame Schema; SA=Social Anxiety

There was no gender difference in social anxiety (t ₍₄₁₈₎ = 1.9, P > .05). The one-way analysis of variance test showed that there is no significant difference in social anxiety in the three educational levels (F _{(2, 417)} = 2.43; P > .05). There was no difference in the marital status between the three groups in terms of social anxiety (F _{(2, 417)} = 0.08; P > .05).

According to Pearson correlation coefficient results (see Table 2), self-criticism (r = 0.64, P = .001) shame/deficiency schema (r = 0.62, P = .001), and traumatic childhood experiences (r = 0.45 P = .001) are positively and significantly correlated with social anxiety.

Table 2.

Means, standard deviations, and correlations for the study variables along with scale reliabilities

Variables		Cronbach’s α		M		SD		2		3		4
1. Social anxiety		0.94		26.1		15.6		0.45**		0.62**		0.64**
2. Childhood trauma		0.90		72.5		23.7		1		0.72**		0.60**
3. Shame		0.94		13.2		6.5				1		0.68**
4. Self-Criticism		0.95		36.02		20.1						1

SPSS macro PROCESS was used to determine the significance of the entire model as well as the predictive variance value (R-squared value). Accordingly, it was seen that the entire model was statistically significant and explained about 20% of the variance (R = 0.45, R² = 0.20, F (1, 418) = 110.10, P < .001). Bootstrapping (n = 5,000) was used to construct 95% bias-corrected, accelerated (BCa) confidence intervals (CIs) for each effect; significant mediation (P < .05) was established if CIs did not contain zero.

The coefficient of determination (R-squared correlation) R² is specifically calculated for the endogenous variables within the model, with a value of zero assigned to exogenous constructs. A higher value associated with the endogenous constructs indicates a stronger fit of the model. Three thresholds of 0.19, 0.33, and 0.67 have been established to represent low, moderate, and high levels of the R².[31] Based on this, the following model is a good fit.

The second indicator of the predictive power of the model is the predictive correlation index or Q². This criterion, which was introduced by Stone and Geiser (1975), determines the predictive power of the model in endogenous constructs. The values 0.02, 0.15, and 0.35 indicate small, medium, and large predictive power for endogenous constructs, respectively.[32] In this study, the Q² index for all three variables of social anxiety, shame, and self-criticism was higher than 0.3.

If the SRMR index value below 0.10 or equal to 0.08 (in a more conservative version) is seen as a strong fit. Henseler et al.[33] introduced the SRMR as a measure of goodness of fit for PLS-SEM, which helps in preventing model misrepresentation.[33] Based on this, this study model is a good fit. The rest of the important indicators are listed in Table 3.

Table 3.

Model fit indices

Index (Acceptable Value)		The Obtained Value
Average extracted variance (AVE >0.50)		0.77−0.83
Composite reliability index (CR >0.70):		0.94−0.96
VIF <5 or VIF between 5 and 10		2.19−6.93
SRMR <0.08		0.08
NFI >0.90		0.85
Aston-Geisser Q² index (between 0.15 and 0.35: strong)		>0/30

According to the results obtained, traumatic childhood experiences were found to have significant positive associations with shame schema (R² = 0.54, β = 0.73, t = 24.00, P = .001, 95% CI [0.67-0.79]). In the continuation of the sequential model, shame has a significant and positive relationship with self-criticism (R² = 0.46, β = 0.67, t = 16.7, P = .001, 95% CI [0.60-0.76]). Furthermore, self-criticism was a significant predictor of social anxiety (R² = 0.46, β = 0.50, t = 8.2, P = .001, 95% CI [0.38-0.62]).

The findings show significant indirect effects of traumatic childhood experiences through shame schema and self-criticism (β = 0.25, t = 6.68, P = .001, 95% CI [0.18-0.33]) as well as a significant direct effect of traumatic childhood experiences on social anxiety (β = 0.178, t = 2.97, P = .002, 95% CI [0.60-0.75]). The total effects (direct and indirect) of childhood traumas on social anxiety were also significant (β = 0.42, t = 8.49, P = .001, 95% CI [0.33-0.53]).

A mediation model was conducted to examine the mediating role of shame and self-criticism in the relationship between traumatic childhood experiences and social anxiety, as shown in Figure 1.

Figure 1.

Mediation Model

Estimates of the effect and BCa 95% CIs from all pathways are reported in Table 4.

Table 4.

Path coefficients and indirect effects for mediation model

Variables		Path Coefficients (P)						Indirect Effects
								Bias Corrected
		To SA		To SS		To SC		Estimate		Bootstrap 95% Confidence Interval
Effects from CT to SA
CT (direct)		0.178 (.002)		0.73 (.001)
SS						0.68 (.001)
SC		0.50 (.001)
Total effects (direct and indirect)		0.42 (.001)
CT → SS → SC → SA								0.25 (.001)		0.18		0.33

CT=Childhood Trauma; SC=Self-Criticism; SS=Shame Schema; SA=Social Anxiety

Discussion

The aim of this study was to examine the mediating relationship between shame and self-criticism stemming from childhood traumatic events and social anxiety, using Elliot and Shahar’s emotion-focused approach model. The research confirmed the hypothesis that self-criticism functions as a mechanism for managing shame and contributes to the development of symptoms of social anxiety. Additionally, the study demonstrated that traumatic experiences have an indirect predictive influence on social anxiety that surpasses their direct effects.

This finding aligns with a study by Shahar, Doron, Szepsenwol,[16] which showed that shame and self-criticism mediate the relationship between childhood maltreatment experiences and social anxiety and that self-criticism is a way to cope with dealing with shame.

Furthermore, these results are consistent with the outcomes of Lazarus and Shahar,[15] which demonstrated that traumatic experiences can predict social anxiety through the mechanisms of shame and self-criticism employed as coping strategies. Asmari Brdezard et al.[34] showed that humiliating and early negative life experiences can lead to shame and predict social anxiety.

Shame is a highly unpleasant and distressing emotion, eliciting an immediate initiation of a safety mechanism (e.g., self-criticism) aimed at avoiding shame and preventing its recurrence. This is consistent with the idea of shame avoidance proposed by Schoenleber and Berenbaum.[34] They argued that concentrating only on the tendency to experience shame as a risk factor for different psychological problems is inadequate. They emphasized the significance of considering how individuals perceive shame as exceptionally distressing and undesirable. Therefore, self-criticism serves as a defensive strategy to prevent the reoccurrence of shame, even if the experience of that shame is not self-conscious. Indeed, developing a self-organization based on shame creates a weak self that is easily vulnerable, and because this fragility is unbearable and a sign of weakness, it must be strongly protected. This protection may manifest as setting high personal standards for how social interactions and relationships with others and self-blame for not meeting them.

Limitation and recommendation

One of the limitations of this research is that the sampling method was available in the nonclinical community, which limits the generalization of the results. Additionally, the design of this research was cross-sectional, which is better to use longitudinal designs for causal conclusions.

In addition, emotion-focused therapy is a neo-humanistic approach that emphasizes the phenomenal world of clients in the psychopathology model for social anxiety.

Therefore, the measurement of primary and secondary emotions and even the emotional scheme of shame, with standardized questionnaires that do not measure the unique characteristics of the individual’s phenomenal world of experience, was one of the weaknesses and serious limitations of this research.

It is suggested to investigate this explanatory and pathological model for social anxiety in qualitative research.

Conclusions

Based on the results, it was observed that shame schema and self-criticism mediate the relationship between childhood traumatic experiences and social anxiety. These results also indicated that self-criticism is developed as a coping strategy for shame to regulate it. Moreover, based on the structural equation model, it was shown that childhood traumatic experiences predict social anxiety symptoms from an indirect path (through shame and self-criticism) with a higher effect coefficient than directly (path from traumatic experiences to social anxiety).

Conflicts of interest

There are no conflicts of interest.

Funding Statement

This research was financially supported by the Psychosis Research Center, University of Social Welfare and Rehabilitation Sciences, Tehran, Iran.