According to a meta-analysis by Mayo-Wilson et al [10], interventions with shortened sessions based on the Clark and Wells model for social anxiety have demonstrated large effect sizes. Based on these findings, the expected effect size for this study was conservatively set at 0.80. A priori power analysis using the G*Power 3.1 by Faul et al [44] indicated that a total sample size of 52 would be required. Accounting for an anticipated dropout rate of 20%, the target sample size was set at 65. Accordingly, a minimum of 65 participants were targeted.

Participants were recruited online and were required to meet the following eligibility criteria: (1) current enrollment in a university program, with no restrictions on degree level; (2) availability to attend a 2-day remote workshop and willingness to be randomly assigned; (3) no suicidal ideation and no formal diagnosis of any psychiatric disorder; (4) no psychotropic medication use within the past year and no history of psychological treatment or counseling; and (5) no specific cutoff on social anxiety scores was required for inclusion. As a psychoeducational group, our primary goal was to reach individuals who experienced social anxiety−related difficulties and were motivated to make changes, regardless of their symptom severity. Therefore, although a cutoff value of the Social Phobia Inventory (SPIN) is commonly used to indicate clinically significant symptoms, we did not adopt this cutoff during recruitment. Individuals were eligible for this workshop as long as they perceived themselves affected by social anxiety. Despite the absence of a formal cutoff, participants’ baseline levels of social anxiety were relatively high. Among those enrolled, 62 (90%) of the 69 participants scored above the clinical threshold of 19 on the SPIN, with a mean score of 37.91 (SD=13.67).

Participants were randomly assigned to either the intervention group (n=39) or the waitlist control group (n=39). Prior to the start of the intervention, 7 participants from the intervention group and 2 from the control group withdrew. A total of 69 participants ultimately took part in the study (n=32 in the intervention group; n=37 in the control group).

This study received approval from the Ethical Review Committee of the Department of Psychology at Renmin University of China (IRB-24‐041). All participants provided written informed consent before participating. Individuals who completed all 4 assessments received both monetary compensation of 40 RMB (US $ 5.70) in total (10 RMB [US $1.42] per assessment) and a commemorative gift. To ensure the protection of participants’ privacy, the raw data were accessible only to the research team and numerical IDs were assigned in lieu of personally identifiable information during data handling. All reported results were fully anonymized, and no identifiable personal features are presented in the manuscript.

Simple randomization was employed to assign participants to groups. Random sequences were generated using Microsoft Excel, with half of the participants allocated to the intervention group and the other half to the waitlist control group.

Before the intervention, an online questionnaire was administered to all participants to collect baseline data (T1). The posttest (T2) was conducted 1 week after the intervention, with follow-up assessments at 1 month (T3) and 3 months (T4) postintervention. Participants in the waitlist control group received the same 2-day online workshop intervention as those in the intervention group after all data collection was completed.

This study implemented a CBT-based MBPG intervention, delivered via videoconferencing. The intervention was delivered synchronously via Tencent Meeting, an online platform similar to Zoom. All needed materials were presented in the form of slides shared on screen during the online sessions. The intervention program was named Joymaster Workshop to engage participants’ interest, reduce associated stigma, and symbolize mastery of social enjoyment. For CBT-based counseling groups targeting social anxiety, a recommended group size is approximately 6 participants [45], whereas psychoeducational groups are typically larger [30]. To avoid reductions in interaction, experiential engagement, and opportunities for practice associated with overly large groups, we set the size of the psychoeducational group at 15‐20 participants. Therefore, we divided the 32 participants in the intervention condition into 2 workshops. In allocating participants, we considered their time availability and aimed to maintain balance across the 2 groups in terms of size and gender distribution to minimize potential biases arising from group composition. The 2 workshops ultimately included 17 and 15 participants, respectively. Both workshops were delivered by the same group leaders on 2 consecutive weekends, with identical content to ensure treatment fidelity. The timing of the post-intervention and follow-up assessments was adjusted according to the specific workshop each participant attended.

The term massed refers to the delivery format, which consisted of 2 consecutive days, a total duration of 12 hours. As a brief intervention, the program selectively incorporated effective techniques from full-protocol CBT, focusing on cognitive restructuring and behavioral experiments. Contemporary CBT interventions for social anxiety typically follow either the Heimberg protocol or the Clark protocol [21]. Given the massed format and the aim of achieving effective outcomes within a short timeframe, this study’s program prioritized cognitive change over habituation through exposure, as the latter generally requires more extended treatment. Therefore, we adapted the cognitive restructuring technique from Hope and Heimberg’s model, which involves challenging automatic thoughts using supporting and opposing evidence [45].

Behavioral experiments were based on the approach by Leigh and Clark [46] as well as Hofmann and Otto [47], in which participants design and engage in social tasks to test and challenge their automatic thoughts. Behavioral experiments also enhanced the interactivity of the massed intervention and promoted participant engagement. A summary of the main intervention modules is presented in Table 1. Among these, the out-of-session behavioral experiments required participants to complete a selected task during a 2-hour lunch break, based on plans developed at the end of the morning session. Examples included asking a passerby to take a photo of them with a trash can or initiating a brief conversation with someone new in the campus canteen. All participants completed a behavioral experiment.

This intervention was conducted as a psychoeducational group rather than a counseling group. First, we employed online slide presentation as visual aids, combining didactic instruction for efficient knowledge delivery with opportunities for experiential practice of therapeutic techniques. Second, the intervention accommodated a larger number of participants than is common in counseling groups, with each workshop hosting 15‐20 individuals.

The intervention was delivered by a leader and an assistant. The leader was a PhD-level counselor intensively trained in CBT, and the assistant was a master’s student in psychology. Psychoeducational content was presented primarily through PowerPoint slides, with all procedures structured around the slide content. This ensured a highly standardized delivery and strong adherence to the intervention protocol.

The data were analyzed following the intent-to-treat principle, including all randomized participants. The Shapiro-Wilk test was employed to assess residual normality for each scale across groups and time points. Intervention effects over time were examined using linear mixed models, which are less sensitive to missing data. Given that residuals for certain variables at specific time points were skewed and the sample size was limited, a bootstrap method with 1000 resamples was applied to estimate the linear mixed model fixed-effect parameters and their 95% CIs, thereby enhancing the robustness of statistical inferences and avoiding strict distributional assumptions [57].

Cohen d was calculated to determine both between-group and within-group effect sizes. Owing to baseline differences between the intervention and waitlist control groups, Morris’ baseline-adjusted formula was applied to adjust the between-group effect sizes at posttest and follow-up [58]. The effect size was based on mean pre-post change (M_{post, T} – M_{pre, T}) in the intervention group minus the mean pre-post change (M_{post, C} – M_{pre, C}), divided by the pooled pretest standard deviation (SD_pre). All analyses were conducted using RStudio (version 4.5.0).

where the pooled standard deviation is defined as

The detailed flow of participants’ recruitment, allocation, and analysis is presented as Figure 1. No significant differences were found between the intervention and control groups in demographic variables or baseline outcome scores (see Table 2).

Shapiro-Wilk normality tests were conducted separately for the intervention and waitlist control groups at each time point for all outcome variables. The results showed that for both groups, the residuals of the 5 outcome variables across 4 time points were generally normally distributed (W=0.93‐0.98, P’s=.052-.902, n=35). Exceptions were observed in a few cases: the BFNES in the waitlist group at T1 (W=0.93, P<.05) and T4 (W=0.89, P=.004) and the DASS-21 in the intervention group at T2 (W=0.91, P<.05) and T4 (W=0.90, P<.05) and in the waitlist control group at T4 (W=0.89, P=.003). Overall, these findings suggest that the residual distributions did not exhibit significant skewness.

According to the results (Table 3), the main effects of group and time on SPIN scores were not significant. However, the group×time interaction was significant (β=−4.00, 95% bootstrap CI −6.55 to −1.22), indicating that compared to the waitlist control group, the intervention group experienced a significant reduction in social anxiety over time (see Figure 2 for the trend). Moreover, the intervention group exhibited significant reductions with large effect sizes at T2 (d=−0.81, 95% CI −1.28 to −0.33), T3 (d=−0.97, 95% CI −1.44 to −0.49), and T4 (d=−0.81, 95% CI −1.28 to −0.33; Table 4).

The main effects of group and time on BFNES scores were not significant. However, the group×time interaction effect was significant (β=−1.37, 95% bootstrap CI −2.64 to −0.08), indicating that compared to the waitlist control group, the intervention group experienced a significant reduction in fear of negative evaluation over time. However, this reduction was statistically significant only at T3, with a moderate effect size (d=−0.56, 95% CI −1.03 to −0.09).

For DASS-21 scores, neither the main effects of group and time nor the group×time interaction effect was significant (β=-1.77, 95% bootstrap CI −3.42 to 0.17). This indicates that no significant differences existed between the groups in terms of depression, anxiety, or stress symptom trajectories over time.

The main effects of group and time on SAKT scores were not significant. However, the group×time interaction effect was significant (β=.62, 95% bootstrap CI 0.05 to 1.17), indicating that compared to the waitlist control group, the intervention group showed a significant increase in social anxiety knowledge over time. Large effect sizes were observed at T2 (d=1.53, 95% CI 1.06-2.00), T3 (d=0.86, 95% CI 0.38-1.33), and T4 (d=0.98, 95% CI 0.51-1.45).

For SASI scores, the main effects of group and time, as well as the group×time interaction (β=−.43, 95% bootstrap CI −1.43 to 0.62), were not significant. This suggests that there were no significant differences between the groups in terms of social anxiety stigma over time. Similarly, no significant effects were found for ATSPPH-SF scores. The pattern of results mirrored that of the SASI, indicating no significant between-group differences in attitudes toward seeking professional psychological help over time.

This study examined the effectiveness of a CBT-based MBPG delivered via videoconference for addressing social anxiety in university students. The results showed that the intervention led to significant improvements in social anxiety, including reductions in participants’ fear of negative evaluation over a certain period and a marked increase in social anxiety knowledge. However, no significant improvements were observed in emotional distress (ie, depression, anxiety, and stress as measured by the DASS-21), social anxiety stigma, or attitudes toward seeking professional psychological help.

First, compared to the waitlist control group, participants in the intervention group reported significant reductions in social anxiety, with large effect sizes observed in both the short term (1-week posttest and 1-month follow-up) and longer term (3-month follow-up). Moreover, fear of negative evaluation significantly decreased 1 month after the intervention, consistent with the study’s hypotheses. A recent meta-analysis reported that various forms of remote CBT yield medium-to-large effect sizes in treating social anxiety [19]. While this study did not directly compare videoconference delivery to other remote CBT modalities, the observed effect sizes align with those reported in prior research on remote CBT for social anxiety, with relatively durable effects. This further suggests that core CBT techniques, particularly cognitive restructuring and behavioral experiments targeting automatic thoughts, can be effectively delivered in both in-person and remote formats, likely without significant differences in efficacy.

Second, this study employed a massed brief intervention, delivering 12 hours of intervention intensively over 2 consecutive days. High-frequency delivery of this kind has been shown to sustain participants’ motivation and engagement [36]. The selected intervention components in this study are well-established techniques for reducing social anxiety, and the massed format enables tighter integration across modules, allowing therapeutic gains to accumulate within a short period [59]. Therefore, despite involving fewer sessions than full-protocol interventions, this brief intervention yielded robust therapeutic outcomes. Given that fear of negative evaluation is a core cognitive mechanism in the development and maintenance of social anxiety [60], its reduction further supports the intervention’s efficacy. However, in this study, improvement in fear of negative evaluation was significant only at the 1-month follow-up. This suggests that the observed reduction in social anxiety may involve additional mediating mechanisms beyond cognitive change alone, which warrants further investigation in future research.

However, the intervention did not have significant effects on anxiety, depression, or stress, which was inconsistent with the study’s hypotheses. Cognitive restructuring, a core component of this intervention, is a widely applicable CBT technique and is theoretically effective for treating anxious or depressive symptoms as well [61]. Moreover, prior studies using videoconferencing group CBT for social anxiety have reported significant reductions in DASS-21 scores [40,41]. A 1-day CBT-based workshop for secondary vocational students with social anxiety also significantly reduced DASS-21 scores, with moderate effect sizes [29]. However, not all prior findings are consistent: an in-person group CBT intervention for social anxiety using a full treatment protocol found no improvements in depressive symptoms [62]. Taken together, the absence of significant change in DASS-21 scores in this study may suggest that the MBPG via videoconference has limitations in the intervention’s depth. Compared to full-protocol CBT, the brief, massed format involves fewer techniques and a considerably shortened duration. Specifically, spaced CBT interventions commonly provide opportunities for participants to practice newly learned skills between weekly sessions through homework assignments, thereby consolidating gains and facilitating symptom improvement [63]. Although the MBPG also allowed participants to practice during sessions and lunch breaks, these chances were still limited, which in turn constrained the extent to which negative beliefs and cognitive biases could be corrected. This may inherently limit its capacity to address broader emotional symptom domains beyond social anxiety or reduce the extent of therapeutic change achievable within participants. In addition, the DASS-21 primarily assesses physiological arousal and subjective emotional experiences, rather than cognitive content. As this intervention targeted participants’ cognitive biases, improvements in social anxiety were likely driven by cognition-focused changes. Such changes, however, are not readily captured by the DASS-21. Although the DASS-21 is commonly used in social anxiety interventions, it may be less suitable for detecting the specific types of cognitive changes targeted in this study.

The outcomes related to mental health literacy on social anxiety were somewhat complex, with the corresponding hypotheses only partially supported. Participants’ knowledge of social anxiety significantly improved, reflecting the most direct impact of the psychoeducational component and aligning with the study’s hypotheses. However, no significant changes were observed in social anxiety stigma or in attitudes toward seeking professional help. Prior research has shown that when individuals’ mental health literacy regarding social anxiety improves, they tend to adopt a more objective view of the condition, experience reduced stigma, and show a greater willingness to seek professional help [64]. Furthermore, Cui et al [54] conducted a 6-week, in-person psychoeducation group with the same measurements and found significant improvements in knowledge, stigma, and help-seeking attitudes. We believe several factors may account for the discrepancy between their findings and ours. First, disclosure among people with mental illness is an effective strategy to manage stigma [65], yet opportunities for interpersonal contact were highly constrained in the online format compared with in-person groups. This limitation likely hindered meaningful changes in stigma. Second, help-seeking attitudes are influenced by multiple factors, including mental health literacy and perceived service accessibility [66], and improving these attitudes requires shifts in participants’ deeper beliefs and motivations [64]. However, participants in the online workshop were recruited from diverse regions across the country, resulting in variability in access to mental health resources. For participants from low-resource backgrounds, severely limited perceived accessibility of services may have constrained improvements in help-seeking attitudes. Therefore, no significant change in help-seeking attitudes was observed at the group level. Nonetheless, these findings tentatively suggest that improvements in social anxiety symptoms may occur independently of changes in stigmatizing attitudes toward social anxiety.

This study has several notable strengths. First, it employs a videoconference platform to deliver a new format of intervention, that is, the MBPG, which offers numerous advantages. The videoconferencing delivery format overcomes geographical barriers, increasing access to evidence-based interventions. At the same time, however, online delivery inevitably reduces opportunities for in-person exposure. It must be carefully considered when selecting intervention techniques, as the effectiveness of exposure-based strategies may be more substantially limited. In this intervention, both the cognitive restructuring and behavioral experiments were cognition-targeted techniques that do not rely on exposure in vivo, thereby minimizing the potential impact of the videoconferencing format on treatment effects. The massed format—delivered over a short, tightly scheduled period—may improve efficiency and reduce dropout. Moreover, the psychoeducational group format accommodates more participants per session, and its highly structured, slide-based delivery reduces the demands on group leaders while ensuring treatment fidelity. Finally, the use of a randomized controlled trial design provides preliminary yet promising evidence for the efficacy of this intervention model.

The study also has some limitations. First, the sample size was relatively small, and we conducted our intervention only in the nonclinical sample. Future research should expand upon these findings by conducting large-scale, multicenter randomized controlled trials in both clinical and community samples to further evaluate the effects of the MBPG-based program for social anxiety. Second, videoconferencing-based CBT has inherent constraints compared to the face-to-face format. These include increased exposure to distractions, reduced attentional engagement, and limited ability for leaders to observe and monitor participant cues [67], which further reduce opportunities for direct contact and spontaneous communication both between the leaders and participants and among participants themselves. Such limitations may reduce intervention efficacy, warranting improved technological solutions and content organization to address these challenges. In addition, compared to traditional treatments, the massed brief intervention format offers limited opportunities for participants to practice newly acquired cognitive skills. As a result, these skills may not be sufficiently reinforced, which in turn constrains the depth of therapeutic change. Future iterations of the intervention protocol should incorporate additional strategies to address this limitation. Finally, the follow-up period was limited to 3 months. Longer-term follow-up (eg, 6 months, 1 year, or beyond) is necessary to assess the maintenance of treatment gains over time.

We finally derived several implications from this intervention for future research. First, although online delivery is convenient and lowers access barriers, it is important to select a platform that supports essential interactive features, particularly breakout rooms and spaces that allow participants to communicate more freely, which may partially mitigate the limitations associated with reduced in-person exposure. Second, because the content of this workshop was delivered in a highly structured, slide-based format, future dissemination could broaden not only the reach of the program but also the range of potential leaders. Beyond mental health professionals, trained paraprofessionals or nonspecialists (eg, peer counselors or students majoring in psychology or related fields) could deliver the intervention using standardized materials, such as guidance menus and slide decks, thereby enhancing the accessibility of mental health care.

This study demonstrates that a videoconferencing CBT-based massed brief psychoeducational group can produce meaningful and sustained improvements in social anxiety among university students. Future research should focus on scaling up this model through larger randomized controlled trials, adapting it for more diverse populations, and exploring the feasibility of delivery by nonspecialist leaders.