Group sports games for social anxiety in university students: a randomized controlled trial revealing gender-specific efficacy

Abstract

Background

Social anxiety is a prevalent issue among college students, with many experiencing chronic anxiety in social situations and a strong fear of negative evaluation. This condition can severely impair both occupational and social functioning, leading to significant consequences.

Objective

This study aimed to evaluate the efficacy of Therapeutic Group Sports Games (TGSG) for reducing social anxiety. The TGSG intervention incorporates Yalom’s therapeutic principles into structured sports activities.

Methods

A total of 105 college students with elevated social anxiety were randomly assigned to either the Therapeutic Group Sports Games (TGSG) group (n = 53), which systematically integrated Yalom’s therapeutic principles into structured group sports activities, or an active control group engaging in standard aerobic exercise (AE, n = 52). Both interventions were delivered twice weekly for ten weeks, for 60 min per session. The primary outcomes were scores on the Interaction Anxiousness Scale (IAS) and the Brief Fear of Negative Evaluation Scale (BFNES), assessed at baseline and post-intervention.

Results

A significant time × group interaction was observed, which was moderated by gender. For female participants, the TGSG intervention demonstrated superior efficacy over aerobic exercise on both the IAS (F (1, 50) = 25.313, p < .001, partial η²= 0.336) and the BFNES (F [1,, 50] = 7.208, p = .010, partial η²=0.126), representing large effect sizes. For male participants, both the TGSG and AE groups showed significant within-group improvements over time (p < .01 for both measures), but no significant between-group differences emerged (p >.09).

Conclusions

The TGSG intervention is a highly effective, gender-sensitive approach for alleviating social anxiety in female college students. Its superiority over physical activity alone confirms that the therapeutic mechanisms reside in its structured social and psychological components. These findings offer a promising, scalable mental health intervention that can circumvent the avoidance characteristic of social anxiety.

Trial registration

Chinese Clinical Trial Registry (ChiCTR) ChiCTR2500108203. Registered on 26 August 2025. Retrospectively registered.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40359-025-03756-z.

Introduction

Social anxiety disorder (SAD), as defined in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), is characterized by a marked and persistent fear of social or performance situations in which the individual is exposed to unfamiliar people or possible scrutiny by others [1]. Core diagnostic features include fear of negative evaluation, avoidance of social situations, and significant distress or functional impairment [2, 3]. Epidemiological studies indicate that SAD is among the most prevalent mental health disorders globally, with the World Mental Health (WMH) Survey Initiative reporting a lifetime prevalence of 4.0% in the general population [4]. However, this prevalence is considerably higher among university students, with rates ranging from 23.7% to 32.8% across different regions [5, 6]. Furthermore, recent data indicate a significant increase in SAD prevalence among young adults over the past decade, a trend that is particularly pronounced in women [7, 8].

Existing literature highlights notable gender differences in the experience of social anxiety. Women generally exhibit a higher lifetime prevalence of SAD (5.67% in women compared to 4.20% in men) and tend to report greater clinical severity and subjective distress associated with SAD [9, 10]. Symptom presentation also differs, women are more likely to experience comorbid specific phobias, generalized anxiety disorder, and post-traumatic stress disorder, whereas men are more prone to comorbid substance use disorders and conduct disorders [11]. Regarding help-seeking behaviors, men with SAD may seek treatment more frequently than women, while women are more likely to receive pharmacological interventions for SAD [12]. Social anxiety is also linked to lower perceived social support, with distinct communication patterns: men show lower expressivity and precision, while women exhibit lower verbal aggression and higher emotionality [13]. Information regarding gender-specific responses to interventions, particularly group-based approaches, remains limited. However, the inclusion of social support modules in cognitive behavioral therapy has been suggested to be beneficial, especially for young, unmarried women [14]. This underscores the importance of considering gender-specific factors in both the manifestation and treatment of social anxiety.

While SAD represents the severe end of the social anxiety spectrum, contemporary mental health research increasingly embraces a dimensional perspective, viewing social anxiety as a continuum rather than a rigid categorical diagnosis. This shift highlights a critical, yet often overlooked, population: university students experiencing subclinical social anxiety. These individuals grapple with significant social anxiety and avoidance that, while not meeting full diagnostic criteria for SAD, demonstrably impairs their daily functioning. This subclinical form is considerably more widespread than diagnosed SAD on campuses, a substantial proportion of students report impairing social anxiety symptoms [15, 16]. For these students, the immediate consequences of social anxiety are profound and tangible. They frequently struggle with academic engagement, encounter significant challenges in forming meaningful social connections, and often withdraw from campus activities, all contributing to heightened psychological distress [17, 18]. Such pervasive difficulties not only compromise their current university experience but also erect substantial barriers to their future personal and professional development [19].

Given these immediate challenges and long-term risks, addressing social anxiety at the subclinical level represents a critical strategy. This approach moves beyond reactive treatment, adopting a proactive, preventative public health strategy. Early intervention offers two distinct advantages: first, it provides immediate relief and improved functioning for a large, underserved group currently struggling with academic and social impairments. Second, and equally crucial, it holds the potential to redirect developmental trajectories, preventing the escalation to chronic SAD and its associated lifelong personal and societal burdens. However, this preventative goal, particularly salient for social anxiety, confronts a formidable challenge: the pervasive pattern of avoidance. Individuals experiencing social anxiety, regardless of severity, are inherently prone to sidestepping feared social situations, which creates a substantial barrier to engaging with traditional therapeutic approaches and, critically, creates a profound therapeutic paradox. While evidence-based interventions like exposure therapy demand direct confrontation of feared social situations, the very individuals who stand to benefit most are precisely those predisposed to disengage. This inherent conflict frequently leads to high attrition rates, poor treatment adherence, and consequently, suboptimal therapeutic outcomes [20, 21].Thus, avoidance stands as a central impediment to effective intervention across the social anxiety spectrum, highlighting an urgent need for innovative approaches capable of fostering engagement and retention.

To address this critical challenge of avoidance, this study proposes and evaluates Therapeutic Group Sports Games (TGSG), a novel intervention specifically designed to mitigate this barrier. TGSG’s primary innovation lies in strategically leveraging the intrinsic appeal and structured nature of sports games to enhance participant engagement. By situating social interaction within a playful, inherently goal-oriented, and less overtly “therapeutic” context, TGSG aims to lower perceived social threat and augment intrinsic motivation, thereby facilitating a reduction in avoidance behaviors [22]. Concurrently, TGSG functions as a structured therapeutic modality, integrating key therapeutic factors from evidence-based group psychotherapy. Drawing upon principles articulated in Yalom’s influential interpersonal group psychotherapy model [23], TGSG specifically fosters curative factors such as group cohesiveness (built through collaborative team play and shared objectives), interpersonal learning (gained from real-time social interactions and feedback within the group), universality (realizing shared anxieties among peers), and the development of socializing techniques (practiced in a supportive, lower-threat environment). Within this facilitated environment, participants are also encouraged to engage in graduated social exposure. By synergistically integrating the motivational properties of games with established therapeutic principles, TGSG represents a potentially potent, accessible, and engaging treatment modality for social anxiety.

Despite the recognized efficacy of traditional group-based interventions for social anxiety [24, 25] and the growing interest in gamified or activity-based approaches for various mental health concerns [26, 27], a critical research gap persists. Rigorous empirical evaluation of an integrated, games-based approach like TGSG, which systematically combines the intrinsic motivational elements of structured sports games with established psychotherapeutic principles specifically to target avoidance in subclinical social anxiety, particularly with an emphasis on potential gender-specific outcomes, remains notably scarce. While some studies have explored physical activity or social skills training in group settings [28, 29], few have comprehensively investigated such a synergized intervention for this specific population and its nuanced gender effects. The present study aims to address this gap by investigating the efficacy of TGSG for reducing social anxiety among university students. Based on TGSG’s theoretical underpinnings and the existing literature on gender differences in social anxiety, it is also important to consider that gender can influence preferences for and engagement with group physical activities, potentially leading to varied experiences and benefits from such an intervention. Therefore, we hypothesize that: (1) The TGSG intervention will lead to a significant reduction in social anxiety symptoms from pre-test to post-test within the intervention group (2). The TGSG group will demonstrate significantly greater improvements in social anxiety symptoms compared to an active control group, attributable to its unique ability to enhance engagement and deliver integrated therapeutic elements (3). The efficacy of the TGSG intervention will be moderated by gender, with a potentially more pronounced or distinct effect observed in women, given their higher prevalence rates and unique symptomatic and help-seeking profiles.

Methods

Design and setting

This is a prospective, assessor-blinded, active-controlled, parallel-design randomized controlled trial. The study protocol was designed to recruit university students with elevated social anxiety and randomize them in a 1:1 ratio to either an experimental group (TGSG) or an active control group (AE). To prevent cross-contamination, interventions for the two groups were conducted at separate venues. Given established gender differences in social anxiety, this study employed a gender-stratified randomization design. Participants were first stratified by gender, and then randomized within their respective gender strata into either the TGSG or AE group. This ensured that all intervention groups (TGSG and AE) were single-gender, facilitating a focused examination of gender-specific effects and minimizing potential confounding from mixed-gender group dynamics. Consequently, data analyses were planned to be conducted separately for male and female cohorts to examine intervention effects within each gender. The intervention was delivered twice weekly, with each session lasting approximately 60 min, over a ten-week period. The protocol was approved by the Ethics Committee of Ocean University of China (approval number: OUC-HM-032) and this report adheres to the Consolidated Standards of Reporting Trials (CONSORT) 2010 statement.

Participants

Participants were recruited from the university-wide student population through multiple channels, including posters on campus bulletin boards, announcements via official university social media, presentations in mental health education courses, and referrals from the university counseling center. The recruitment materials detailed the study’s purpose, procedures, time commitment, potential risks and benefits, and the voluntary nature of participation. Interested students contacted the research team via email or telephone for an initial screening.

Inclusion criteria were: (1) physically healthy university students; (2) meeting the threshold for elevated social anxiety, defined as an Interaction Anxiousness Scale (IAS) score ≥ 38 or a Brief Fear of Negative Evaluation Scale (BFNES) score ≥ 32; and (3) providing voluntary written informed consent. Exclusion criteria were: (1) any medical contraindications to physical exercise; (2) current engagement in other treatments for social anxiety (pharmacotherapy or psychotherapy); (3) presence of severe depression, suicide risk, psychotic disorders, or substance use disorders; and (4) any serious physical illness that could interfere with participation.

The IAS and the BFNES were selected as primary screening instruments due to their established psychometric properties and their focus on distinct yet crucial dimensions of social anxiety. The IAS primarily assesses anxiety experienced during social interactions, reflecting behavioral and physiological aspects [30], while the BFNES focuses on cognitive concerns related to the fear of negative evaluation by others [31]. The rationale for employing both scales was to gain a comprehensive, multi-dimensional understanding of social anxiety among the subclinical population. This approach also allowed for the exploration of whether the TGSG intervention differentially impacts distinct facets of social anxiety, specifically examining its effects on interaction-based anxiety versus fear of negative evaluation.

The thresholds for elevated social anxiety (IAS ≥ 38 or BFNES ≥ 32) were carefully chosen to identify university students experiencing elevated, subclinical social anxiety, rather than those meeting full diagnostic criteria for SAD. It is acknowledged that neither scale has universally established clinical diagnostic cut-offs for SAD, particularly for subclinical populations. The selection of these thresholds was guided by existing normative data from relevant student populations: an IAS score of 38 represents the mean score for a large sample of university students (n = 1060) [32]. Similarly, a BFNES score of 32 aligns closely with the mean scores reported for non-clinical student populations in validation studies [33, 34]. Crucially, these empirically derived thresholds were further refined and confirmed through consultation with experienced mental health professionals from the university counseling center, ensuring they captured students experiencing clinically relevant levels of subclinical social anxiety that warrant intervention. By setting the threshold at or slightly above these established means for student samples, the aim was to capture individuals experiencing symptoms beyond typical levels of social apprehension, indicative of subclinical distress and functional impairment, consistent with the dimensional perspective of social anxiety and the study’s focus on early intervention.

Furthermore, the use of an “either/or” inclusion criterion (meeting the threshold on either the IAS or BFNES) was a deliberate strategy to recruit a broad spectrum of university students experiencing elevated social anxiety across its various manifestations, rather than being confined to a single, narrow symptomatic profile. This approach enhances the ecological validity of the sample for a broad-spectrum intervention like TGSG, which targets multiple facets of social anxiety. It ensures that the study population represents the diverse ways subclinical social anxiety presents in university settings, making the intervention’s potential benefits more generalizable to the broader at-risk student population.

Interventions

To ensure consistency and minimize potential facilitator effects, all TGSG and AE sessions were conducted by a dedicated team of trained facilitators. Specifically, a team of four facilitators (two with physical education backgrounds and two qualified counselors) were involved in delivering both interventions. Each facilitator was randomly assigned to lead a combination of TGSG and AE groups across the study period, ensuring that any individual facilitator’s style or expertise was balanced across the experimental and control conditions. All facilitators received comprehensive, standardized training on both intervention protocols (TGSG and AE) for two weeks prior to the study commencement, ensuring fidelity to the manuals and consistent delivery standards.

Therapeutic group sports games (TGSG)

The TGSG intervention was based on Yalom’s theory of eleven therapeutic factors in group counseling, including instillation of hope, universality, imparting information, altruism, recapitulation of the primary family group, development of socializing techniques, imitative behavior, interpersonal learning, group cohesiveness, catharsis and existential factors [23]. This study integrated these therapeutic factors into sports games, aiming to reduce participants’ social anxiety levels through the combination of physical activity and group interaction. The TGSG intervention was a structured, 10-week program consisting of 20 twice-weekly, 60-minute sessions. The intervention was theory-driven, systematically translating Yalom’s 11 therapeutic factors into a phased curriculum of sports-based games. Each 60-minute session followed a consistent three-part structure: a 10-minute Warm-up and Ice-breaker, a 40-minute Themed Therapeutic Game, and a 10-minute Guided Reflection and Processing. The detailed 10-week curriculum, progressively structured into four phases, is outlined in Table 1. Each session was guided by a detailed manual, and research team members conducted on-site monitoring using structured observation forms. Weekly meetings were held to review fidelity and address any implementation issues.

Table 1.

Therapeutic group sports games (TGSG) intervention protocol

Phase	Weeks	Core Therapeutic Objectives	Targeted Yalom’s Therapeutic Factors	Example Session Activities
1	1–2	Establishing Safety and Universality: Reduce initial participation anxiety, build group trust, and help members recognize the commonality of social anxiety struggles.	Instillation of Hope, Universality, Group Cohesiveness	“Trust Fall”: Members fall backward into teammates’ arms, directly experiencing tangible safety and challenging negative cognitions. "Common Goal”: Group transports a ball without hands, fostering shared awkwardness and initial group cohesion.
2	3–4	Information Exchange and Altruistic Experience: Practice effective communication through cooperative tasks and enhance self-worth by helping others, shifting attentional focus from internal anxiety to an external task.	Imparting Information, Altruism	“Blind Square”: Blindfolded group forms a square with a rope, practicing clear verbal communication and active listening. "Rescue Mission”: Members act as rescuers, fostering altruism and enhancing self-value through helping teammates.
3	5–8	Developing Socializing Techniques and Interpersonal Learning: Directly rehearse and learn specific social behaviors, expand social skill repertoire through observation and modeling, and receive feedback in authentic interactions.	Development of Socializing Techniques, Imitative Behavior, Interpersonal Learning	“Team Obstacle Course”: Group navigates complex challenges, revealing authentic interaction styles for interpersonal learning and providing opportunities for imitative behavior and feedback.
4	9–10	Leverage built trust and cohesion for safe emotional expression and connect in-game insights to broader life themes (e.g., choice, responsibility, personal growth).	Catharsis, Existential Factors	“Human Knot”: Group untangles a physical knot, facilitating emotional release (catharsis) and prompting reflection on agency, choice, and responsibility in life (existential factors).

Aerobic exercise

Participants in the control group engaged in structured aerobic exercise, also for ten weeks with two sessions per week, each lasting approximately 60 min to match the duration of the experimental intervention. To ensure consistency in delivery and participant engagement across both intervention arms, the same team of trained facilitators led all sessions, specifically ensuring correct exercise form and adherence to the target intensity for the control group. This study selected aerobic exercise as an active control, rather than a no-treatment or placebo condition, because exercise itself has been shown to have significant anxiolytic effects [35, 36]. Furthermore, while aerobic exercise is not typically considered a first-line treatment, several studies indicate that it can provide a beneficial effect on social anxiety symptoms [37]. Notably, a randomised controlled trial found that the magnitude of weekly symptom reduction from an aerobic exercise programme was equivalent to that of a well-regarded psychological intervention, Mindfulness-Based Stress Reduction, even if their underlying neurocognitive mechanisms of change appeared to differ [38, 39]. This robust active-control design is a key methodological strength. It allows for a more precise examination beyond the broad question of whether exercise is effective, focusing instead on the critical research question of whether the TGSG’s integrated psychosocial component provides additional clinical benefits beyond those attributable to physical activity alone.

The aerobic exercise protocol included: (1) Warm-up Phase: Including joint mobilization and light stretching (2). Main Exercise Phase: Participants performed moderate to low-intensity aerobic exercises, including brisk walking, jogging, aerobics, and aerobic dance activities. Exercise content was adjusted every 2–3 weeks to maintain participants’ interest, while maintaining similar exercise intensity (3). Cool-down Phase: Including concluding movements and stretching relaxation. To ensure physiological load was comparable between groups, participants in both the experimental and control groups wore heart rate monitors, with the target intensity for the main activity phase set at 50–63% of maximum heart rate. While participants exercised in a group setting, the protocol did not include structured group interaction or psychological processing elements. Intervention fidelity for the control group was also monitored to ensure adherence to the prescribed exercise protocol.

Control group activities did not include any intentionally designed group interaction elements or psychological counseling components. Participants primarily engaged in individual exercise, and although conducted in the same space, organized group interaction or emotional sharing was not encouraged.

Randomization, allocation concealment, and masking

A computer-generated random allocation sequence was created by a statistician independent of the research team using Research Randomizer software (www.randomizer.org). This study employed a gender-stratified block randomization. Participants were first stratified by gender, and then randomized within their respective gender strata to either the TGSG or AE group, ensuring a balanced 1:1 allocation ratio within each gender stratum. This design resulted in single-gender intervention and control groups, consistent with our aim to examine gender-specific effects. Allocation concealment was strictly maintained. The sequence was stored in an encrypted document accessible only to the independent statistician. After a participant was confirmed eligible and provided consent, the research coordinator, who was not involved in assessment or intervention delivery, requested the group assignment. The assignment was provided using a sequentially numbered, opaque, sealed envelope (SNOSE) system to prevent selection bias.

Masking (or blinding) was implemented to the fullest extent possible. Crucially, all personnel involved in outcome assessment were, and remained, blinded to participants’ group assignments throughout the entire study to protect against measurement bias. Due to the distinct nature of the interventions, it was not feasible to blind participants or the intervention facilitators to group allocation. However, to mitigate potential expectancy and performance biases, a partial blinding strategy was employed, consistent with the informed consent process. During recruitment and consent, participants were informed of the study’s general purpose—to compare two different types of physical activity programs for improving student mental health. Crucially, they were kept unaware of the specific primary outcome (social anxiety) and the specific hypotheses regarding the superiority of one intervention over the other. Facilitators were similarly trained only on their respective intervention protocols and remained unaware of the study’s primary research hypotheses.

Outcome measures

Interaction Anxiousness Scale [40]. The IAS was selected as the primary outcome measure because it directly assesses subjective social anxiety experiences independent of self-reported behaviors. This 15-item self-report measure evaluates anxiety in interpersonal interactions, specifically assessing subjective anxiety (tension and neuroticism) or its opposite (relaxation, calmness) across various potentially challenging social situations. Items are rated on a 5-point Likert scale ranging from 1 (not at all characteristic of me) to 5 (extremely characteristic of me), yielding total scores ranging from 15 to 75. Higher scores indicate greater interaction anxiety.

The Brief Fear of Negative Evaluation Scale (BFNES) [41]was used to assess fear of negative evaluation, a core cognitive component of social anxiety. The scale comprises 12 items rated on a 5-point Likert scale ranging from 1 (not at all characteristic of me) to 5 (extremely characteristic of me), yielding total scores from 12 to 60. Higher scores are indicative of a greater fear of negative evaluation. Previous studies have established the scale’s excellent reliability and validity in undergraduate populations [42].

Sample size calculation

Sample size was determined using G*Power software (version 3.1.9.7) [43] for the required sample size calculation. The following parameters were specified: repeated measures between-groups analysis of variance, large effect size (f = 0.4), significance level (α) of 0.05, statistical power (1-β) of 0.8, allocation ratio of 1:1, and 2 groups. Based on these parameters, the a priori power analysis indicated a minimum total sample size of 52 participants (n = 26 per group).

Data analysis

All data analyses were conducted using SPSS 26.0 software. First, descriptive statistical analyses were performed for all variables, including calculation of means, standard deviations, frequencies, and percentages to summarize participants’ basic characteristics and the distribution of each measured variable. The Shapiro-Wilk test was used to assess the normal distribution of primary outcome variables. To evaluate the effectiveness of randomization, independent samples t-tests were used to compare demographic characteristics and primary outcome measures between the experimental and control groups at baseline.

The primary analysis strategy was designed to rigorously evaluate gender-specific intervention effects, consistent with the single-gender intervention group design. This involved conducting separate two-way repeated measures ANOVAs for male and female participants, with intervention type as the between-subjects factor and time as the within-subjects factor. The IAS total score and BFNES total score were used as primary outcome variables. For each outcome variable within each gender stratum, the main effects between groups, main effects of time, and group × time interaction effects were examined. Additionally, to assess the overall efficacy of the TGSG intervention across all participants, a two-way repeated measures ANOVA was conducted, with intervention type as the between-subjects factor and time as the within-subjects factor.

Results

Participant characteristics

The flow of participants through each stage of the trial is presented in the CONSORT diagram (Fig. 1). A total of 200 students were assessed for eligibility, of whom 115 met the inclusion criteria and were randomly assigned to either the TGSG group (n = 57) or the AE group (n = 58). Over the course of the 10-week intervention, 10 participants (8.7%) discontinued participation (4 from the TGSG group and 6 from the AE group), with reasons detailed in Fig. 1. The final per-protocol analysis included the 105 participants (53 in the TGSG group, 52 in the AE group) who completed the post-intervention assessment.

Fig. 1.

CONSORT 2010 flow diagram

The demographic characteristics and baseline scores for the analyzed sample are presented in Table 2. There were no statistically significant differences between the experimental and control groups at baseline in terms of gender, major, only-child status, household registration, or age (all p >.05). Furthermore, no significant baseline differences were observed between the two groups for Interaction Anxiousness Scale (IAS) scores or Brief Fear of Negative Evaluation Scale (BFNES) scores for either male or female participants (all p >.05), indicating successful randomization.

Table 2.

Demographic characteristics of participants in TGSG and AE

Variables	Therapeutic Group Sports Games(N = 53)	Aerobic Exercise(N = 52)	x2/t	P
Gender (n, %)			x2=0.001	0.923
Female	26 (49.1%)	26 (50.0%)
Male	27 (50.9%)	26 (50.0%)
Major (n, %)			x2=3.219	0.073
Science/Engineering	31 (58.5%)	39 (75.0%)
Humanities	22 (41.5%)	13 (25.0%)
Only-child Status (n, %)			x2=0.091	0.762
Only child	28 (52.8%)	29 (55.8%)
Non-only child	25 (47.2%)	23 (44.2%)
Household Registration (n,%)			x2 = 0.246	0.620
Urban	27 (50.9%)	29 (55.8%)
Rural	26 (49.1%)	23 (44.2%)
Age, years (mean ± SD)	20.58 ± 1.89	20.52 ± 1.61	t = 0.18	0.86
Male Participants (mean ± SD)
IAS	41.52 ± 6.11	41.77 ± 6.09	t=−0.150	0.882
BFNES	40.96 ± 8.19	40.54 ± 9.57	t = 0.174	0.863
Female Participants (mean ± SD)
IAS	48.58 ± 5.47	47.08 ± 4.61	t = 1.069	0.290
BFNES	42.19 ± 8.63	42.46 ± 7.45	t=−0.120	0.905

Baseline social anxiety scores, as measured by IAS and BFNES, were compared between male (n = 53) and female (n = 52) participants (Table 3). Female participants reported significantly higher IAS scores(Mean = 47.82 ± 5.07), reflecting greater interaction anxiety, compared to male participants (IAS: Mean = 41.64 ± 6.04), t(103) = 5.68, p <.001. However, no significant difference was observed in BFNES scores between female (Mean = 42.33 ± 7.99) and male (Mean = 40.75 ± 8.81) participants, t(103) = 0.96, p =.341.

Table 3.

Baseline social anxiety scores by gender (Mean ± SD)

Outcome measures	Male(n = 53)	Female(n = 52)	t	p
IAS	41.64 ± 6.04	47.82 ± 5.07	5.68	< 0.001
BFNES	40.75 ± 8.81	42.33 ± 7.99	0.96	0.341

Primary outcomes

To evaluate the overall efficacy of the TGSG intervention, a series of two-way repeated measures ANOVAs were conducted with intervention type (TGSG vs. AE) as the between-subjects factor and time (pre-test vs. post-test) as the within-subjects factor. The results are summarized in Table 4. For the IAS scores, there was a significant main effect of time, F = 67.386, p <.001, Partial η² = 0.395. A significant main effect of group was also observed, F = 7.333, p =.008, Partial η² = 0.066. Crucially, a significant group × time interaction effect was found, F = 19.263, p <.001, Partial η² = 0.158. Similarly, for BFNES scores, a significant main effect of time was observed, F = 28.187, p <.001, Partial η² = 0.215. There was also a significant main effect of group, F = 7.090, p =.009, Partial η² = 0.064. Importantly, a significant group × time interaction effect was found, F = 7.705, p =.007, Partial η² = 0.070.

Table 4.

Results of two-way repeated measures ANOVA for outcome measures

Outcome measures	Time effect			Group effect			Interaction effect
	F	P	Partial η²	F	P	Partial η²	F	P	Partial η²
IAS	67.386	<0.001	0.395	7.333	0.008	0.066	19.263	<0.001	0.158
BFNES	28.187	<0.001	0.215	7.090	0.009	0.064	7.705	0.007	0.070

Descriptive statistics (mean ± SD) for IAS and BFNES scores at pre-test and post-test by gender and group are presented in Table 5 (Fig. 2). Two-way repeated measures ANOVAs were conducted separately for male and female participants to examine the effects of the interventions on IAS and BFNES scores over time (Table 6).

Table 5.

Descriptive statistics of outcome of participation

Gender	Outcome measures	Experimental group pretest	Experimental group posttest	Control group pretest	Control group posttest
Male	IAS	41.52 ± 6.11	33.79 ± 7.29	41.76 ± 6.09	38.19 ± 7.28
	BFNES	40.96 ± 8.19	30.82 ± 13.51	40.54 ± 9.58	37.31 ± 11.04
Female	IAS	48.58 ± 5.47	36.58 ± 6.84	47.08 ± 4.61	44.69 ± 4.97
	BFNES	42.19 ± 8.63	30.96 ± 8.61	42.46 ± 7.46	39.00 ± 9.53

Fig. 2.

Changes in IAS and BFNES scores by gender and intervention group from pre- to post-intervention

Table 6.

Results of two-way repeated measures ANOVA for outcome of gender

Gender	Outcome measures	Time effect			Group effect			Interaction effect
		F	P	Partial η²	F	P	Partial η²	F	P	Partial η²
Male	IAS	21.897	< 0.001	0.300	2.797	0.101	0.052	2.964	0.091	0.050
	BFNES	10.605	0.002	0.172	2.029	0.160	0.038	2.835	0.098	0.053
Female	IAS	56.650	< 0.001	0.531	7.561	0.008	0.131	25.313	< 0.001	0.336
	BFNES	25.778	< 0.001	0.340	4.811	0.033	0.088	7.208	0.010	0.126

Female participants

For female participants, the analysis revealed a strong and statistically significant time × group interaction for both primary outcomes, indicating that the therapeutic trajectory differed markedly between the TGSG and AE groups. On the IAS, this interaction was pronounced (F (1, 50) = 25.313, p <.001), accounting for 33.6% of the variance (partial η² = 0.336), a large effect. Follow-up simple effects analysis (Table 7) elucidated this finding: the TGSG group experienced a substantial and highly significant decrease in IAS scores from pre- to post-intervention (Mean Difference = −12.00, p <.001). In sharp contrast, the change observed in the AE group was minimal and not statistically significant (p =.084). Consequently, at post-test, the TGSG group’s IAS scores were significantly lower than those of the AE group (p <.001), confirming the superior efficacy of the TGSG intervention. A similar pattern emerged for the BFNES. A significant time × group interaction was found (F (1, 50) = 7.208, p =.010, partial η² = 0.126), again representing a large effect size. As detailed in Table 8, this was driven by a significant reduction in BFNES scores within the TGSG group (p <.001), while the AE group showed no significant improvement (p =.097). This resulted in significantly lower BFNES scores for the TGSG group at the study’s conclusion compared to the AE control group (p =.002).

Table 7.

Simple effects analysis for IAS scores in female participants (Significant time x group Interaction)

Comparison	Mean Difference (I-J)	Std. Error	Sigb	95% Confidence Interval for Differenceb (Lower, Upper)	Effect Size (Partial η²)	F-statistic (df = 1, 50)
Within-Group Effects (Time 2 vs. Time 1)
TGSG Group (Experimental)	−12.000*	1.351	0.000	(−14.714, −9.286)	0.612	78.849a
AE Group (Control)	−2.385	1.351	0.084	(−5.099, 0.330)	0.059	3.114a
Between-Group Effects (Experimental vs. Control)
At Pre-test (Time 1)	1.500	1.403	0.290	(−1.319, 4.319)	0.022	1.142
At Post-test (Time 2)	−8.115*	1.658	0.000	(−11.447, −4.784)	0.324	23.944

Note：*The mean difference is significant at the .05 level

^aExact statistic

^bAdjustment for multiple comparisons: Bonferroni

Table 8.

Simple effects analysis for BFNES scores in female participants (Significant time x group Interaction)

Comparison	Mean Difference (I-J)	Std. Error	Sigb	95% Confidence Interval for Differenceb (Lower, Upper)	Effect Size (Partial η²)	F-statistic (df = 1, 50)
Within-Group Effects (Time 2 vs. Time 1)
Sports Games Group (Experimental)	−11.231*	2.046	0.000	(−15.341, −7.121)	0.376	30.124a
Aerobic Exercise Group (Control)	−3.462	2.046	0.097	(−7.571, 0.648)	0.054	2.862a
Between-Group Effects (Experimental vs. Control)
At Pre-test (Time 1)	−0.269	2.237	0.905	(−4.762, 4.224)	0.000	0.014
At Post-test (Time 2)	−8.038*	2.521	0.002	(−13.102, −2.975)	0.169	10.167

Note：*The mean difference is significant at the .05 level

^aExact statistic

^bAdjustment for multiple comparisons: Bonferroni

Male participants

In contrast to the findings for females, the analyses for male participants did not reveal a significant time × group interaction effect for either IAS (F (1, 50) = 2.964, p =.091) or BFNES (F (1, 50) = 2.835, p =.098). This lack of a significant interaction indicates that the TGSG intervention was not demonstrably superior to aerobic exercise for males.

However, a significant main effect of time was observed for both IAS (F (1, 50) = 21.897, p <.001, partial η² = 0.300) and BFNES (F (1, 50) = 10.605, p =.002, partial η² = 0.172). This suggests that both interventions were effective in reducing social anxiety symptoms over the 10-week period for male students, with no statistically discernible difference in their degree of efficacy.

Exercise intensity control

To verify that the therapeutic effects observed were not confounded by differences in physiological exertion, heart rate was monitored as an objective measure of exercise intensity. According to the American College of Sports Medicine (ACSM) guidelines, the heart rate range for low-intensity aerobic exercise is 50–63% of maximum heart rat. Throughout the 10-week intervention period, 10 randomly selected participants from each group wore heart rate monitoring devices during each session. The analysis revealed no statistically significant difference in mean heart rate between the control group (AE: M = 119.83 bpm, SD = 5.35) and the experimental group (TGSG: M = 118.28 bpm, SD = 6.15), t(198) = −1.91, p =.057. Although the result approached the threshold of significance, it did not meet the conventional criterion. Furthermore, the mean heart rates for both groups were well within the pre-specified target range for low-intensity aerobic exercise.These results provide strong evidence that the two groups engaged in physical activity at a comparable physiological intensity.

Discussion

Overall efficacy and gender specific patterns

This randomized controlled trial evaluated the efficacy of an innovative intervention, Therapeutic Group Sports Games (TGSG), in alleviating social anxiety among university students, compared to an active control of standard aerobic exercise (AE). The TGSG intervention demonstrated superior efficacy in reducing social anxiety symptoms, with participants experiencing significantly greater reductions in Interaction Anxiousness Scale (IAS) and Brief Fear of Negative Evaluation Scale (BFNES) scores from pre- to post-intervention compared to those in the AE control group. This was supported by significant group × time interactions for both measures, confirming our initial hypothesis regarding TGSG’s overall effectiveness in mitigating social anxiety. Beyond this general efficacy, our results further uncovered a critical gender-specific pattern. TGSG demonstrated significant efficacy in reducing social anxiety specifically among female university students, who showed significantly greater reductions in both interaction anxiety and fear of negative evaluation compared to the AE control group. In contrast, male participants demonstrated comparable improvements in social anxiety symptoms across both interventions, as evidenced by a significant main effect of time but no significant group × time interaction.

A critical consideration for these gender-specific patterns is the baseline symptom severity. Our analyses revealed that female participants reported significantly higher IAS scores compared to male participants at baseline. While no significant baseline difference was observed in BFNES scores between genders, the higher initial IAS scores in females suggest they may have had more room for improvement. Conversely, the relatively lower baseline IAS scores among male participants could potentially contribute to a “floor effect,” where their symptom levels were already closer to the minimum measurable range, thus limiting the observable magnitude of further reduction and making it more challenging to detect differential intervention effects beyond the general benefits of physical activity. These gender-specific findings partially support our initial hypotheses. Specifically, TGSG’s design appears particularly adept at addressing the core challenge of avoidance in socially anxious women, a factor often exacerbated by their heightened fear of negative evaluation and lower perceived social support. This was achieved through several synergistic mechanisms, deeply rooted in Yalom’s interpersonal group psychotherapy model, that extend beyond the effects of physical activity alone.

TGSG mechanisms for female participants

One key mechanism involved the structured yet low-threat social exposure embedded within engaging sports games, which allowed female participants to gradually confront feared social situations. Unlike traditional exposure therapy, the playful, goal-oriented nature of TGSG, combined with the diversion of attention toward game objectives rather than self-focused anxiety, effectively reduced the perceived social threat. This facilitated genuine social interaction and skill practice in a naturalistic, less clinically-framed context. For women, who frequently report greater subjective distress and exhibit heightened physiological and neural reactivity to social threats [9, 44], this gradual and indirect exposure might have been particularly critical in overcoming initial avoidance, as direct confrontation could be overwhelmingly anxiety-provoking. In contrast, men, who may be more prone to externalizing behaviors or substance use as coping mechanisms and often engage in more problem-solving focused coping [9, 45], might find the “low-threat” aspect less uniquely impactful compared to the direct utility of the physical activity itself.

Another crucial aspect was the supportive, cohesive peer setting that fostered psychological safety, a critical element for individuals with social anxiety. This aspect holds particular salience for women, given their often higher rates of internalizing comorbidities and a greater reported link between social anxiety and lower perceived social support [13]. This environment, promoting Yalom’s factors of group cohesiveness and universality, allowed for corrective social experiences where positive feedback and acceptance directly challenged maladaptive cognitions, leading to a significant reduction in fear of negative evaluation. The emphasis on emotional expression and interpersonal connection within the group might resonate more strongly with female participants, who typically exhibit higher emotionality and are often more receptive to relational therapies and group-based emotional processing [46]. While men also benefit from group support, their communication patterns might make them less immediately engaged with or less profoundly impacted by the purely emotional and relational aspects of group cohesion within this specific context.

Furthermore, TGSG effectively facilitated interpersonal learning and the development of socializing techniques. Within the guided group dynamics, participants could observe, imitate, and practice adaptive communication patterns and social behaviors. For female participants, who may experience social anxiety with higher emotionality and potentially more nuanced social anxieties related to interpersonal dynamics [47], the structured practice of adaptive communication and the opportunity for interpersonal feedback in a safe space could be uniquely valuable. This contrasts with men, whose primary social anxiety fears such as dating anxiety or reliance on substance use as coping [10] might require different or more direct forms of intervention not fully captured by this specific group sports format, and who may also be less inclined to seek psychological treatment.

These psychosocial elements structured exposure, psychological safety, and focused skill development are the crucial differentiators that elevate TGSG beyond mere physical exercise, and their differential impact on female participants appears to be a key driver of our gender-specific findings. While aerobic activity offers anxiolytic benefits, it lacks the targeted, relational, and cognitive restructuring components inherent in TGSG’s integration of Yalom’s model and CBT principles. For female participants, this comprehensive, experiential learning approach produced substantially larger effects than aerobic exercise alone, suggesting a particular resonance with their social anxiety presentation and effectively overcoming the pervasive pattern of avoidance.

Methodological control and psychosocial attribution

A critical methodological strength of our study was the careful control of exercise intensity across both interventions. Heart rate monitoring confirmed that no statistically significant difference in physiological exertion between the TGSG group and the AE control group (t(198) = −1.91, p =.057). This finding is pivotal. It allows us to confidently attribute the superior outcomes in the female TGSG group to the specific psychosocial components of the intervention, rather than to differences in physical activity levels. The TGSG condition’s distinctive feature was its systematic integration of physical activity with targeted psychosocial elements. This multidimensional approach simultaneously engaged cognitive processes (through strategic decision-making and reframing of social situations), behavioral components (via structured social interaction and skills practice), emotional regulation (through supportive group experiences), and physiological mechanisms (via moderate physical activity). For female participants specifically, this integrated approach produced effect sizes (Partial η²=0.612 for IAS; Partial η²=0.376 for BFNES) substantially exceeding those typically reported in randomized clinical trial of either standalone psychological interventions or exercise-only programs for anxiety [48, 49]. The pronounced efficacy differential, despite comparable physical exertion, provides compelling evidence that the psychosocial elements derived from Yalom’s model—rather than the physical activity alone—were the primary therapeutic agents driving the superior outcomes.

Male participant response and underlying pathways

The differential response pattern observed in male participants, where both TGSG and AE led to comparable improvements in social anxiety symptoms, offers important insights into intervention mechanisms. Our analysis demonstrated that both interventions effectively reduced social anxiety in male participants, suggesting that distinct yet effective therapeutic pathways were likely at play.

A primary contributor to the observed improvements in both groups is the inherent therapeutic power of physical activity itself. Exercise directly modulates neurobiological mechanisms underlying anxiety, for instance, by increasing BDNF production, enhancing endocannabinoid signaling, and improving HPA axis regulation [50–52]. These physiological changes effectively counteract the autonomic hyperarousal characteristic of social anxiety, likely accounting for a substantial portion of the symptom reduction seen in both conditions.

Beyond the physiological benefits, the social context of the AE group also warrants consideration. While the AE protocol involved individual exercise, it occurred within a shared space, constituting a form of passive, low-dose social exposure. This minimal but consistent exposure to the presence of others, without the pressure of direct interaction, may have facilitated habituation to being observed in a non-threatening environment. In contrast, the TGSG intervention provided active, structured, and goal-directed interpersonal interaction. Although TGSG produced numerically larger improvements in males (IAS reduction: 7.73 vs. 3.58 points), this difference did not reach statistical significance (p =.091).

This pattern suggests that for male participants, the potent combination of exercise physiology and even minimal, passive social exposure was sufficient to produce significant gains. It appears that the additional, more complex psychosocial elements integrated into TGSG, while beneficial for females, may not have conferred a statistically detectable additional advantage for males within the scope of this study. This observation aligns with findings from broader psychotherapy research, which sometimes indicates that core “active ingredients” of an intervention can drive a significant portion of the therapeutic effect, making the incremental benefits of more elaborate components challenging to detect statistically in certain populations or sample sizes [53, 54]. For male participants, whose social anxiety presentation may differ, the fundamental anxiolytic effects of exercise combined with a low-pressure social setting might have already met a substantial portion of their therapeutic needs. Nonetheless, the moderate effect size for the interaction term (Partial η² = 0.050) warrants attention, suggesting a potential additive benefit of the TGSG intervention that might become significant with larger samples or a longer intervention period.

Practical and clinical implications

These results offer a compelling rationale for universities and community mental health services to develop, implement, and evaluate psychologically-informed sports and activity groups as a core offering. Such programs represent a potentially less stigmatizing, and cost-effective means of promoting social-emotional well-being and addressing social anxiety, especially for female students who demonstrated a clear and superior benefit in this trial. This approach could significantly broaden the spectrum of available mental health support, appealing to individuals who might not otherwise seek or adhere to traditional forms of therapy, and could serve as a valuable early intervention or preventative strategy within high-density young adult populations.Finally, the relatively simple structure of the intervention—sports games infused with group psychological counseling principles—suggests it could be scalable and cost-effective in resource-limited settings. Physical education departments and student services could potentially collaborate to implement similar programs with appropriate training and supervision, creating sustainable mental health support options that simultaneously promote physical well-being.

Limitations and future directions

Several limitations warrant careful consideration when interpreting our findings, each pointing to crucial avenues for future research. First, our sample consisted exclusively of undergraduate students from a single university in China. This limits the generalizability of our results, as social anxiety expression and its correlates can vary across different cultural contexts. Future research should therefore explicitly aim to replicate these findings in diverse East Asian student populations and other cultural settings, acknowledging potential cultural variations in social anxiety presentation and responses to intervention. Second, participants were selected based on elevated self-report symptoms rather than a formal clinical diagnosis of SAD. While this subclinical population is a critical target for preventative interventions, this study should be viewed as a crucial preliminary investigation. Future research is needed to validate the efficacy of TGSG in clinically diagnosed populations, to confirm its therapeutic benefits across the full spectrum of social anxiety. Third, while our 10-week intervention demonstrated significant short-term effects, we lack follow-up data to determine the longevity and sustainability of these benefits. Future longitudinal studies are essential to evaluate whether TGSG produces lasting reductions in social anxiety symptoms and maintains its gender-specific advantages over extended periods. Fourth, our measures relied on self-reported questionnaires, which may be subject to social desirability bias and recall inaccuracies. Future studies could incorporate objective behavioral assessments to provide a more comprehensive evaluation of symptom reduction and functional improvement. Fifth, our assessment focused primarily on symptom reduction rather than direct functional improvement. Future research should integrate these functional outcomes to provide a more holistic understanding of TGSG’s real-world impact.

Conclusion

This randomized controlled trial provides compelling evidence that Therapeutic Group Sports Games (TGSG), an intervention integrating Yalom’s group psychotherapy principles with physical activity, is a highly effective, gender-sensitive treatment for social anxiety in university students. The study’s key finding is the demonstrably superior efficacy of TGSG for female participants in reducing both interaction anxiety and fear of negative evaluation, compared to an exercise intensity-matched aerobic exercise control. This crucial attribution stems from the integrated psychosocial components of the TGSG model, rather than from physical activity alone. While both interventions yielded improvements for male participants, TGSG did not demonstrate statistically significant superiority over aerobic exercise in this cohort. These findings underscore the critical importance of considering gender as a moderating factor in the design and application of social anxiety interventions, challenging the notion of a uniform approach to mental health interventions and highlighting the therapeutic potential of TGSG as a scalable, low-stigma, and early intervention option for specific populations.

Authors’contributions

J.N.F: Writing – original draft, Investigation, Methodology, Formal analysis, Conceptualization. : Y.C.G Writing – review & editing, Investigation, Supervision, Data curation. X.Z : Writing – review & editing, Supervision, Data curation. W.B.Y: Writing – review & editing, Supervision, Methodology, Data curation.All authors read and approved the final manuscript and are accountable for the content of the article.

Funding

This research received no specific grant from any funding agency, commercial or not-for-profit sectors.

Data availability

The data that support the findings of this study are available from the corresponding author, [WENBING YU], upon reasonable request.

Declarations

Ethics approval and consent to participate

The study protocol was approved by the Ethics Committee of Ocean University of China (approval number: OUC-HM-032). All procedures were performed in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki Declaration and its later amendments. Written informed consent was obtained from all individual participants included in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.