Abstract

Background:

Autistic traits are prevalent among college students and are associated with impaired social functioning and mental health. Although physical activity is considered a protective factor, the underlying mechanisms, particularly the mediating role of social anxiety and the heterogeneity of autistic traits, remain unclear. This study aimed to examine the association between physical activity and autistic traits, test the mediating role of social anxiety, and identify latent subgroups among female college students.

Methods:

A cross-sectional survey was conducted among 2,137 female college students (M_age = 19.31, SD = 1.17). Autistic traits, social anxiety, and physical activity were assessed using the Autism-Spectrum Quotient-10 (AQ-10), the Social Interaction Anxiety Scale-6 (SIAS-6) and the Social Phobia Scale-6 (SPS-6), and the Physical Activity Rating Scale-3 (PARS-3), respectively. Mediation analysis was performed using PROCESS (Model 4), and latent profile analysis (LPA) was conducted to identify subgroups.

Results:

Physical activity was significantly negatively associated with social anxiety (β = -0.205, p < 0.001) and autistic traits (β = -0.197, p < 0.001), while social anxiety was positively associated with autistic traits (β = 0.425, p < 0.001). Social anxiety partially mediated this relationship (indirect effect = -0.087, 95% CI [-0.106, -0.067]), accounting for 31% of the total effect. Latent profile analysis identified two profiles: low autistic traits (74.54%) and high autistic traits (25.46%). The high trait group showed lower physical activity (t = 7.91, p < 0.001, d = 0.39) and higher social anxiety (t = -23.25, p < 0.001, d = -1.16).

Conclusions:

Physical activity and social anxiety are key modifiable factors associated with autistic traits. Interventions should promote physical activity while reducing social anxiety, particularly for students with elevated autistic traits.

1 Introduction

Autistic traits are defined as autism-spectrum-like characteristics manifested in social interaction, communication styles, and behavioral interests, originating primarily from heightened sensitivity to uncertainty and evaluative cues in social contexts (1, 2). These traits are dimensional and widely distributed in the general population, including difficulties in social communication (e.g., reduced eye contact), atypical language use, cognitive inflexibility, and restricted interests (3, 4). It is important to distinguish autistic traits from autism spectrum disorder (ASD), a clinically diagnosed neurodevelopmental condition defined by the DSM-5 and characterized by significant functional impairment (5). Individuals with elevated autistic traits may not meet ASD diagnostic criteria but can still experience notable psychosocial challenges (6, 7).

Autistic traits are prevalent among youth and are particularly common in college students, with approximately 15%–30% exhibiting moderate-to-high levels and about 1%–2% reaching clinically relevant thresholds (8–11). These traits have been linked to increased academic pressure and developmental challenges during the college years (12, 13) and are associated with social anxiety, loneliness, and maladaptive behaviors (14, 15). Female college students warrant particular attention in this field. Compared with male college students, females with elevated autistic traits are more likely to exhibit subtler or camouflaged presentations, which may obscure identification while leaving associated internalizing problems, particularly social anxiety, insufficiently recognized (16). At the same time, the college period involves heightened interpersonal and identity-related demands, which may render female students with elevated autistic traits especially vulnerable to psychosocial difficulties (17).

Accordingly, it is important to examine modifiable factors associated with autistic traits in this population. Physical activity has been widely linked to improved psychological well-being and social functioning (18, 19). In contrast, social anxiety, defined as the fear of negative evaluation in social situations (20), may serve as a key explanatory mechanism. Therefore, the present study aimed to examine the associations among physical activity, social anxiety, and autistic traits in female college students, and to test whether social anxiety mediates the relationship between physical activity and autistic traits.

1.1 Physical activity and autistic traits

In recent years, physical activity has been widely recognized as a cost-effective and accessible approach for promoting psychological and behavioral health (21, 22). According to the World Health Organization, physical activity refers to any bodily movement produced by skeletal muscles that results in energy expenditure. In contrast, physical exercise is a structured and planned subset of physical activity (23, 24). The present study focuses on broader engagement in physical activity rather than solely structured exercise.

Existing research indicates that physical activity contributes to enhanced social functioning and psychological adaptation (25–28). Drawing on the Broaden-and-Build Theory, physical activity can elicit positive emotional experiences such as enjoyment and relaxation, thereby expanding attentional scope and enhancing cognitive flexibility, which may help alleviate cognitive rigidity and stereotyped patterns associated with autistic traits (29, 30). Meanwhile, the Distraction Hypothesis suggests that physical activity facilitates a shift of attention from internal rumination to external movement processes, thereby interrupting repetitive thought patterns (31, 32). In addition, Social Cognitive Theory (SCT) posits that emotional states play a central role in behavioral regulation (33). When individuals perceive social threat or anxiety, they are more likely to adopt avoidance behaviors. Physical activity may reduce such avoidance tendencies by improving emotional states and enhancing self-regulation capacity, thereby promoting more adaptive social engagement (34, 35).

At the contextual level, physical activity also provides a natural platform for social interaction, which may contribute to improved social adaptation (36). Empirical evidence suggests that participation in group-based physical activities can enhance social communication and adaptability among individuals with autistic traits, and regular physical activity is associated with better mental health and lower levels of internalizing problems (37). Longitudinal studies further indicate that physical activity predicts better psychological functioning and reduces behaviors such as social withdrawal, which substantially overlap with autistic traits (38–40).

1.2 The mediating role of social anxiety

Social anxiety may play a pivotal mediating role between physical activity and autistic traits. Social anxiety is typically defined as a persistent fear of negative evaluation in social or performance situations, accompanied by heightened distress and avoidance tendencies (16). Within the framework of Social Cognitive Theory (SCT), emotional states play a central role in behavioral regulation. When individuals experience elevated levels of social anxiety, they are more likely to adopt avoidance strategies, such as social withdrawal or reduced interaction, which are also characteristic features of autistic traits (41).

Physical activity may alleviate social anxiety through multiple pathways. Specifically, engaging in physical activity can enhance individuals’ sense of competence and provide positive social feedback, thereby reducing perceived social threat and promoting more adaptive emotional regulation (42–44). As social anxiety decreases, individuals experience less fear in social situations and improved self-regulation capacity, which manifests behaviorally as reduced social withdrawal and fewer stereotyped or avoidant responses, ultimately reflecting lower levels of autistic traits (45, 46).

A substantial body of empirical evidence supports these theoretical assumptions. Studies focusing on college student populations have shown that regular physical activity is associated with lower levels of social anxiety and better emotion regulation (47, 48). As a key psychological construct, social anxiety reflects individuals’ tendency toward tension and avoidance in socially evaluative contexts. It has been identified as a significant predictor of social withdrawal and maladaptive behaviors (49). Higher levels of social anxiety are associated with increased avoidance behaviors and interpersonal difficulties, which may exacerbate behavioral manifestations related to autistic traits (50–52). In contrast, lower levels of social anxiety are typically linked to better social adaptability and fewer autism-like behaviors (53). Although existing literature has addressed these variables individually, the specific intrinsic integrative mechanisms linking physical activity, social anxiety, and autistic traits warrant further systematic exploration.

1.3 Latent profile analysis

Existing scholarship has predominantly adopted a variable-centered perspective. However, such methodologies, which rely on sample averages, often fail to adequately capture inter-individual heterogeneity (54). By analyzing distinct subgroups of individuals, researchers can gain a deeper understanding of variations in physical activity and social anxiety, as well as how these differences influence autistic traits (55). Consequently, analyzing the latent categories of autistic traits from a person-centered perspective is of significant importance. Latent Profile Analysis (LPA) constitutes a person-centered statistical approach that clarifies the links among observed variables through latent continuous variables and further elucidates relationships among observed variables using latent class variables (56, 57).

In recent years, a limited number of studies have employed this methodology to explore latent categories of autistic traits. For instance, certain studies have identified two distinct types of autistic traits (58), while others suggest that a three-class model captures the nuances more effectively (59). Nevertheless, research on the latent categories of autistic traits remains relatively scarce, and classification results across studies have yet to reach consensus. This implies that a definitive consensus regarding the underlying structure and characteristics of autistic traits remains elusive. Based on current research, this study aims to further refine the latent patterns of autistic traits among female college students. Identifying distinct types of autistic traits facilitates a more profound grasp of the relationships between physical activity, social anxiety, and autistic traits of varying levels and characteristics. This approach allows for transcending the limitations of average effects to reveal internal sample heterogeneity, thereby providing a basis for formulating more targeted intervention strategies.

1.4 Current research

Synthesizing the aforementioned literature, the present research aims to investigate the potential inverse relationship between physical activity and autistic traits, while also exploring the mediating role of social anxiety in this association. Moreover, this research employs LPA to identify distinct latent profiles of autistic traits among female college students and to explore variations in physical activity and social anxiety among the profiles. This comprehensive analytical approach, combining both variable-centered and person-centered perspectives, not only enriches existing literature by elucidating the complex mechanisms underlying autistic traits but also offers novel translational perspectives for tailoring mental health interventions for female college students.

Building on relevant theoretical models and prior research, we formulate three key hypotheses: (H1) Physical activity is inversely and markedly associated with autistic traits. (H2) Social anxiety plays a significant mediating role in the relationship between physical activity and autistic traits. (H3) Autistic traits among female college students exhibit distinct latent profiles, and these profiles exhibit significant disparities in physical activity and social anxiety levels.

2 Materials and methods2.1 Participants

The present study employed a cross-sectional survey design. Data were collected between September 10 and December 22, 2025, using offline paper-and-pencil questionnaires administered in classrooms and libraries across universities in China. To ensure adequate statistical power for the planned analyses, an a priori power analysis was conducted using G*Power 3.1 for the intended regression-based mediation model (60). The results indicated that, given a medium effect size (f² = 0.15), a significance level of α = 0.05, and a desired power of (1−β) = 0.95, a sample size of at least 124 participants is necessary to detect the target effect. The actual effective sample size obtained in this study (N = 2,137) substantially exceeded this threshold, thereby providing a robust statistical basis for subsequent model testing and interpretation.

A convenience sampling strategy was adopted to recruit participants from universities in China. Eligibility criteria were as follows: (a) biologically female; (b) currently enrolled as full-time undergraduate students; (c) no physical disabilities that prevent participation in physical activity; and (d) voluntary participation with informed consent. To ensure data quality, all returned questionnaires underwent manual verification, and the following exclusion criteria were applied: (a) self-reported history of clinically diagnosed psychiatric or neurological disorders (e.g., depression, schizophrenia, or clinically diagnosed autism spectrum disorder); (b) questionnaires with significant missing data; and (c) invalid questionnaires characterized by regular response patterns (e.g., straight-lining) or obvious inattentive responding.

All participants provided written informed consent before participation. They were fully informed of the study’s purpose and procedures, the voluntary nature of participation, and their right to withdraw at any time without penalty. The anonymity and confidentiality of all responses were strictly ensured, and no personally identifiable information was collected.

The study protocol was reviewed and approved by the Institutional Review Board of Jiangxi Normal University (IRB-JXNU-PEC-20240106). All procedures were conducted in accordance with the ethical standards of the Declaration of Helsinki.

2.2 Measurements2.2.1 Autistic traits scale

Autistic traits were measured using the 10-item Autism Spectrum Quotient (AQ-10) developed by Baron-Cohen (61). Comprising 10 items, this scale is widely employed for the rapid screening and measurement of adult autistic traits due to its brevity and ease of administration. Items are rated on a 4-point Likert scale (0 = definitely disagree, 3 = definitely agree), and reverse-scored items were recoded before statistical analysis. Total scores range from 0 to 30, with higher scores indicating more pronounced autistic traits. The widely used Chinese version of the AQ-10 was adopted in this study, which has demonstrated good applicability in previous studies involving Chinese university students (62). In the current study, the Cronbach’s α coefficient for this scale was 0.950.

2.2.2 Social anxiety anxiety scale

Social anxiety levels were assessed using the short forms of the Social Interaction Anxiety Scale (SIAS-6) and the Social Phobia Scale (SPS-6), as revised by Peters (63). The instrument consists of 12 items across two subscales: social interaction anxiety (SIAS-6) and social phobia (SPS-6), each with 6 items. All items were rated on a 5-point Likert scale ranging from 0 (“not at all characteristic of me”) to 4 (“extremely characteristic of me”). Scores for each subscale range from 0 to 24, with higher scores indicating greater levels of social interaction anxiety or social phobia, respectively. The total social anxiety score is obtained by summing all items, with a possible range from 0 to 48, where higher scores reflect higher overall levels of social anxiety. The Chinese version of the SIAS-6/SPS-6 has been validated in college student populations and demonstrates good reliability and validity (64). In the current study, the Cronbach’s α coefficient for the scale was 0.926.

2.2.3 Physical activity

Participants’ physical activity levels were evaluated using the Physical Activity Rating Scale-3 (PARS-3) developed by Liang (65). This scale consists of three distinct items designed to measure the Intensity, duration, and Frequency of exercise. The total score for physical activity was calculated using the formula: Physical Activity Score = Intensity × (Duration - 1) × Frequency. Scores range from 0 to 100, with higher scores indicating higher levels of physical activity. Tailored specifically to Chinese university students, this instrument has demonstrated robust reliability and validity across extensive empirical research (66). In the current study, the Cronbach’s α coefficient for this scale was 0.748.

2.2.4 Control variables

Given that autistic traits and social anxiety in female university students may be contingent upon specific background factors, particularly those related to demographic characteristics (67), variables such as grade, place of residence, and family structure were identified. Consequently, grade, place of residence, and only-child status were included as control variables in the analysis. These were dummy coded as follows: place of residence (1 = urban, 2 = rural), only-child status (1 = non-only child, 2 = only child), and grade (1 = freshman, 2 = sophomore, 3 = junior, 4 = senior).

2.3 Data analysis

Data analyses were conducted using SPSS 26.0 and Mplus 8.3 statistical software (68, 69). Preliminary analyses included descriptive statistics and Pearson correlation coefficients for all variables. The normality of the data was assessed using skewness and kurtosis indices. When the absolute values of skewness were less than 2, and those of kurtosis were less than 7, the data were considered not to deviate substantially from normality (70). Given that the skewness of social anxiety slightly exceeded the recommended threshold, both Pearson correlation and Spearman rank-order correlation analyses were performed to ensure the robustness of the findings. To assess potential common-method bias in self-reported data, Harman’s single-factor test was conducted.

For the variable-centered analysis, the PROCESS macro (Model 4) in SPSS was utilized to examine the mediating mechanism of social anxiety in the relationship between physical activity and autistic traits. Grade, place of residence, and only-child status were included as covariates in the analysis. The bias-corrected nonparametric percentile Bootstrap method (with 5,000 resamples) was used to estimate the 95% confidence interval (CI) for the mediation effect; a CI that does not contain zero indicates statistical significance.

For the person-centered analysis, latent profile analysis (LPA) was conducted using Mplus 8.3. Model fit was evaluated using the Akaike Information Criterion (AIC), the Bayesian Information Criterion (BIC), and the adjusted BIC (aBIC), with lower values indicating better fit. Entropy values were used to assess classification accuracy, with higher values indicating better classification. The optimal number of profiles was determined based on the Lo–Mendell–Rubin likelihood ratio test (LMR-LRT) and the bootstrap likelihood ratio test (BLRT), with statistical significance set at p < 0.05. A minimum class size of 5% was required to ensure interpretability (71). For group comparisons between latent profiles, independent-samples t-tests were conducted. To account for potential deviations from normality, Mann–Whitney U tests were also performed as a robustness check. Effect sizes were reported using Cohen’s d for t-tests and rank-biserial correlation for Mann–Whitney U tests where appropriate. Across all statistical analyses, a two-tailed p-value of < 0.05 was adopted as the criterion for statistical significance, unless otherwise specified.

3 Results3.1 General information about respondents

As shown in Table 1, the mean age of the participants was 19.31 years (SD = 1.17). In terms of academic year, the sample was predominantly composed of sophomores (38.20%) and freshmen (34.10%), followed by juniors (22.20%) and seniors (5.50%). Most participants were only children (72.60%), and a greater proportion of the sample came from rural areas (59.00%) compared to urban areas (41.00%).

VariableCategoryn (%)M ± SDAge (years)——19.31 ± 1.17Academic yearFreshman728 (34.10)—Sophomore816 (38.20)—Junior475 (22.20)—Senior118 (5.50)—Only-child statusOnly child1,551 (72.60)—Non-only child586 (27.40)—Place of originUrban876 (41.00)—Rural1,261 (59.00)—

Demographic characteristics of the participants (N = 2,137).

3.2 Common method bias test

To assess common method bias, Harman’s single-factor test was conducted. As shown in Table 2, the results indicated that four factors with eigenvalues greater than 1 were extracted. The first unrotated factor accounted for 41.23% of the total variance, which is below the commonly accepted threshold of 50% and close to the more conservative 40% criterion (72). These findings suggest that common method bias is unlikely to pose a serious threat to the validity of the results. In addition, procedural remedies, such as ensuring anonymity and voluntary participation, were implemented to further mitigate potential common method bias.

ComponentEigenvalue% of VarianceCumulative %110.30741.22741.22723.48113.92355.15031.7446.97862.12741.0714.28566.413

Total variance explained from Harman’s single-factor test.

Only factors with eigenvalues greater than 1 are presented. Extraction method: Principal Component Analysis.

3.3 Correlational analysis

Table 3 summarizes the descriptive statistics and correlation coefficients for physical activity, social anxiety, and autistic traits. Most variables exhibited acceptable levels of skewness (|skewness| < 2) and kurtosis (|kurtosis| < 7) (70). However, the skewness of social anxiety slightly exceeded the recommended threshold. Therefore, Spearman’s rank-order correlation was conducted as the primary analysis, while Pearson correlation analysis was additionally performed as a robustness check. The results indicated that physical activity (M = 42.30, SD = 35.93) was significantly negatively correlated with social anxiety (r = -0.378, p < 0.01) and autistic traits (r = -0.401, p < 0.01). Furthermore, social anxiety (M = 11.44, SD = 8.57) was significantly positively correlated with autistic traits (r = 0.566, p < 0.01). In addition, Pearson correlation analysis was conducted as a robustness check, and the results are presented in Table 4. The Pearson correlation results showed the same directional patterns and statistical significance as the Spearman correlations, further supporting the stability and reliability of the findings.

VariableMSDSkewnessKurtosis1231. Physical activity42.3035.930.641-0.75512. Social anxiety11.448.572.0525.179-0.378**13. Autistic traits7.967.7101.0720.479-0.401**0.566**1

Descriptive statistics and Spearman rank-order correlations among variables (N = 2137).

Variable1231. Physical activity12. Social anxiety-0.208**13. Autistic traits-0.293**0.489**1

Pearson correlations among variables (N = 2,137).

3.4 Variable-Centered centered analysis

Hayes’ PROCESS macro (Model 4) was employed to test whether social anxiety mediated the association between physical activity and autistic traits. Origin, only-child status, and grade were included as control variables. The findings indicated a significant negative association between physical activity and social anxiety (β = -0.205, p < 0.001). When both physical activity and social anxiety were entered into the regression model predicting autistic traits, social anxiety demonstrated a significant positive association with autistic traits (β = 0.425, p < 0.001). Importantly, the direct effect of physical activity on autistic traits remained statistically significant and negative (β = -0.197, p < 0.001). This pattern suggests that social anxiety serves as a partial mediator of the association between physical activity and autistic traits (see Table 5, Figure 1).

VariableSocial anxietyAutistic traitsβtβtPlace of Residence0.1483.498***0.2586.952***Only Child Status-0.417-8.972***-0.134-3.229**Grade0.0040.173-0.004-0.174Physical activity-0.205-9.813***-0.197-10.553***Social anxiety0.42522.395***R20.0840.296F48.715***179.516***

Testing the mediation model.

The dependent variable is autistic traits; the independent variable is physical activity; the mediating variable is social anxiety; and the control variables are place of residence, only child status, and grade. **p < 0.01, ***p < 0.001.

Path diagram of the mediation model. ***p < 0.001.

The mediation effect was further evaluated using a Bootstrap procedure with 5,000 resamples. As reported in Table 6, the analysis revealed a significant indirect effect of social anxiety on the relationship between physical activity and autistic traits, with an effect estimate of -0.087 (95% CI: -0.106 to -0.067). Since the 95% confidence interval did not include “0,” the mediating effect is statistically significant. The mediated (indirect) pathway explained 31% of the overall effect. These results offer empirical evidence that social anxiety constitutes a key mechanism linking physical activity to autistic traits.

PathEffectBootSEBootLLCIBootULCIRelative proportionTotal effect-0.2840.020-0.324-0.244100%Direct effect-0.1970.019-0.234-0.16069%Indirect effect-0.0870.010-0.106-0.06731%

Results of mediation effect analysis.

3.5 Latent profile analysis

Employing a person-centered perspective, the study further explored the complex relationships between these variables. LPA was conducted in Mplus 8.3 using the 10-item autistic traits scale to identify homogeneous subgroups. Models ranging from one to five profiles were estimated (see Table 7).

ProfilesAICBICaBICLMR-LRTBLRTEntropy12345157544.72457658.06757594.524N/AN/A11246117.75946293.44046194.950<0.001<0.0010.95674.5425.46341122.77341360.79441227.355<0.001<0.0010.94630.0058.9111.09439959.11440259.47440091.086<0.001<0.0010.89232.1011.0421.9934.87538792.59939155.29838951.962<0.001<0.0010.9189.8332.9412.5433.65