Physical activity is a low-cost, effective health intervention that has drawn growing attention from researchers and practitioners (Graham et al., 2025; Bermejo-Cantarero et al., 2024; Hu et al., 2025). It involves planned, structured exercise aimed at improving physical fitness and cardiorespiratory function, developing muscular strength and endurance, and promoting mental health and social functioning (Hulteen et al., 2017; Yuaqing et al., 2025). Common forms include aerobic exercise, strength training, team sports, and flexibility work (Kern et al., 2020; Cho et al., 2023). Such activities help release excess energy, reduce stress, enhance self-control, and strengthen social connections through teamwork and interaction (Eather et al., 2023; Wu et al., 2025; Chiou et al., 2020).

Four main theoretical pathways are used to explain how physical activity relates to aggression: emotion regulation (Gross, 2002), cognitive control (Diamond, 2013), social bonding (Loudin et al., 2003), and social identity (Harwood, n.d.). Emotion regulation accounts posit that exercise reduces anxiety and anger, thereby curbing impulsive aggression (Ouyang and Liu, 2023; Hamer et al., 2012). Cognitive control accounts emphasize improvements in executive function that bolster impulse control (Zhu et al., 2023; Cya et al., 2019). Social bonding and social identity perspectives highlight how behavioral synchrony and group belonging reduce within- and between-group conflict, suppressing aggressive behavior (Allen et al., 2018; Karriker-Jaffe et al., 2013; Hirsh and Kang, 2016). Empirical cross-sectional studies have also confirmed that there is a significant negative correlation between physical activity and aggressive behavior in adolescents (Xu et al., 2025). In other words, adolescents who regularly engage in physical activity tend to exhibit lower rates of aggressive behavior (Huang et al., 2021).

However, most existing research relies on cross-sectional designs. While useful for identifying contemporaneous associations, these designs cannot capture within-person change over time or establish temporal ordering (Wang and Cheng, 2020; Vanderweele et al., 2020). This limitation restricts our in-depth understanding of the relationship between physical activity and aggressive behavior. Moreover, there is still a lack of in-depth empirical research on the underlying mechanisms linking physical activity to aggressive behavior. Uncovering these mechanisms is of great theoretical significance for understanding aggressive behavior and can provide a scientific basis for future prevention and intervention measures aimed at addressing it in adolescents.

Self-control may be a key mediator in the longitudinal link between physical activity and aggressive behavior. Physical activity serves as a structured intervention that inherently involves goal-setting, discipline, and emotional regulation, all of which contribute to strengthening self-control (Biddle, 2016; Vella et al., 2023; Gratz et al., 2018). According to the strength model of self-control (Baumeister et al., 2007), self-control functions as a limited psychological resource that can be depleted through use but also replenished and strengthened through consistent practice. Physical activity, particularly in forms such as aerobic exercise and team sports, provides repeated opportunities to practice self-regulation, thereby increasing self-control capacity over time (Cheng et al., 2025; Zhong et al., 2021). Moreover, engaging in physical activity has been shown to lower baseline arousal and foster adaptive coping mechanisms, both of which are critical components of self-control (Kruschwitz et al., 2024; Potoczny et al., 2025; Tse, 2020).

Self-control theory (Gul and Pesendorfer, 2004) posits that self-control is a psychological capacity enabling individuals to inhibit impulsive behaviors and choose actions aligned with societal norms. As both a finite resource that can be temporarily depleted and a skill that can be enhanced through training, self-control can be strengthened via goal setting, repeated practice, and feedback (Cen et al., 2022; Jiang and Dong, 2022; Yang et al., 2022). Adolescence, a critical period for the development of neurocognitive systems like the prefrontal cortex, provides a window for structured activities such as physical exercise to foster self-control (Denson et al., 2012; Plessen et al., 2023). Higher self-control has been consistently linked to better emotional regulation and behavioral inhibition, reducing the likelihood of impulsive and aggressive responses. Meta-analyses confirm a robust negative association between self-control and aggression, highlighting self-control as a protective factor that helps adolescents adopt adaptive coping strategies like problem-solving and seeking social support instead of resorting to aggression (Lei et al., 2020; Li et al., 2024).

Although the connections between physical activity and self-control, as well as between self-control and aggression, are well-documented, notable research gaps persist. Many earlier studies have predominantly used cross-sectional designs, restricting the capacity to determine causal relationships or the sequence of events over time. Furthermore, limited research has specifically examined self-control as a mediator in the link between physical activity and aggressive behavior.

Grounded in gender socialization theory (Eagly and Wood, 2016), boys and girls develop distinct behavioral tendencies and emotional regulation strategies that shape their responses to physical activity and self-control. Societal expectations encourage girls to exhibit cooperation, emotional regulation, and prosocial behaviors, while boys are more likely to display competitiveness, dominance, and externalizing behaviors such as aggression (Van Der Graaff et al., 2018; Wang et al., 2021). These differences suggest that physical activity may have stronger effects on enhancing self-control and reducing aggression in girls, as their prosocial behaviors are more closely tied to peer relationships (Chu and Cui, 2024). In contrast, boys’ tendencies toward withdrawal and externalizing behaviors may moderate the impact of physical activity on self-control and aggression (Wang et al., 2015; Malonda et al., 2019).

In addition, grade level corresponds to different developmental stages and school ecologies that may modulate the strength and expression of the associations among study variables. Our sample spans Grade 7 (early adolescence) and Grade 10 (mid-adolescence). Early adolescence is marked by rapid biological and socioemotional changes, heightened peer influence, and still-developing emotion regulation (Ho et al., 2025), whereas mid-adolescence features more mature executive functioning and self-control alongside greater academic pressure and peer competition (Sahi et al., 2023). These developmental and contextual differences may alter the strength and direction of the physical activity→self-control→aggressive behavior pathway (e.g., a stronger mediating role of self-control with advancing grade, or shifts in how aggression manifests). Therefore, in addition to testing the overall longitudinal mediation while controlling for gender and grade, we examine potential differences by gender and grade to identify heterogeneous effects and inform targeted, tiered interventions.

Earlier studies investigating the links between physical activity, self-control, and aggressive behavior have predominantly relied on cross-sectional methods (Xu et al., 2025; Yu et al., 2024). While valuable, such studies are limited in their ability to establish causal relationships or determine the directionality between variables. Additionally, they do not account for the dynamic changes individuals experience over time. To overcome these limitations, this study integrates the theories of emotion regulation, cognitive control, and self-control, adopting a longitudinal design with data collected at three time points. This approach enables a more precise analysis of the relationships among physical activity, self-control, and aggressive behavior. The study’s hypotheses are as follows: H1: Physical activity at Time 1 (T1) will negatively predict adolescents’ aggressive behavior at Time 3 (T3). H2: Self-control at Time 2 (T2) will mediate the association between physical activity at T1 and aggressive behavior at T3. H3: The indirect effect of physical activity at T1 on aggressive behavior at T3 via self-control at T2 will differ between boys and girls. H4: The indirect effect of physical activity at T1 on aggressive behavior at T3 via self-control at T2 will differ between Grade 7 and Grade 10 students. The hypothesized model is illustrated in Figure 1.

This study is a longitudinal cohort study on the relationship between physical activity and mental health among Chinese adolescents, with data drawn from the first three waves of surveys. Given the need to maintain the longitudinal nature of the study design, only students in grades 7 and 10 were selected as participants. The initial survey was conducted from March to April 2024 in eight middle schools in central China, chosen due to considerations of convenience and budget constraints. After removing invalid questionnaires at baseline (T1), a total of 2,011 eligible participants were retained. All participants at baseline were reassessed 6 months later (T2), and another 6 months after T2 for the third assessment (T3). Ultimately, 1,927 participants remained at T2, and 1,866 participants completed the T3 assessment. The sample included 1,021 males (54.7%) and 845 females (45.3%), with 1,069 from urban areas (57.3%) and 797 from rural areas (42.7%). There were 1,122 students in grade 7 (60.1%) and 744 in grade 10 (39.9%), with a mean baseline age of 13.90 ± 1.65 years. The recruitment and retention of participants are depicted in Figure 2. Variables measured at each wave were as follows: T1 (baseline), demographics, physical activity, and aggressive behavior; T2 (6-month follow-up), self-control; and T3 (12-month follow-up), aggressive behavior.

The main reasons for sample attrition included students being absent or refusing to participate on the day of the assessment. Ultimately, participants who completed all three waves were included in this study. A comparison of aggressive behavior scores was conducted between participants retained through all three waves and those who dropped out after the first wave. The results showed no significant differences between the attrition and retained samples on the outcome variable (aggressive behavior; p > 0.05).

Participants were informed that the study followed a longitudinal design, with additional rounds of data collection planned over the next year. They were asked to use a code name to ensure their responses could be matched across the three rounds. Participation was entirely voluntary, with no compensation provided. Data collection occurred in participants’ classrooms, where students completed the survey electronically in the school’s computer lab. To ensure the validity of the data, staff received uniform training prior to data collection to maintain consistent protocols, an online system recorded total response times, and a lie-detection item was included to assess the attentiveness of participants’ responses. All participants were assured that their privacy would be protected and were informed of their right to withdraw at any time if they felt uncomfortable. The study procedures were approved by the Institutional Review Board (IRB-JXNU-PEC-20231107).

The study used SPSS 26 to conduct descriptive statistics, reliability tests, and Pearson’s correlation analyses for the variables. Prior studies have indicated that gender and grade level are correlated with aggressive behavior among adolescents (Chun and Mobley, 2010; Cheng et al., 2024). Moreover, initial levels of aggressive behavior can impact subsequent changes in aggression. Consequently, gender, grade level, and baseline aggressive behavior were included as control variables in the analysis. Gender was dummy-coded (0 = male, 1 = female), and grade was dummy-coded (0 = grade 7, 1 = grade 10).

Before conducting the mediation analysis, we tested the assumptions. Results showed that the skewness and kurtosis coefficients of the variables were all within the ±1.5 range, indicating that the data met the assumption of approximate normality (Mardia, 1970). The Variance Inflation Factor (VIF) values were all below 5, confirming the absence of multicollinearity issues (Kim, 2019). These findings confirmed that the dataset met the necessary assumptions for mediation analysis.

To address the limitations of cross-sectional data, this study employed a semi-longitudinal design and conducted simple mediation analysis using Model 4 in the Process Macro software. This analysis examined the relationship between physical activity at T1 and aggressive behavior at T3 through self-control at T2 in the longitudinal data. Gender, grade level, and baseline aggressive behavior were included as control variables in the analysis. In addition, we conducted multi-group structural equation modeling in AMOS 28.0 (maximum likelihood estimation) to test whether the mediation pathway differed by gender (boys vs. girls) and by grade (Grade 7 vs. Grade 10).

Table 1 presents the descriptive statistics and intercorrelations of the main variables. Physical activity at T1 is negatively correlated with aggressive behavior at T1 and T3 and positively correlated with self-control at T2; self-control at T2 is also negatively correlated with aggressive behavior at T1 and T3, providing a basis for the mediation analysis. The variables exhibit approximately symmetric distributions (e.g., physical activity: skewness = −0.058, kurtosis = −1.171; the skewness and kurtosis of the other variables fall within ranges consistent with approximate normality). Scatterplot diagnostics indicate relationships that are roughly linear and homoscedastic, with no influential outliers detected, thereby supporting the use of Pearson’s correlation coefficients. Meanwhile, physical activity (0–7 days in the past week) was treated as a quasi-continuous count variable to preserve information, in line with Huang et al.’s approach (Huang et al., 2025a). As a robustness check, Spearman’s rank correlations showed directions and significance levels that were broadly consistent with those of Pearson’s correlations.

We utilized Model 4 of the SPSS macro developed by Hayes (Preacher and Hayes, 2008) to examine the longitudinal mediating effect of self-control. After accounting for grade, gender, and baseline aggressive behavior, the results presented in Table 2 demonstrate that physical activity at T1 significantly and positively predicts self-control at T2 (β = 0.164, t = 7.610, p < 0.001) and significantly and negatively predicts aggressive behavior at T3 (β = −0.083, t = −4.810, p < 0.001). Furthermore, self-control at T2 significantly and negatively predicts aggressive behavior at T3 (β = −0.215, t = −11.689, p < 0.001). The mediation effect estimate is −0.035, with a Bootstrap 95% confidence interval ranging from −0.046 to −0.025. These findings indicate that self-control at T2 partially mediates the relationship between physical activity at T1 and aggressive behavior at T3, thereby supporting hypothesis 2 (see Figure 3 and Table 3).

To assess the generalizability of the indirect-effects model (physical activity at T1 → self-control at T2 → aggressive behavior at T3), we conducted multi-group comparisons following Shao et al. (2018). Specifically, we ran two sets of multi-group SEM analyses by gender (boys vs. girls) and by grade (Grade 7 vs. Grade 10). For gender, freely estimated group models fit well (Table 4), supporting multi-group comparability. Constraining measurement weights (Model 2 vs. Model 1) did not worsen fit (Δχ² = 0.38, Δdf = 2, p = 0.83), indicating measurement-weight invariance, whereas constraining structural weights (Model 3 vs. Model 2) significantly degraded fit (Δχ² = 8.17, Δdf = 3, p = 0.04), indicating gender differences. For grade, freely estimated group models also fit well (Table 4). Measurement-weight constraints preserved fit (Model 2 vs. Model 1: Δχ² = 3.22, Δdf = 2, p = 0.20), but structural-weight constraints significantly worsened fit (Model 3 vs. Model 2: Δχ² = 28.01, Δdf = 3, p < 0.001), indicating grade differences.

Subsequently, we conducted pairwise multi-group comparisons by gender and by grade, using the critical ratio (CR) to test between-group path differences, with |z| ≥ 1.96 indicating significance. A positive CR value indicates that the path coefficient is larger in the first group (e.g., boys or Grade 10), whereas a negative CR value indicates that the path coefficient is larger in the second group (e.g., girls or Grade 7).

The gender comparison showed a significant gender difference in the negative effect of self-control at T2 on aggressive behavior at T3 (CR = −2.03, p < 0.05), indicating that this path was stronger for girls than for boys. However, no significant gender differences were observed for the paths from physical activity at T1 to self-control at T2 or to aggressive behavior at T3 (boys: β = 0.27, −0.18; girls: β = 0.25, −0.11; all p < 0.05). See Figure 4 for the multi-group comparisons by gender.

The grade comparison indicated a significant grade difference in the positive effect of physical activity at T1 on self-control at T2 (CR = −4.08, p < 0.05), suggesting that this path was stronger for Grade 7 students than for Grade 10 students. No significant grade differences were found for the paths from self-control at T2 to aggressive behavior at T3 or from physical activity at T1 to aggressive behavior at T3 (Grade 7: β = −0.68, −0.17; Grade 10: β = −0.61, −0.13; all p < 0.05). See Figure 5 for the multi-group comparisons by grade. Overall, gender primarily moderates the “self-control at T2 → aggressive behavior at T3” path, whereas grade primarily moderates the “physical activity at T1 → self-control at T2” path; the other paths are significant but not significantly different between groups.

Consistent with Hypothesis 1 and prior work (Khazaie et al., 2023; Garcia et al., 2021; Yu et al., 2024), we found that higher frequencies of adolescent physical activity predicted lower levels of aggressive behavior 1 year later. Physical activity not only improves physical fitness but also appears to curb aggression by enhancing emotion regulation and strengthening self-control, reinforcing its benefits for mental health and behavioral regulation. This aligns with findings that physical activity is positively associated with mental health, including better emotion regulation and greater subjective wellbeing (Zhu et al., 2022; Potoczny et al., 2022; Carl et al., 2024; Stocker et al., 2020; He et al., 2023; Li et al., 2022; Wennerhold and Friese, 2023).

Physical activity reduces aggressive behavior by enhancing psychological regulation, improving behavior management, and fostering social interaction. According to self-control theory, regular training and goal setting can strengthen individuals’ ability to control impulsive behaviors, helping adolescents better manage their emotions and behaviors, thereby reducing impulsivity and aggression (Franco et al., 2016). From the perspective of emotion regulation, physical activity can lower anxiety and anger, alleviate stress, and decrease the likelihood of aggressive behavior (Ouyang and Liu, 2023; Hamer et al., 2012). For example, regular physical activity helps adolescents remain calm when facing stress and setbacks, preventing emotional outbursts that may lead to aggression (Norris et al., 1992; Wang et al., 2024). It also boosts self-confidence and self-esteem, further reducing the likelihood of aggressive behavior (Ruiz-Ranz and Asín-Izquierdo, 2024; Yin et al., 2025).

Additionally, physical activity creates a positive social environment, enhancing adolescents’ social skills and sense of belonging through teamwork and fair competition, which reduces both intra-group and inter-group conflicts and suppresses aggressive behavior (Sansi et al., 2021; Yang et al., 2024; Kwon et al., 2024). Regular participation in physical activity also strengthens self-discipline and goal orientation, supporting better behavior and emotion management (Li et al., 2024; Van Sluijs et al., 2021). Through physical activity, adolescents can release excess energy, relieve stress, and learn cooperation and respect, all of which contribute to further reductions in aggressive behavior (Nie et al., 2025; Ahmed et al., 2023). Moreover, the mechanisms through which physical activity exerts its protective effects can also be understood through cognitive control theory (Zhu et al., 2023; Cya et al., 2019). This theory posits that physical activity enhances executive functions, thereby improving impulse control and reducing the likelihood of aggressive behavior.

Consistent with Hypothesis 2, self-control at T2 exerted a significant longitudinal mediating effect on the association between physical activity at T1 and aggressive behavior at T3. This result indicates that physical activity may reduce aggressive behavior by strengthening self-control. As a key component of psychological resilience, self-control helps individuals manage stress and challenges that might otherwise trigger aggression (Chester, 2024; Bahjatunnufuz et al., 2024). This conclusion aligns with previous cross-sectional research, which has similarly reported that physical activity is associated with lower levels of aggression through enhanced self-control (Xu et al., 2025).

Self-control theory (Gul and Pesendorfer, 2004) posits that self-regulatory capacities can be strengthened via systematic behavioral training. Physical activity provides a prototypical training context that combines repeated practice, goal setting, feedback, and incremental challenge (Rahayu et al., 2020; Murray et al., 2021). Adolescence is a sensitive period for the maturation of neurocognitive systems supporting self-control; repeated practice in structured, socially scaffolded contexts facilitates the internalization of behavioral norms (Zeinab Mousavi et al., 2024). Accordingly, regular exercise routines and goal-directed activity schedules may foster inhibitory control, delay of gratification, and emotion regulation (Berli et al., 2016). Experimental and neurobiological evidence is consistent with this account: lifestyle interventions incorporating exercise have improved cognitive control and self-regulation (Xiang et al., 2019), and 12 weeks of moderate-intensity aerobic exercise has been shown to increase gray matter density in the dorsolateral prefrontal cortex, a key substrate of inhibitory control (Northey et al., 2020).

Building on these findings, it is important to consider the protective role of self-control in mitigating aggression during adolescence. Higher self-control is robustly and negatively associated with aggressive behavior (Plessen et al., 2023; Wang et al., 2025). Individuals with stronger self-control demonstrate better emotion regulation and behavioral inhibition, making them more likely to select norm-consistent, prosocial coping strategies in conflict situations (Van Den Bekerom et al., 2025; Wenzel et al., 2024). For adolescents, this translates into greater use of adaptive responses—such as problem-solving and help-seeking—rather than impulsive aggression when confronted with interpersonal stressors (Tamm et al., 2018; Schwartz-Mette et al., 2021).

The multi-group analyses revealed that the effect of self-control on aggressive behavior differed by gender, whereas the effect of physical activity on self-control differed by grade. These findings suggest that both sociocultural factors and developmental stages jointly shape the mechanisms linking physical activity, self-control, and aggression.

First, regarding gender, the results supported Hypothesis 3. Our study found that the inhibitory effect of self-control on aggressive behavior was more pronounced among girls, which is inconsistent with the findings of Bürgler and Hennecke (2024). This difference can be explained by sociocultural norms and gender socialization. In East Asian cultural contexts, girls are typically expected to display qualities such as obedience and harmony (Li and Xie, 2020; Dong et al., 2025). Consequently, when girls engage in aggressive behavior, it is often perceived as a greater deviation from social norms. Under such conditions, even small differences in self-control may have amplified effects on whether girls suppress aggression (Siregar, 2020). In contrast, boys’ aggression is, to some extent, reinforced through socialization processes, as competitiveness and toughness are socially encouraged (Stanaland et al., 2024; Björkqvist, 2018). In conflict situations, even boys with a high level of self-control may display aggression due to peer pressure or social expectations, thereby weakening the inhibitory role of self-control (Farrell et al., 2017). Previous studies have also demonstrated that gender moderates the association between callous-unemotional traits and antisocial or aggressive behaviors, with stronger effects observed among girls (Falcón et al., 2021). These findings indicate that gender not only influences the mean levels of self-control but, more importantly, determines the extent to which differences in self-control translate into behavioral outcomes.

Second, regarding grade, the results supported Hypothesis 4. We found that the facilitative effect of physical activity on self-control was significantly stronger among seventh graders than among tenth graders. From the perspective of developmental psychology and neuroscience, early adolescence (approximately 12–14 years) is a critical period for the development of self-control, as the prefrontal cortex and executive functions undergo rapid maturation, and adolescents exhibit heightened neural plasticity in emotion and behavioral regulation (Zondervan-Zwijnenburg et al., 2020). During this stage, physical activity can significantly enhance the development of self-control through mechanisms such as goal setting, adherence to rules, and emotional regulation (Graham et al., 2025; Bermejo-Cantarero et al., 2024; Pascoe et al., 2020). In contrast, during middle adolescence (approximately 15–17 years), self-control becomes relatively more stable as the maturation of the prefrontal cortex slows and neural plasticity decreases. Consequently, the capacity of physical activity to further enhance self-control is reduced during this stage. Additionally, adolescents in middle adolescence face higher cognitive resource demands, primarily due to academic pressure and the burden of high-stakes examinations (Wan et al., 2025; Cheng et al., 2025). This pressure may compete with the cognitive and emotional resources required for the self-regulatory benefits of physical activity. For instance, the cognitive and emotional efforts needed to manage academic stress might limit adolescents’ ability to fully absorb the benefits of exercise (Guan et al., 2024; Recchia et al., 2023; Yan et al., 2024). Therefore, developmental differences in neural plasticity and competing cognitive demands likely contribute to the stronger relationship between physical activity and self-control observed among younger adolescents.

In summary, the moderating roles of gender and grade highlight the joint influences of sociocultural expectations and developmental stages on the relationships between physical activity, self-control, and aggression. Specifically, the stronger inhibitory effect of self-control on aggressive behavior among girls reflects the impact of sociocultural norms and gender socialization, while the more pronounced facilitative effect of physical activity on self-control in early adolescence underscores the importance of developmental timing and neural plasticity. These findings suggest the need for tailored interventions: for girls, programs should focus on leveraging physical activity to strengthen self-control, helping them better manage aggression within the context of sociocultural expectations. For younger adolescents, particularly in middle school, schools should prioritize providing ample opportunities for physical activity to capitalize on neural plasticity, enhance self-regulation, and reduce the risk of aggression.

This study investigated the relationships between physical activity, self-control, and aggressive behavior in adolescents, emphasizing the mediating role of self-control and the moderating effects of gender and grade level. However, several limitations should be noted. First, the data were collected from schools in central China using convenience sampling. While this method facilitated efficient data collection, it limits the generalizability of the findings, as the sample may not fully represent adolescents from other regions or cultural backgrounds. Future studies should include samples from diverse geographic areas and employ random sampling to improve representativeness and generalizability. Second, the reliance on self-report scales may introduce social desirability bias. To mitigate this, future research could incorporate additional methods, such as behavioral observations or reports from teachers and parents, to provide a more comprehensive understanding of the variables. Third, the study focused solely on the relationship between physical activity, self-control, and aggressive behavior. Other potential protective factors, such as family support and socioeconomic status, were not examined. These factors may interact with self-control and aggression in meaningful ways, and future studies should explore their roles in these relationships. Fourth, physical activity was assessed using a single-item measure of weekly frequency. While this approach simplified data collection, it has limited reliability and does not capture other dimensions of physical activity, such as intensity, duration, and context (e.g., team vs. individual activities). This may have weakened the observed effect sizes. Future research should use validated multi-item scales or objective measures (e.g., accelerometers) to assess physical activity more comprehensively in terms of frequency, intensity, duration, and type. Finally, although the longitudinal design strengthens causal inference, the study remains observational. Future research should adopt experimental designs to rigorously test the causal effects of physical activity and self-control on aggressive behavior in adolescents. Such methods would provide stronger evidence for causality and enhance the applicability of the findings in real-world contexts.