Growing evidence suggests that physical activity, which is defined as bodily movement produced by skeletal muscles that expends energy, is associated with a range of health benefits and specifically can increase academic engagement through psychosocial and neurocognitive mechanisms (Owen et al., 2016). Researchers have addressed the relationship between physical activity and academic engagement from diverse perspectives. On the one hand, a meta-analysis of the positive correlation between physical fitness, a significant byproduct of regular physical activity, and academic achievement indicates that active students tend to perform better academically (Castelli et al., 2007). Moreover, one study investigated the acute cognitive effects of physical activity, demonstrating that brief bouts of exercise can enhance attention, memory, and executive functioning, which are crucial for academic participation (Tomporowski et al., 2011).

On the other hand, focused interventions might trigger scholarly involvement. Doolittle researched structured physical activity programming in schools to demonstrate their potential to enhance students’ enthusiasm for learning and classroom participation (Doolittle, 2016). Similarly, researchers analyzed resource utilization strategies for promoting physical activity, identifying key facilitators that increase students’ physical and academic engagement (Wang et al., 2023). Classroom research has discovered that the integration of physical activity breaks into the curriculum improves students’ concentration, motivation, and overall participation in the learning process (Spring et al., 2023). Notably, some evidence reported that chronic physical exercise is linked to long-term cognitive gains from longitudinal measurement (Lambourne and Tomporowski, 2010) and proved the high correlation between physical activity and engagement, specifically in obese children (McCoy and Rupp, 2021) validating the long-term benefit of chronic exercise for academic performance, especially in vulnerable populations. Despite these advances, critical gaps persist in the literature. The mechanisms through which physical activity influences academic engagement across development are limited, especially among middle school students undergoing fast physical, cognitive, and social changes. Thus, we propose the following hypothesis:

Interpersonal relationships—such as with family, peers, romantic partners, and community members—are a fundamental part of human life that has a significant influence on behavior, health, and well-being across the lifespan. Klem and Connell (2004) first proposed the paramount position of interpersonal relationships in school engagement. Since then, research has investigated the relationship between interpersonal relationships and personal aspirations (Collie et al., 2016). Most informative are peer relationship studies (Shao et al., 2024), flipped classroom settings (Li, 2021), teacher-student, and peer relationship interactions (Košir and Jugović, 2025). Specifically, peer relationships are impactful in fostering feelings of belonging and cooperation for middle school students (Wentzel, 1991).

There is robust evidence demonstrating a bidirectional relationship between physical activity and social relationships. Evidence indicates that physical activity enhances interpersonal relationships through the direct construction of self-confidence and social competence, resulting in better peer relations (Spring et al., 2023). Physical activity is a buffer against stress for university students, increasing the quality of relationships and overall psychological health (Joo, 2018). Physical activity improves social skills in teens, particularly in key transitional periods such as middle school (Hu and Tan, 2025). The above indicates not just the promotion of health behavior through physical activity but also stress reduction, enhanced self-efficacy, and improved social skills at various stages of life.

Additionally, empirical evidence consistently highlights the pivotal role played by interpersonal relationships in academic engagement. Interpersonal relationships between teachers and students, and peers are direct motivators of engagement (Collie et al., 2016), and indirect influences through psychological mediators such as personal best goals and self-efficacy (Benlahcene et al., 2024). These relationships manifest distinct roles in social contexts: teacher interactions facilitate cognitive engagement through intellectual challenge and feedback, peer relationships facilitate emotional engagement by satisfying belonging needs, and family relationships offer instrumental support that removes practical barriers to learning (Zhu et al., 2025). During critical developmental phases, the teacher assumes a relational amplifier role, augmenting the positive impact of peer interactions on engagement by satisfying students’ fundamental need for relatedness (Košir and Jugović, 2025). Combined, these findings exhibit interpersonal relationships that promote academic interaction through multiple channels. Thus, we hypothesize the following:

Interpersonal relationships serve as the foundational framework through which social support influences health behaviors, as articulated by social support theory (Baron et al., 1990). Instrumental support reduces practical barriers to physical activity, and parental modeling of active lifestyles and joint activities (e.g., family walks, cycling) correlate with higher physical activity levels in adolescents (Sallis et al., 2015). Emotional support addresses psychological barriers like lack of confidence or isolation; spouse support among the elderly, for example, ensures prolonged physical activity maintenance by inspiring opportunity for group exercise, which encourages motivation and reduces seclusion (FioRito and Routledge, 2020). Through peer networks that provide informational support and encouragement for physical activity, adolescents are encouraged to participate in moderate-to-vigorous activity to attain group membership (Martins et al., 2021).

Support from teachers is particularly important in fulfilling students’ fundamental psychological needs for relatedness and competence (Klem and Connell, 2004), encouraging engagement directly while facilitating achievement indirectly (Tao et al., 2022). This teacher-student relationship functions with other origins of critical support: parents have valuable structural and instrumental resources to offer, as well as good emotional support and cooperative learning provision (Chen, 2005; Qudsyi et al., 2020). These relational supports indicate that teacher scaffolding is particularly essential for underachievers, whereas high achievers are more likely to be helped by peer-based learning relations (Moreira et al., 2018). These findings highlight the multidimensional nature of academic support systems, with different relationships playing different and interdependent roles in promoting student achievement.

Evidence of the relationship between support and physical activity shows age-dependent patterns and contextual variations across populations. In later life, social support is essential to facilitate physical activity by removing some primary barriers, such as social isolation, mobility issues, and motivational deficits (Lindsay Smith et al., 2017). In contrast, peer support is the predominant predictor of adolescent physical activity participation, especially for team sport and vigorous activity, which attests to the developmental importance of social belonging and identity formation during this life stage (Prochaska et al., 2002; Mendonça et al., 2014). Meta-analytic findings specifically reveal a moderate positive association between peer support and physical activity in adolescent girls, with peer influence outperforming parental support effects consistently, while these relations demonstrate cultural and methodological variability (Laird et al., 2016). These general research findings reveal the relationship between support and physical activity. The elderly influence physical activity by support, but in adolescents, there is an association, and the direction of causality is uncertain. Probably, a bidirectional influence exists between them.

Besides interpersonal relationships, other research shows that a wide support system can enhance student engagement and academic achievement through unique and complementary processes. Landmark research put in place the foundation of knowledge of academic support systems as an aggregate process of enabling student achievement (Chen, 2005). Subsequent research has all examined these processes, demonstrating the way school-based support fits with academic achievement (Moreira et al., 2018) and operates dynamically through day-to-day interactions (Robayo-Tamayo et al., 2020). Meta-analytic estimates (Tao et al., 2022) have subsequently measured these associations, with concurrent studies on social support constructs (Jayarathna, 2015; Gutiérrez et al., 2017) further elucidating their role in explaining variance in student engagement (Wilcox et al., 2016). These initial studies collectively emphasize support systems as multifaceted networks that extend beyond interpersonal relationships to include structural and institutional levels. Strong interpersonal relationships, forged through physical activity, cultivate a network of trust that enables students to effectively seek and receive academic support—whether from peers, teachers, or parents—by reducing perceived barriers to support-seeking and enhancing the perceived reliability of assistance.

The emerging theoretical framework comprises a series of initial findings. Physical activity is association with social competence (Doolittle, 2016) and academic resource utilization (Wang et al., 2023) has been well-documented, as have the bidirectional relationships between relational quality and support systems (Robayo-Tamayo et al., 2020). Additional insights come from studies examining parent–child communication (Geng et al., 2023) and the interplay between self-esteem and engagement (Zhao et al., 2021), collectively contributing to a comprehensive understanding of academic engagement determinants.

Based on the relationships between variables identified in the above literature and using relevant scales that can measure variables, we explored variables that may be associated with academic engagement, including physical activity, interpersonal relationships, and academic support, which can measure academic resources and different sources of support, interpersonal relationships and academic support are related to other variables among them, and hypothesized that there is a mediation effect between the four variables. Interpersonal relationships and academic support were hypothesized to be mediating variables, playing a sequential mediating role in the physical activity and academic engagement of middle school students. This research addresses critical gaps in existing literature by unraveling the cognitive mechanisms through which physical activity influences academic engagement among middle school students. Specifically, we move beyond documenting mere correlational associations to explore how social interactions and psychological resources sequentially mediate this relationship. Furthermore, the findings offer tangible implications for educational practice, including the design of school - based intervention programs, formulation of evidence - informed policies, and development of parental support strategies. These dual contributions—advancing theoretical understanding of adolescent learning processes and providing actionable insights for real - world application—highlight the study’s interdisciplinary significance.

Within this model framework, we proposed the following hypotheses:

All hypotheses are shown in Figure 1.

Before the research, the participants were adequately trained to complete the test with a complete understanding of the requirements and instructions. Cluster sampling was employed in the study, where middle school students from two schools in Sichuan Province, China, were sampled. Having obtained informed consent from schools, teachers, and parents, the students completed the survey anonymously. Written informed consent was elicited from each participant, and they were informed of their right to refuse to participate.

760 students aged between 13 and 17 years were involved in the survey, out of which 746 participants (96.84%) provided valid information. 365 participants were male (48.9%), and 381 were female (51.1%). Participants were divided grade-wise as follows: 161 students of seventh standard (21.6%), 154 students of the eighth standard (20.6%), 144 students of the ninth standard (18.8%), 145 students of tenth standard (19.4%), and 146 students of eleventh standard (19.6%).

The Chengdu Sports University Research Ethics Committee permitted the current research (NO. CTYLL2025004). Procedures were carried out according to proper guidelines and rules and in harmony with the Declaration of Helsinki on ensuring the safety of human participants in research. Written informed consent was voluntarily given by all the participants and their parents.

The present study employed a double-platform analysis approach using RStudio (Version 4.4.2; R Core Team, 2024) and IBM SPSS Statistics (Version 27.0) and the results from both software tools were fully congruent to ensure the robustness of analysis and verification of findings. The main analysis was conducted using the BruceR package (Bao, 2024), which is a fast statistical package designed for behavioral and social science studies. Integrated functionality offered by BruceR touches every phase of the research cycle, starting from high-level data management capabilities like clever data import/export operations in different file formats to variable recoding procedures that consume less memory. The psychometric analysis feature within the package simplifies scale scoring through coding routines as well as reliability checking. To deal with more sophisticated statistical modeling, BruceR incorporates an easy-to-use extension of Hayes’ (Hayes, 2018) PROCESS procedure for moderation and mediation analysis with bootstrap estimation of confidence intervals - a feature that is particularly helpful in testing complex theoretical models. The package also shines with an automated reporting process producing publication-quality output and visual tools incorporated.

Our analysis process involved two consecutive steps. The initial data preparation stage was typified by assertive data cleansing methods, including outlier identification and exclusion criteria application, followed by careful descriptive statistics to identify the underlying characteristics of our dataset and examine relationships. The second analytical phase, theoretically grounded in the current literature, aimed to test a serial mediation model to explain complicated relationships between our four key variables. By contemporary methodological convention (Hayes, 2018), we employed bias-corrected bootstrap estimation with 5,000 resamples to yield 95% confidence intervals for all indirect effects. The statistical significance criterion was that the confidence intervals did not contain zero at the conventional α = 0.05 level.

To assess the possibility of common method variance (Aguirre-Urreta and Hu, 2019) in using only self-reports, we conducted a Harman single-factor test (Podsakoff et al., 2003). Exploratory factor analysis without rotation indicated that 13 distinct factors emerged from the test with a cumulative variance of 61.00%. Most significantly, the first principal component explained a mere 21.10% of variance, a far cry from the magic mark of 40%, which would indicate common method bias (Zhou and Long, 2004). The pattern of results thus suggests that common method bias is not a threat to the validity of our findings.

Table 1 presents the descriptive statistics for all study variables among our sample of middle school students (N = 746). Physical activity levels, as measured by the Physical Activity Rating Scale, were found to be moderate (M = 43.82, SD = 22.65) with a nearly normal distribution. Academic engagement yielded the highest mean score (M = 86.01, SD = 22.71), while interpersonal relationship difficulties showed moderate levels (M = 18.77, SD = 6.13). Academic support scores were positively skewed (M = 81.66, SD = 16.58), with generally high perceived support. Demographic variables gave measures, with grade levels being almost symmetrical (M = 8.95, SD = 1.43), genders almost an even split (51.10% female), and 71.60% singletons only child. All variables gave adequate parameters of normality (Skewness: −0.94 to 0.40; Kurtosis: −2.00 to 3.05), and there were no signs of floor or ceiling effects in measures.

The analysis revealed significant Pearson correlations among the variables shown in Figure 2. Several significant relationships emerged: (i) Grade level showed negative correlations with physical activity (r = −0.15, p < 0.001), interpersonal relationships (r = −0.27, p < 0.001), and academic engagement (r = −0.28, p < 0.001); (ii) Physical activity was positively correlated with interpersonal relationships (r = 0.21, p < 0.001), academic support (r = 0.16, p < 0.001), and academic engagement (r = 0.39, p < 0.001); (iii) Interpersonal relationships showed positive relationships with both academic support (r = 0.13, p < 0.001) and engagement (r = 0.45, p < 0.001); (iiii) Academic support was positively correlated with academic engagement (r = 0.17, p < 0.001).

Notably, the strongest correlation was found between interpersonal relationships and academic engagement (r = 0.45). In contrast, demographic variables such as gender and only child status were generally weakly associated with the other measured variables. These results highlight the complex interactions among grade, physical activity, interpersonal relationships, and academic factors, providing valuable insights into the phenomenon under study.

The regression analysis (Table 2) yielded several key findings regarding the factors influencing academic engagement. Physical activity showed a strong positive effect on academic engagement (β = 0.360, p < 0.001), while interpersonal relationships demonstrated a substantial positive prediction with academic engagement (β = 0.338, p < 0.001). Academic support exhibited a modest positive effect on engagement (β = 0.078, p < 0.05). Grade level negatively predicted academic engagement (β = −0.224, p < 0.01) and interpersonal relationships (β = −0.241, p < 0.001) significantly, but positively predicted academic support (β = 0.060, p > 0.05). Notably, physical activity had a positive effect on interpersonal relationships (β = 0.173, p < 0.001), suggesting a potential protective effect. The model diagram of serial mediation is shown in Figure 3.

The final comprehensive regression model (Model 4) provided a robust explanation for the variance in academic engagement, accounting for 31.8% of it (R² = 0.317). This outcome underscores the combined influence of physical activity, interpersonal relationships, academic support, and grade level in shaping academic engagement.

Bootstrap analysis indicated that the mediation effect was strong in the hypothesized model. The total indirect effect was statistically significant and accounted for roughly 20.28% of the total effect. Of the individual indirect paths, the mediation effect through interpersonal relationships was the strongest and accounted for roughly 16.54% of the total effect. The indirect effect through academic support and serial mediation of interpersonal problems and academic support was small, at 3.27% and 0.42%, respectively, but significant. Importantly, the direct effect was significant, accounting for 79.72% of the total effect (see Table 3).

Our findings indicate several significant trends in adolescent development that need to be carefully monitored. The moderate physical activity levels observed are consistent with recent longitudinal studies (Guthold et al., 2020); similar decline is most likely indicative of increased scholastic burdens and shifting social priorities during this age bracket. The relatively high academic engagement scores, despite significant individual differences, are consistent with cross-cultural studies of East Asian education systems (Kitamura et al., 2022), in which academic achievement is highly prized, causing a situation of fewer social relationships. These are consistent with contemporary developmental theories because of the especially crucial social adjustment in early adolescence (Blakemore, 2019). Our results extend this by showing how physical activity can serve as a buffer against social adversity, in line with new neurodevelopmental evidence (Orben et al., 2020). Academically, our findings contribute to the mounting literature on school effects (Wang and Hofkens, 2020), especially in collectivist cultures where institutional support is key to student outcomes.

In correlational terms, the negative correlation between grade level and physical activity is consistent with recent multinational surveys (Aubert et al., 2022). While the strong correlation between physical activity, like exercise, and academic engagement validates emergent hypotheses of exercise-mediated neuroplasticity (Morris et al., 2020) and is consistent with emerging intervention trials in schools (Daly-Smith et al., 2021). The correlations of academic support may imply complex influences, which require additional examination using experience sampling (Larson et al., 2022).

This suggests the need for intervention at critical developmental stages to sustain motivation. The physical activity prediction of academic motivation is in keeping with Hypothesis H1, reinforcing earlier work (Sallis et al., 2015) that has identified physical activity as enhancing cognitive resources (e.g., attention, memory, executive function) and motivational states, and thus providing a stronger foundation for active learning. For instance, Donnelly et al. (2016) found that school physical activity interventions significantly improved on-task behavior and academic achievement, particularly in math and reading. These findings are in line with neuroscientific research (Hillman et al., 2014) that aerobic exercise increases prefrontal cortex activity and neurogenesis, both of which are critical for self-regulation and sustained attention during academic performance.

Given this strong evidence, this study underscores the importance of physical education as an effective means of strengthening academic involvement. Further support comes from a longitudinal study (Booth et al., 2014), in which adolescents engaging in regular physical activity had increased levels of academic attainment over time, which was mediated by increased concentration and reduced absenteeism. Policymakers and educators should thus prioritize physical activity for its health impact and as a key initiator of academic success.

Evidence highlights that shared physical activity experiences, such as team sports and group exercises, act as powerful boosters for cooperation, trust, and social interaction (Blakemore, 2019), aligning with Hypothesis H2. Eime et al. (2013) demonstrated that participation in sports in an organized setting produces social capital by enhancing the peer network and an overwhelming feeling of belonging, both of which feature prominently in adolescent psychosocial development. More studies revealed that school-based physical activity programs reduce social isolation and enhance prosocial behavior among vulnerable youth by creating environments in which positive social norms are supported (Lubans et al., 2016). Moreover, longitudinal studies reveal that students whose peer relationships were developed through cooperative physical activities have increased classroom engagement and report increased teacher and peer academic support (Oberle et al., 2019). This aligns with social interdependence theory (Johnson and Johnson, 2009), which contends that cooperative environments—like team sports—develop mutual accountability, shared goals, and motivational spillover into academic domains. A meta-analysis showed that physically active adolescents develop stronger teacher-student relationships and social skills, processes statistically mediating increased academic achievement (Visier-Alfonso et al., 2021).

Together, these studies illuminate the psychosocial process by which exercise enhances academic engagement, secondarily augmenting the social scaffolding (interpersonal relationships, teacher support) that supports it. For this to be maximized, schools would emphasize collective physical activities, e.g., team sports, class athletic competitions, or collective fitness challenges, whose explicit aim is both physical health and interpersonal quality.

This model examines the impact of academic support, where physical activity and social relationships are the prime roles. Physical activity’s positive impact on academic support (by way of greater peer/teacher interactions) is consistent with social support theory (Baron et al., 1990), in which active students are able to gain more instrumental benefits (e.g., study skills guidance) and emotional support. This is consistent with a longitudinal study (Howie et al., 2018) because physical activity had minimal direct influences on school performance, but its influence was powerfully mediated through increased teacher support and peer interaction. Similarly, Michael et al. (2023) showed in an intervention school that engaged students established more supportive teacher-mentor relationships that enabled increased support-seeking behaviors regarding studies. The weak but positive pathway from physical activity to participation in school work through support structures copies evidence (Owen et al., 2016), showing school sport activity enhanced social capital, indirectly impacting school results by opening access to learning assets.

This model emphasizes the interlinked physical activity and relational determinants of support system development, providing support for Hypothesis H3. The linear link between physical activity and participation in academic activity, and augmented mediation through support systems, is supported by evidence, whose meta-analysis established that when physical activity’s academic effect was greatest, it was when mediated by psychosocial factors like teacher support and classroom belonging Vasilopoulos et al. (2023). This suggests that schools might be able to reap academic dividends on exercise through explicitly developing social-support mechanisms in which such dividend payments are achieved.

This model specifies one critical serial mediation path through which physical activity influences academic engagement: by improving interpersonal relationships first, which in turn raises perceived academic support, and ultimately enhancing academic involvement. This chained process is squarely within Hypothesis H4 and is rooted in theoretical and empirical foundations. The sequence begins with the well-documented contribution of physical activity to social relationships (Eime et al., 2013). This first step in mediation is further supported, and network analysis found that physically active students hold more central positions in classroom social networks, providing them greater access to academic resources (Janssen and Connelly, 2021). These positive relationships are then succeeded by actual academic support, as predicted by social support theory (Baron et al., 1990). The longitudinal findings powerfully speak to this second link - though physical activity had modest direct contributions to academic performance, its influence was largely mediated through increased teacher support and peer collaboration (Howie et al., 2018). The complete serial mediation pathway echoes the social-academic integration model proposed (Oberle et al., 2021), who demonstrated that peer acceptance mediated the effect of physical activity on classroom engagement through subsequent increases in teacher support. Notably, Parker et al. (2022) demonstrated that interventions explicitly designed to leverage this two-pathway pathway (e.g., cooperative PE paired with peer tutoring) resulted in academic competencies larger than activity-only programs.

Our serial mediation model generates major theoretical contributions. The interpersonal challenges pathway informs updated social competence theories (Durlak et al., 2022), showing how physical activity promotes prosocial behaviors via skills acquisition and offering opportunities (Doolittle, 2016). The academic support findings inform existing models of resilience (Martin et al., 2022), with physical activity being a potential stimulus for support utilization. The biopsychosocial model is supported by our evidence of parallel mechanisms (Smith et al., 2021). The serial mediation model emphasizes that the academic advantages of physical activity are not isolated but emerge from an active network of relationships and support. While physical activity exerts a robust direct influence, its impact is further shaped by social dynamics in educational settings. By organizing programs on purpose to enhance interpersonal connections with collaborative physical activity and connect the links to academic supports. These approaches not just align with developmental science but also offer a scalable, affordable way to enhance overall student well-being and academic success.