Self-Determination Theory (SDT) proposes that empowering, task-involving climates support the internalization of sport motivation, whereas controlling, performance-pressuring climates are more likely to be associated with controlled forms of motivation. In line with this view, studies in sport settings have reported opposite links between climate and aggressive conduct: among collegiate athletes, disempowering climates are associated with higher aggression while empowering climates are associated with lower aggression, with reduced moral disengagement emerging as a mediator (Zhong and Wang, 2025; Gürpınar, 2023); at the team level, coaches’ autonomy support has been linked to stronger collective efficacy which, in turn, is associated with lower aggression, and the association between autonomy support and aggression is attenuated once efficacy is included (Kim and Choi, 2024). Because motivational climate is theorized to relate to the quality of sport motivation (autonomous/internalized vs. controlled) and to broader norm adherence and emotion–behavior tendencies, these findings provide contextual support for an individual-level association between sport motivation and aggression. At the same time, the association is contingent on sport type and population characteristics: converging surveys indicate that men, full-contact sports, and higher competitive tiers show elevated physical aggression and antisocial behavior (Mazzanti et al., 2025; Bovolon et al., 2024; Vertommen et al., 2022), and national data suggest that a substantial proportion of adolescents report exposure to psychological or physical violence within competitive systems (Parent and Vaillancourt-Morel, 2021); within-team dynamics also matter, as stronger positional competition and a violent motivational climate have been prospectively associated with later aggression (Worley et al., 2022; González-Hernández et al., 2025). Mechanistically, meta-analytic evidence shows a moderate-to-strong association between moral disengagement and antisocial behavior (Zhu et al., 2025), and longitudinal studies suggest that higher baseline self-compassion is associated with faster declines in antisocial behavior across a season and with a weaker positive association between narcissism and aggression (McEwan and Zhang, 2024; Zhang et al., 2024). Notwithstanding, much of the literature treats contextual/climate routes or moral/emotional pathways in isolation; few investigations simultaneously consider emotion and control channels within the same model before re-examining the direct sport-motivation–aggression link (i.e., reasons/quality of engagement rather than participation volume or sport type), and evidence in general undergraduate samples remains limited. Accordingly, we hypothesize H1: sport motivation is negatively associated with aggressive behavior.

The Process Model of Emotion Regulation partitions regulation into situation selection, attentional deployment, cognitive reappraisal, and response modulation; the more proficient individuals are with antecedent-focused strategies—especially reappraisal—the better they may be able to dampen negative arousal before impulses form (Gross, 2015). Building on this, the ability model of emotional intelligence (EI) specifies four branch skills—perceiving, using, understanding, and regulating emotion—proposing that higher EI is associated with greater use of reappraisal relative to late-stage suppression, which may relate to fewer aggression-relevant responses (Mayer et al., 2016). From a motivational standpoint, Self-Determination Theory (SDT) suggests that autonomous motivation is linked to more positive affect and stable engagement opportunities, which may support the acquisition and enactment of emotional skills; within the General Aggression Model (GAM), such skills may reduce aggressive responding by shaping appraisal and response selection processes (Laborde et al., 2016).

Empirical work aligns with these linkages on both segments of the chain connecting sport and aggression through EI. In university and secondary-school samples, either higher physical-activity frequency or stronger intrinsic motivation correlates positively with EI facets such as clarity and repair (Wang et al., 2020; Vaquero-Solís et al., 2020), and a cross-cultural survey indicates that athletes exhibit higher EI and that EI covaries positively with autonomous motivation (Rogaleva et al., 2024). At the curricular level, reviews of cooperative or reflective physical-education programs consistently report gains in EI accompanied by reductions in misconduct or aggression (Rico-González, 2023). On the EI-to-aggression link, undergraduates with higher EI score lower across multiple violence dimensions (Lindell-Postigo et al., 2023); mechanistic studies further suggest that EI is indirectly associated with lower aggression via higher positive affect, lower negative affect, and stronger emotion-regulation self-efficacy (Chamizo-Nieto et al., 2020; Mesurado et al., 2018; Gao et al., 2023). Classroom-based EI training has been reported to yield sustained decreases in verbal and physical aggression at follow-up (Castillo-Gualda et al., 2018), and in competitive contexts, EI attenuates the positive association between narcissism and aggression (Jang et al., 2025). Notably, most studies test only one segment—either sport/motivation to EI or EI to aggression—rather than estimating mediation within a single model that simultaneously includes sport motivation, EI, and aggression; moreover, differences between ability-based and trait EI measures and questions of directionality remain to be adjudicated in unified analyses. The present study tests a theory-driven indirect pathway, while acknowledging that temporal ordering cannot be established with cross-sectional data. Accordingly, we hypothesize H2: emotional intelligence accounts for an indirect association between sport motivation and aggressive behavior in the hypothesized model.

The strength model of self-control conceptualizes self-control (SC) as a trainable yet depletable “mental muscle,” suggesting that emotionally driven impulses may be more likely when regulatory resources are low (Englert, 2016). Within the revised General Aggression Model (GAM), SC is posited to function as a brake at the appraisal–decision stage; once affect is activated, insufficient inhibition may increase the likelihood of aggressive output (Allen et al., 2018). A social-learning–self-discipline perspective adds that sport contexts may provide opportunities to practice SC through rule reinforcement, delay of gratification, and immediate feedback, while acknowledging that in climates emphasizing retaliation or dominance, these capacities may be co-opted in service of instrumental or premeditated aggression (Chester, 2024). Framed around impulsive aggression, higher-quality sport motivation—more autonomous, internalized, and stable—may be linked to sustained participation and goal management, which may help build and maintain SC resources that, in turn, could interrupt the emotion-to-action transition toward aggression.

Evidence across levels of analysis is broadly consistent with this account. Longitudinal and chain models have reported that greater physical-activity engagement is prospectively associated with higher baseline self-control and slower growth in aggression (Park et al., 2017; Rohlf et al., 2018); in university and combat-sport samples, a “sport motivation/activity → self-efficacy → SC → aggression” pathway has been supported, with indirect effects accounting for nearly half of the total association (Yu et al., 2024; Xu A. et al., 2025; Xu T. et al., 2025; Chen et al., 2019). Additional evidence comes from interventions and meta-analytic work: a meta-analysis of martial-arts programs indicates small-to-moderate reductions in aggression among children and youth (Harwood-Gross et al., 2017), while randomized martial-arts training, multicomponent programs, and sports-based courses for juvenile offenders have been shown to improve self-control and reduce both proactive and reactive aggression (Fung and Lee, 2018; Lok et al., 2024). Two meta-analyses and two systematic reviews converge that programs embedding SC exercises and rule reinforcement yield the largest effects, with average impacts in the range of −0.25 to −0.41 on Hedges’ g or Cohen’s d; non-contact formats tend to perform best (Spruit et al., 2016; Castillo-Eito et al., 2020; van der Sluys et al., 2024; Yang et al., 2023). At the same time, context matters: generic smartphone apps or brief drills show inconsistent benefits for aggression in general populations (Kip et al., 2021; Beames et al., 2023). Moderator findings are accumulating—years or rank in martial arts relate to higher SC and lower aggression (de Lorenco-Lima et al., 2025; Zvi and Lavi, 2025); in rural youth, exercise links to lower aggression via a “psychological capital → SC” sequence (Xu A. et al., 2025; Xu T. et al., 2025b); and in correctional settings, the inclusion and intensity of self-control and emotion-management modules are positively associated with the effectiveness of sports programmes in reducing reoffending (Jugl et al., 2023). Notably, much of this literature tests only one segment—either “sport → SC” or “SC → aggression”—often relies on self-report, and focuses on martial-arts or child samples; studies that simultaneously include sport motivation, SC, and aggression with comparable standardized effects in general undergraduate populations remain scarce. The present study tests a theory-driven indirect pathway involving self-control, while acknowledging that temporal ordering cannot be established with cross-sectional data and that self-control may also precede sport motivation. Accordingly, we hypothesize H3: self-control accounts for an indirect association between sport motivation and aggressive behavior in the hypothesized model.

The process model of emotion regulation proposes that individuals with higher emotional intelligence (EI) may be better able to “cool down” earlier in the regulatory sequence—through situation selection, attentional deployment, and cognitive reappraisal—potentially reducing inhibitory load during response execution (Gross, 2015). The strength model of self-control (SC) emphasizes that control resources are limited yet trainable; effects of SC practice alone are modest but may be more reliable when coupled with emotion-regulation strategies (Friese et al., 2017). Within an SDT × GAM integration, higher-quality sport motivation may co-occur with positive affect and meaning while supplying rule-and-feedback cues, creating a context in which emotion–control coupling may operate (Boat and Cooper, 2019). Converging evidence from intervention and longitudinal designs is consistent with links between EI-related skills and inhibitory control: preschool programs combining EI with inhibition training have been reported to improve emotion recognition and impulse control and to reduce aggression (Romero-López et al., 2021); among collegiate athletes, EI training has been associated with fewer Go/No-Go commission errors alongside changes in prefrontal–parietal hemodynamics (Köroglu et al., 2025). A three-wave cross-lagged panel study reported that emotional intelligence prospectively predicted lower levels of later problematic mobile social media use, with no evidence for the reverse path (Peng et al., 2025). A systematic meta-analytic review has reported that emotional intelligence is negatively related to health-related risk behaviors (Sánchez-López et al., 2022). In educational settings, PE programs that enhance students’ emotional intelligence have been reported to reduce verbal and emotional aggression (Ortiz-Franco et al., 2024), and similar protective patterns of higher EI with lower aggression are observed in athlete cohorts (Acebes-Sánchez et al., 2021). Taken together, these findings are broadly consistent with an ordering in which EI-related emotion-processing skills may support self-control processes and, in turn, relate to lower aggressive responses across populations and tasks; however, reciprocal relations between EI and SC are also plausible.

Evidence from a systematic review and meta-analysis of 17 EI-training interventions among healthcare workers suggests post-intervention improvements in emotional intelligence, although effect sizes may be inflated by methodological limitations (Powell et al., 2024); In clinical and elite-sport samples, higher emotional intelligence and related self-regulatory resources tend to co-occur with less dysregulated or risky behavior (Biolcati et al., 2025; Stanković et al., 2022). Recent work shows that higher emotional intelligence is associated with lower impulsivity and sensitivity to reward, which in turn are associated with less health-risk taking (Megías-Robles et al., 2023). From a resource-protection angle, higher EI is associated with better self-regulation, which in turn relates to fewer problematic behaviors such as smartphone addiction and poorer wellbeing (Mascia et al., 2020). Among lower-grade college students, exercise has been reported to relate to lower aggression through a “self-control → resilience” sequence (Yue et al., 2024). Over the life course, childhood SC forecasts broad midlife advantages in health and social functioning (Richmond-Rakerd et al., 2021), while community violence studies identify emotion-regulation breakdowns as an important correlate of aggression (Heleniak et al., 2018). Although the EI–SC–aggression sequence has been replicated across diverse settings, fewer studies have tested an explicit serial mediation while simultaneously including sport motivation, EI, SC, and aggression within a single model. Because the present data are cross-sectional, the EI → SC ordering is treated as a theory-driven, model-implied specification rather than evidence of temporal precedence. Accordingly, we hypothesize H4: emotional intelligence and self-control account for a serial indirect association between sport motivation and aggressive behavior in the hypothesized model, such that higher sport motivation is associated with higher EI, which is associated with higher SC, which is associated with lower aggression.

The target population comprised full-time undergraduates at Shihezi University who were not majoring in sport or physical education. Using randomized classroom intercepts stratified by academic building and time slot, the field team visited classrooms during routine teaching hours from November 1 to 5, 2025. Immediately after class, students present were invited to scan a QR code and complete an online questionnaire. The survey platform time-stamped each submission and generated an anonymous ID; average completion time was approximately 6–10 min. In total, 550 questionnaires were returned; based on two a priori exclusion criteria—completion time under 6 min and substantial item nonresponse—65 cases were removed, yielding a valid analytic sample of N = 485 and an overall validity rate of 88.2%. The sample included 231 men (47.6%) and 254 women (52.4%), ages 18–25 years.

The study protocol and informed consent form were reviewed and approved by the Science and Technology Ethics Committee of the First Affiliated Hospital of Shihezi University (Ethics Review No.: KJ2025-520-01) and complied with the Declaration of Helsinki. All participants provided electronic informed consent before beginning the survey. Data were collected anonymously without personally identifiable information and were used solely for academic research and instructional improvement; no data were disclosed to third parties without authorization.

All analyses were conducted in SPSS 27.0 and AMOS 26; AMOS was used exclusively for confirmatory factor analyses (CFA) and evaluation of the measurement model fit. Following the scoring procedures described above (including reverse-scoring where applicable), we computed composite scores as mean item ratings for SM, EI, SC, and AB. Internal consistency was estimated with Cronbach’s alpha. Descriptive statistics included means, standard deviations, skewness, and kurtosis; bivariate associations were examined using two-tailed Pearson correlations with α = 0.05. Sex differences were tested with independent-samples t tests and accompanied by Cohen’s d. Age differences across four groups were evaluated with one-way ANOVA, reporting the F statistic and partial η², followed by Tukey’s post hoc comparisons. Potential common-method variance (CMV) was assessed using Harman’s single-factor test based on an unrotated principal components solution and by comparing the fit of a one-factor CFA model with the hypothesized multi-factor measurement model. Mediation was tested in PROCESS v4.2 using Model 6 (serial mediation; EI → SC) with percentile bootstrapping (5,000 resamples). Model 6 provides estimates of the total and direct effects as well as the specific indirect effects via EI, via SC, and via the serial EI → SC pathway, which were used to evaluate H2–H4. Sex and age were included as covariates in all mediation models. We report unstandardized coefficients (B) with 95% confidence intervals for the total and direct effects, and 95% percentile bootstrap confidence intervals for specific indirect effects; indirect effects were considered statistically significant when the 95% percentile bootstrap confidence interval did not include zero. Standardized coefficients (β) are reported for the regression equations. Sensitivity analyses (alternative operationalizations). Sensitivity analyses were conducted to examine whether inferences were sensitive to the multidimensionality of the key constructs (Figure 1). Specifically, we re-estimated the serial mediation model (PROCESS Model 6; EI → SC; 5,000 bootstrap resamples; covariates = sex and age) using (a) each SMS-II subscale as X as well as autonomous and controlled motivation indices (and amotivation additionally examined in its raw direction), (b) each EI branch (SEA, OEA, UOE, ROE) as M1 while retaining overall self-control as M2, and (c) each self-control facet (IC, HH, RT, WF, AE) as M2 while retaining overall EI as M1. Results are summarized in Supplementary Tables S2–S4.

All measures were collected in a single session from the same respondents, which may raise concerns about common-method variance (CMV). Procedurally, the survey was anonymous and emphasized voluntary participation and that there were no right or wrong answers, which may reduce evaluation apprehension. Statistically, we conducted Harman’s single-factor test using an unrotated principal components analysis. Four factors had eigenvalues greater than 1, and the largest factor accounted for 36.76% of the total variance, below the commonly cited 40% benchmark. We also compared a one-factor confirmatory factor analysis (CFA) model (all indicators loading on a single latent factor) with the hypothesized multi-factor measurement model; the one-factor model fit substantially worse than the multi-factor model. Together, these checks suggest that a single common factor is unlikely to fully explain the observed pattern of associations; however, residual method variance cannot be ruled out in a single-source, self-report, cross-sectional design and is therefore considered in the limitations. Fit indices for the one-factor versus multi-factor CFA comparison are provided in Supplementary Table S1. Harman’s single-factor test is an insensitive diagnostic and does not rule out residual method variance.

Table 1 summarizes the sample (N = 485): Sport Motivation (SM) had M = 3.45, SD = 0.91; Emotional Intelligence (EI) M = 3.76, SD = 0.85; Self-Control (SC) M = 3.72, SD = 0.77; and Aggressive Behavior (AB) M = 2.14, SD = 0.80. Skewness ranged from −0.714 to 0.580 and kurtosis from −0.549 to 0.135 (absolute values < 1), indicating approximate normality and suitability for parametric analyses. Two-tailed Pearson correlations (α = 0.05) showed that SM correlated positively with EI (r = 0.451, p < 0.001) and SC (r = 0.313, p < 0.001), and negatively with AB (r = −0.349, p < 0.001). EI and SC were positively related (r = 0.462, p < 0.001), and both were inversely related to AB (EI–AB: r = −0.440; SC–AB: r = −0.518; both p < 0.001), with the SC–AB link largest in absolute magnitude. The maximum absolute correlation was |r| = 0.518, suggesting no collinearity concerns for subsequent regression-based models. Overall, higher SM was associated with higher EI and SC and with lower AB; the negative SM–AB association is consistent with H1 and provides descriptive context for the subsequent regression-based mediation analyses.

Table 2 presents independent-samples tests by gender. Female students scored higher than male students on Sport Motivation (SM), Emotional Intelligence (EI), and Self-Control (SC), whereas males scored higher on Aggressive Behavior (AB). Specifically: SM (men 3.33 ± 0.91; women 3.57 ± 0.90), t(483) = −2.924, p = 0.004, Cohen’s d = −0.266 [−0.445, −0.087]; EI (men 3.55 ± 0.89; women 3.95 ± 0.76), variance homogeneity was violated and Welch’s correction was applied, t(453) = −5.288, p < 0.001, d = −0.485 [−0.665, −0.304]; SC (men 3.50 ± 0.75; women 3.92 ± 0.73), t(483) = −6.154, p < 0.001, d = −0.560 [−0.741, −0.378]; AB (men 2.31 ± 0.81; women 1.99 ± 0.76), t(483) = 4.569, p < 0.001, d = 0.415 [0.235, 0.595]. In this sample, women reported higher SM, EI, and SC and lower AB than men, indicating that sex was associated with key study variables; therefore, sex was included as a covariate in the mediation analyses.

Table 3 reports one-way ANOVAs across age groups (18 years n = 121; 19 years n = 109; 20 years n = 111; 21–25 years n = 144), all of which were significant: SM, F(3, 481) = 6.750, p < 0.001, ηp² = 0.040; EI, F(3, 481) = 6.269, p < 0.001, ηp² = 0.038; SC, F(3, 481) = 4.821, p = 0.003, ηp² = 0.029; AB, F(3, 481) = 6.647, p < 0.001, ηp² = 0.040. Group means, respectively, were SM (3.27, 3.30, 3.49, 3.71), EI (3.56, 3.68, 3.77, 3.99), SC (3.54, 3.65, 3.79, 3.87), and AB (2.35, 2.15, 2.18, 1.93). Tukey post-hoc tests showed that for SM and EI the 21–25 group scored higher than the 18–19 groups, with the 20-year group typically intermediate; For SC, the 21–25 group scored significantly higher than the 18-year group; the 20-year group was descriptively higher than the younger groups, but these differences did not reach statistical significance; For AB, the 21–25 group showed the lowest mean aggression; Tukey tests indicated that this group scored significantly lower than the 18-year group, whereas differences relative to the 19- and 20-year groups were in the same direction but did not reach statistical significance. Taken together, older students reported higher SM, EI, and SC and lower AB, which may reflect developmental and contextual differences across undergraduate years. Given these age-related differences, age was included as a covariate in the mediation analyses, and future work could test whether academic year or developmental factors moderate the proposed pathways.

Confirmatory factor analyses in AMOS 26 for the four core constructs indicated acceptable measurement fit, providing a basis for subsequent path testing (χ²/df = 1.368, RMSEA = 0.028, CFI = 0.972, TLI = 0.971). We then tested the serial mediation using PROCESS Model 6 with 5,000 percentile-bootstrap resamples while controlling for sex and age; the regression equations are summarized in Table 4 (standardized coefficients shown in Figure 2), and the total/direct/indirect effects are reported in Table 5. Without mediators, Sport Motivation (SM) was negatively associated with Aggressive Behavior (AB), total effect B = −0.267, 95% CI [−0.342, −0.193] (β = −0.303). With Emotional Intelligence (EI) and Self-Control (SC) entered simultaneously, the direct effect remained significant and negative, B = −0.115, 95% CI [−0.188, −0.042] (β = −0.131), indicating partial mediation. In the regression equations (Table 4 and Figure 2), SM was positively associated with EI (B = 0.378, SE = 0.038, β = 0.406, t = 9.984, p < 0.001) and with SC (B = 0.098, SE = 0.038, β = 0.116, t = 2.614, p = 0.009), and EI was positively associated with SC (B = 0.323, SE = 0.041, β = 0.355, t = 7.836, p < 0.001). With AB as the outcome (Table 4 and Figure 2), the negative association for SC was largest in magnitude (B = −0.383, SE = 0.045, β = −0.368, t = −8.536, p < 0.001), followed by EI (B = −0.179, SE = 0.043, β = −0.189, t = −4.163, p < 0.001), and SM remained negatively associated with AB (B = −0.115, SE = 0.037, β = −0.131, t = −3.098, p = 0.002). All three indirect effects were significant (Table 5): via EI, B = −0.068, 95% CI [−0.109, −0.030]; via SC, B = −0.038, 95% CI [−0.070, −0.008]; and through the serial EI → SC path, B = −0.047, 95% CI [−0.071, −0.028]. Combined, indirect effects summed to B = −0.152, 95% CI [−0.199, −0.106], accounting for approximately 56.87% of the total association. Overall, EI and SC each accounted for significant indirect associations between SM and AB, and the serial indirect effect in the specified order (EI → SC) was also significant; thus, the results were consistent with the independent and serial mediation hypotheses (H2, H3, and H4) in this sample.

Sensitivity analyses are summarized in Supplementary Tables S2–S4. The serial indirect effect remained statistically significant across alternative operationalizations of sport motivation, including each SMS-II regulation and the autonomous/controlled motivation indices (serial indirect effects ranged from B = −0.042 to −0.047, bootstrap 95% CIs excluding zero). When amotivation was examined in its raw direction, the total and indirect effects reversed direction (positive), consistent with its conceptual meaning as reduced motivation. Replacing the overall EI composite with each WLEIS branch (SEA, OEA, UOE, ROE) yielded comparable and significant serial indirect effects (B range = −0.037 to −0.043). Likewise, replacing the overall self-control composite with each SCS-C19 facet (IC, HH, RT, WF, AE) yielded significant serial indirect effects (B range = −0.034 to −0.042). Collectively, these checks indicate that the main inference is not dependent on a single operationalization of sport motivation, emotional intelligence, or self-control.

Consistent with H1, sport motivation was inversely associated with aggressive behavior in this undergraduate sample. This pattern aligns with evidence that structured sport and activity programs are, on average, linked to lower aggression, while effects remain contingent on context and implementation (Yang et al., 2023; Castillo-Eito et al., 2020; Ouyang and Liu, 2023). From an SDT perspective, stronger internalization and value congruence can promote rule adherence and prosocial orientations. Within the General Aggression Model, normative cues and regulated competition may reduce hostile appraisals and facilitate inhibitory responding. In sport settings, autonomy-supportive or empowering climates have similarly been associated with lower aggressive or antisocial conduct, which is compatible with a motivation-based explanation (Krommidas et al., 2022). Longer-term curricula—including team-based and martial-arts–based programs—also report decreases in aggression-related outcomes, suggesting that repeated exposure to clear rules and self-regulatory demands may be one pathway through which motivation relates to lower aggressive tendencies (Trajković et al., 2020; Shachar et al., 2016; Harwood-Gross et al., 2017).

At the same time, the motivation–aggression association likely depends on how sport is experienced (e.g., activity type, level of contact, supervision, and norm climate). Field and laboratory work indicates that non-contact or rhythmic tasks may dampen anger responses relative to more confrontational formats (Pels and Kleinert, 2016). Reviews also emphasize that well-structured supervision and explicit rule reinforcement are key ingredients associated with fewer conflicts and lower aggression (Spruit et al., 2016). Related comparisons highlight discipline and routinized norms as potentially protective features of organized activity contexts (Pačesová and Šmela, 2020; Basiaga-Pasternak et al., 2020). Because sport type, competitive level, and participation dose were not assessed here, future studies should test these contextual factors as moderators to clarify when higher motivation most strongly relates to lower aggression and when this link may weaken. In campus practice, universities may prioritize supervised, rules-based sport opportunities that emphasize autonomy support and cooperative norms, while applying additional oversight to high-contact formats to avoid normalizing aggressive conduct.

Consistent with H2, emotional intelligence accounted for part of the association between sport motivation and aggression. One interpretation is that more self-endorsed engagement in sport is accompanied by more frequent practice of emotion appraisal, interpersonal feedback processing, and regulation strategies in structured contexts, which may be reflected in higher EI. This aligns with evidence that sport/physical-activity engagement and motivation are positively related to EI, and that EI-oriented programs in school or PE settings can reduce aggressive or violent behaviors (Ubago-Jiménez et al., 2021; Ortiz-Franco et al., 2024). Intervention and longitudinal work further suggests that strengthening emotion skills—particularly emotional clarity and regulation—can translate into lower subsequent aggression, although effects depend on context and delivery (Garaigordobil and Peña-Sarrionandia, 2015; Castillo et al., 2013; García-Sancho et al., 2017). Mechanistically, the process model of emotion regulation implies that better emotion identification and earlier, more adaptive regulation (e.g., reappraisal) can reduce anger escalation and hostile responding, thereby lowering aggression propensity (Gross, 2015; Bibi et al., 2020; Newman et al., 2021).

Consistent with H3, self-control also functioned as an independent pathway linking sport motivation to lower aggression. Self-control theories posit that inhibitory capacity constrains impulsive reactions under provocation; when self-control resources are higher, aggressive impulses are more likely to be inhibited (Baumeister et al., 2007; DeWall et al., 2011; Sofia and Cruz, 2015; Chan and Chui, 2017). Meta-analytic evidence reports a robust negative association between self-control and aggression, including in university samples, supporting self-control as a proximal resource relevant to aggressive tendencies (Lei et al., 2020). At an applied level, reviews of sport- and activity-based programs indicate that interventions embedding explicit self-control practice and rule reinforcement are often associated with reductions in aggression, although effects vary by design and population (Ouyang and Liu, 2023; Yang et al., 2023; Denson et al., 2012; Trajković et al., 2020). Taken together, these two indirect pathways suggest that emotion-related skills and inhibitory control each plausibly explain part of the motivation–aggression association, while leaving room for additional mechanisms not included in the current model. For practice, physical education courses and student support services may embed brief emotion-skill training and self-control routines into regular sport sessions. Instructors can reinforce emotion labeling, cognitive reappraisal, and impulse inhibition through structured reflection and feedback during common peer-conflict scenarios.

Consistent with H4, the serial pathway suggests that sport motivation may relate to lower aggression through a coupled sequence of emotional and inhibitory regulation (EI → SC). In structured sport contexts, motivated engagement may provide repeated opportunities for emotion appraisal, social feedback, and regulation practice, which can be reflected in higher emotional intelligence and improved coping-relevant skills. This interpretation is consistent with prior work linking sport engagement or sport-related motivation to higher EI and related adaptive outcomes across samples and settings (Laborde et al., 2016; Ubago-Jiménez et al., 2019; Šukys et al., 2019; Vega et al., 2022). In addition, EI has been associated with self-control and broader self-regulatory functioning in chained associations, supporting the idea that emotion-processing skills may reduce the self-regulatory burden during emotionally charged moments (Gutiérrez-Cobo et al., 2018; Drndarević et al., 2021).

Downstream, self-control represents a more proximal constraint on aggressive impulses, and evidence indicates that higher self-control is linked to lower aggression and related maladaptive behaviors across populations (Li et al., 2017; Yue et al., 2024; Zvi and Lavi, 2025). Because the present data are cross-sectional and self-reported, the EI → SC ordering should be interpreted as a theory-informed specification rather than proof of temporal precedence; reciprocal or reverse processes remain plausible. Future longitudinal and intervention studies can directly test temporal ordering (e.g., whether changes in EI precede changes in self-control) and examine whether the strength of the serial pathway varies across broader student adjustment contexts and stress exposures (La Rosa et al., 2025). In applied settings, universities may pilot sport-based programs that intentionally sequence emotional skill development before repeated self-control practice. Semester-length pilots can monitor feasibility and simple behavioral indicators of conflict and rule violations to inform larger trials.

This study relied on cross-sectional, single-source self-report data and regression-based mediation models; accordingly, the findings should be interpreted as associational and model-based evidence rather than causal effects, even though sex and age were included as covariates in the mediation analyses. To reduce respondent burden and harmonize response formats, all instruments used five-point response options; although internal consistency and CFA fit were acceptable, the response-range modification may affect score variance and comparability with versions using the original response formats. Because all measures were collected at a single time point from the same respondents, common-method variance remains a possibility; Harman’s single-factor test and the one-factor CFA comparison did not indicate a dominant common factor, but residual method bias cannot be ruled out. The sample comprised non–sport-major undergraduates from a single university, limiting generalizability across institutions, regions, and sport backgrounds. Mediation estimates were based on observed composite scores without explicit correction for measurement error and without modeling full multidimensional structures as latent variables. Importantly, sensitivity analyses indicated that the key serial indirect effect was stable across theoretically informed alternative operationalizations of sport motivation (SMS-II regulations and autonomous/controlled indices), EI branches, and self-control facets (Supplementary Tables S2–S4), although these checks do not substitute for latent-variable modeling that explicitly accounts for measurement error. Finally, we did not measure sport participation dose, sport type, competitive level, or contextual features such as supervision and motivational climate, which limits interpretation of boundary conditions.

Future research should strengthen causal inference and clarify directionality using multi-wave longitudinal designs, cross-lagged approaches, and intervention studies that can test whether changes in emotional intelligence precede changes in self-control and subsequent aggression. Measurement can be improved by incorporating multi-informant reports, behavioral tasks indexing emotion processing and inhibitory control, and objective indicators of sport participation alongside records of conflicts or rule violations. Analytically, latent-variable SEM can test whether specific motivational regulations, EI branches, or self-control facets drive the indirect pathways while explicitly accounting for measurement error, and competing specifications such as SC preceding EI can be compared directly. Studies should also examine moderation by sport type, contact level, supervision quality, and competitive context to identify when sport motivation is most strongly linked to lower aggression. Translationally, universities can pilot autonomy-supportive sport environments that embed brief emotion-skill and self-control routines, and evaluate feasibility and maintenance across student subgroups.