This study is a cross-sectional survey based on self-administered questionnaires, with medical college students as research subjects. The research process complies with relevant medical ethics standards.

The minimum sample size required for this study was calculated using the stratified random sampling formula. Based on relevant literature, the expected standard deviation was 2,114.8 METs (Yang and Liu, 2022), with α set at 0.05 (=1.96) and the allowable error specified as 70 METs. The proportion of cases in each stratum relative to the total population was ω ₁ = ω₂ = ω₃ = 0.33. The calculated minimum sample size was n ≈ 1,276. Considering a 20% invalid questionnaire rate, the total required sample size was determined to be 1,531 cases. Calculation formula: n = ∑ W i 2 S i 2 / ω i V + ∑ W i S i 2 / N , where n is the required minimum sample size, 𝑁_𝑖 is the number of units in each stratum, 𝑆_𝑖² is the variance of the i-th stratum, V is the variance for estimating the population mean, V = (δ/u_α/2)², and W_i = N_i/N.

A questionnaire survey was conducted among college students in Jiangsu Province using stratified random sampling. Survey regions were first randomly selected, and Nanjing (Southern Jiangsu), Nantong (Central Jiangsu), and Lianyungang (Northern Jiangsu) were drawn as representatives of the three regions. Next, one university was randomly selected in each city, with its enrolled students as the survey objects. Finally, questionnaires were distributed randomly via Wenjuanxing (an online survey platform), and valid questionnaires were verified based on completeness and rationality of responses. A total of 1,561 questionnaires were distributed, yielding 1,409 valid responses, with a validity rate of 90.26%. After the recovery of valid questionnaires, the data were imported into SPSS 27.0 software for logical verification and data cleaning. Missing value analysis revealed that a small number of missing entries existed in some variables, with the overall data missing rate ≤ 5%, which meets the acceptable range for statistical analysis. In accordance with the data processing norms for quantitative research, listwise deletion was adopted in this study to ensure the reliability and accuracy of the subsequent statistical analysis results.

Participants in this study ranged in age from 15 to 38 years, with a mean age of (20.71 ± 1.918) years, including 453 males (32.15%) and 956 females (67.85%). In terms of grade distribution, freshmen (53.94%) account for the highest proportion. Myopic individuals constituted 84.88%, with non-myopic at 15.12%. Problematic daily mobile phone screen time was prevalent at 86.66%, while normal was only 0.57% and excessive 12.77%. For sleep duration (h), 386 (27.39%) had insufficient sleep (Table 1).

Data were logically checked, described, and statistically analyzed using SPSS 27.0 software, with p < 0.05 defined as statistically significant. Spearman’s correlation analysis was employed to examine the correlations among kinesiophobia level, exercise self-efficacy, physical activity motivation, and physical activity level. Mediation and moderation effect analyses were conducted using the SPSS macro program PROCESS 3.2, and Models 4 and 5 were selected to test the hypotheses.

Normality tests were conducted on all variables, with no variables conforming to a normal distribution. Spearman’s correlation analysis was performed for the four key variables (Table 2).

Physical activity motivation was positively correlated with exercise self-efficacy (r = 0.240, p < 0.01) and physical activity levels (r = 0.201, p < 0.01), while negatively correlated with kinesiophobia (r = −0.283, p < 0.01). Exercise self-efficacy demonstrated a positive correlation with physical activity (r = 0.135, p < 0.01) and a negative correlation with kinesiophobia (r = −0.113, p < 0.01). Kinesiophobia was negatively correlated with physical activity (r = −0.293, p < 0.01). These correlations clarify the interrelationships among the variables, supporting further moderated mediation analysis.

In all effect analyses, gender, grade, vision status, BMI, daily mobile phone screen time, and sleep duration were controlled as covariates. PROCESS macro Model 4 was utilized to examine whether exercise self-efficacy mediated the relationship between physical activity motivation and physical activity. After adjusting for covariates, results (Tables 3 and 4) indicated that physical activity motivation had a significant total effect on physical activity (B = 0.205, p < 0.001), with a persistent significant positive direct effect after controlling for exercise self-efficacy (B = 0.183, p < 0.001). The 95% confidence intervals for both the direct effect of physical activity motivation [0.132, 0.235] and its indirect effect via exercise self-efficacy [0.008, 0.038] excluded zero (Table 4), confirming the statistical significance of both pathways. The direct pathway accounted for 89.268% of the total effect, while the mediated pathway contributed 10.732% (total effect: B = 0.205).

Subsequently, PROCESS macro Model 5 was employed to test the moderated mediation model with covariates controlled. Results (Table 5) showed that after incorporating kinesiophobia into the model, the interaction term (physical activity motivation × kinesiophobia) exerted a significant moderating effect on physical activity (B = −0.052, t = −2.645, p < 0.05), indicating kinesiophobia as a moderator in the relationship between physical activity motivation and physical activity.

Simple slope analysis (Figure 1) revealed an inverse association between kinesiophobia levels and the predictive effect of physical activity motivation on physical activity. Specific effects of kinesiophobia at different intensity levels (M-1SD, M, M + 1SD) are presented in Table 6: the positive predictive effect of physical activity motivation on physical activity was most pronounced under low kinesiophobia (B = 0.182, t = 0.032), and remained significant but weakened under high kinesiophobia (B = 0.078, t = 0.033). Rising kinesiophobia levels gradually attenuated the predictive power of physical activity motivation on physical activity. The complete moderated mediation model is illustrated in Figure 2.

Self-Determination Theory posits that the development of individuals’ motivation to engage in physical activity follows a continuous spectrum, progressing from amotivation to external motivation, and ultimately to internal motivation (Deci and Ryan, 2000). The findings of this study indicated that physical activity motivation positively predicts physical activity levels. Individuals with stronger physical activity motivation tend to proactively seek more exercise-related knowledge and skill training, and actively participate in various sports challenges, thereby enhancing their physical activity levels through practice. Additionally, relevant studies have shown that exercise motivation is correlated with the reinforcement and sustainability of exercise behavior (Flack et al., 2021). Individuals with high motivation are more likely to overcome difficulties and setbacks during exercise and maintain long-term participation in physical activity, with such individuals typically exhibiting higher physical activity levels. Therefore, educational institutions can start by stimulating and strengthening college students’ physical activity motivation—for instance, by promoting the benefits of exercise, offering diverse sports program options, and establishing reasonable exercise goals and reward mechanisms—to inspire students’ intrinsic motivation. Meanwhile, physical education teachers can guide students to connect physical activity with their personal needs, thereby strengthening their physical activity motivation and facilitating the transformation of exercise motivation into actual exercise behavior (Courneya et al., 2012).

The findings of this study demonstrate that physical activity motivation among college students is not only directly associated with their physical activity levels but also indirectly linked to them through exercise self-efficacy, with the mediating role of exercise self-efficacy accounting for 10.732% of the total predictive effect. This result aligns with Self-Efficacy Theory (Bandura, 1977), which posits that an individual’s subjective assessment of their ability to perform a specific behavior is associated with goal attainment. It also helps clarify the specific relationship between physical activity motivation and college students’ physical activity levels. From the perspective of Self-Determination Theory (Miller et al., 1988), when an individual’s needs for autonomy, competence, and relatedness are fulfilled during physical activity, they are more inclined to develop intrinsic motivation, which is associated with greater engagement in physical activity (Lewis et al., 2020). Individuals with stronger physical activity motivation—whether motivated by intrinsic interest or external goals—are more likely to actively participate in exercise (Zhao et al., 2023). In this process, with continuous attempts and progress, their confidence in their own exercise capabilities tends to strengthen; that is, their exercise self-efficacy improves (Robinson et al., 2019). As an individual’s judgment of their own exercise capabilities, exercise self-efficacy functions as a mediator: it is linked to greater enthusiasm and persistence in physical activity participation, which in turn relates to higher physical activity levels. Relevant studies have shown that the physical exercise atmosphere significantly predicts exercise self-efficacy (Zhao et al., 2023); therefore, colleges and universities should foster a positive physical exercise atmosphere, actively offer engaging and personalized emerging sports programs, and encourage college students to experience the pleasure of exercise during physical activities, thereby supporting improvements in exercise self-efficacy.

Numerous studies consistently indicate that self-efficacy and kinesiophobia levels exhibit an inverse relationship: the higher the kinesiophobia level, the lower the self-efficacy (John et al., 2023; Pina et al., 2025). According to the “Fear-Avoidance Model” (Feldberg et al., 2023), when the body experiences fear of movement due to pain, corresponding cognitive and behavioral correlates may emerge; under conditions of high kinesiophobia, the positive association between individuals’ self-efficacy and their behavior is weaker (Zhang et al., 2023). In this context, kinesiophobia levels may be directly associated with physical activity levels rather than functioning as a moderator (Balicka-Bom et al., 2021).

The findings of this study show that kinesiophobia level moderates the direct association in the mediating process through which physical activity motivation relates to physical activity levels via exercise self-efficacy. Moreover, under low kinesiophobia levels, the positive predictive effect of physical activity motivation on physical activity levels is more pronounced, indicating that high kinesiophobia levels weaken the positive association between physical activity motivation and physical activity levels. Specifically, kinesiophobia level negatively moderates the positive relationship between physical activity motivation and physical activity levels. Individuals with high physical activity motivation, under low kinesiophobia levels, are more likely to translate their motivation into actual physical activity behaviors, actively participate in various sports activities, and maintain high physical activity levels. Anxiety and depression are currently major public concerns in the context of mental health, with significant implications for population health (Guan et al., 2025; Formolo et al., 2023; Xu et al., 2025; Miao et al., 2025). Notably, individuals with high kinesiophobia levels may be more prone to negative emotions such as anxiety and depression (Pells et al., 2007), and these emotions are related to reduced motivation and confidence to engage in physical activity. Physiologically, high kinesiophobia is linked to the body’s stress responses (Ramprasad et al., 2011), such as muscle tension and increased heart rate; these physiological correlates may enhance individuals’ discomfort during exercise, which is associated with lower engagement in physical activity. Additionally, individuals with high kinesiophobia levels are more prone to associating movement with negative outcomes—when facing physical activity, their attention is more focused on potential risks (Núñez-Cortés et al., 2023) rather than the positive effects of the activity itself. This cognitive bias is associated with reduced translation of physical activity motivation into actual action (Hammer et al., 2020). In contrast, individuals with low kinesiophobia levels can view exercise more rationally and focus more on the benefits of exercise (Sarac et al., 2024); thus, driven by high physical activity motivation, they are more likely to actively engage in physical activity. Therefore, during college students’ participation in sports, providing timely positive feedback and encouragement—such as the application of mindfulness-based cognitive therapy and confidence intervention (Guetterman et al., 2015)—may alleviate negative emotions, which is associated with reduced kinesiophobia levels, strengthened physical activity motivation, and sustained maintenance of physical activity levels.

This study adopted a cross-sectional design, which only reveals the associative characteristics between variables rather than inferring their causal order and direction of action. The sample has limitations in representativeness due to its single source: the participants were mainly selected from medical colleges and universities in Jiangsu Province, failing to cover college students from different geographical regions nationwide, various types of universities, and non-medical majors. The research data were mainly based on participants’ self-reports. Affected by subjective cognition, recall accuracy, and social desirability bias, there may be recall and reporting bias. Future studies could expand the geographical coverage of samples, increase the sample size, and integrate objective measurement tools to improve data objectivity, thereby enhancing the generalizability of the research conclusions.