Body image and loneliness as mediators of the relationship between physical activity and exercise self-efficacy in college students

Abstract

The purpose of this study is to investigate the relationship between physical activity and exercise self-efficacy among college students and to examine the mediating effects of loneliness and body image. Through an analysis of parallel mediating effects, the intrinsic mechanisms underlying these variables are revealed and provide theoretical support for promoting healthy behaviours among college students. The study, conducted in April 2025, employed a cross-sectional design to survey college students across northeast, North, central, East, and South China. A total of 1102 valid questionnaires were collected. The level of physical activity was assessed using the Physical Activity Rating Scale (PARS-3). Exercise self-efficacy was measured using the Exercise Self-Efficacy Scale. Body image was assessed using the Physical Self-Esteem Scale (PSPP), and loneliness was calculated using the UCLA Loneliness Inventory (3rd edition). The data were analysed using SPSS 27 software for descriptive statistics, correlation analyses, and mediation effect analyses. A significant direct positive correlation was found between physical activity and exercise self-efficacy (β = 0.2313, p < 0.001). Furthermore, loneliness and body image played important mediating roles in the relationship between physical activity and exercise self-efficacy. Specifically, physical activity indirectly improved exercise self-efficacy by reducing loneliness (β=-0.124, p < 0.001) and enhancing body image (β = 0.383, p < 0.001). The mediating effect accounted for 32.1% of the total impact, and the mediating effect of body image (28.5%) was significantly more potent than that of loneliness (3.5%). This study reveals the mechanism by which physical activity indirectly increases college students’ sense of exercise self-efficacy by reducing loneliness and improving body image. This finding shows the importance of focusing on individual psychological states when promoting physical activity. Interventions should promote both positive body image and a favourable social environment to improve college students’ exercise self-efficacy and overall psychological well-being.

Introduction

Numerous studies have examined the relationship between physical activity and health. Because regular physical activity promotes good health, the World Health Organisation (WHO) physical activity recommendations suggest that adults should engage in at least 150 min of moderate-intensity aerobic exercise, 75 min of vigorous-intensity aerobic exercise, or an equivalent combination of the two weekly to significantly reduce the risk of chronic diseases and promote mental health¹. However, according to previous studies, most people do not engage in sufficient physical activity to obtain these physical and psychological health benefits², especially in the college population. According to Hallal’s research, data from 122 countries for adults (≥ 15 years) and 105 countries for adolescents (13–15 years) were synthesised to map global physical-activity levels. The findings reveal that 31.1% (95% CI 30.9–31.2) of adults are physically inactive, with pronounced regional variation: Southeast Asia reports the lowest prevalence at 17.0% (16.8–17.2), whereas the Americas and the Eastern Mediterranean reach approximately 43%³. These data indicate that physical inactivity has become a familiar and persistent problem. College students are at a critical stage of their physical and mental development, and continued physical inactivity affects their physical health, mental health, and social adaptability.

Perceived exercise self-efficacy is closely related to physical activity. Exercise self-efficacy refers to individuals’ belief in their ability to engage successfully in physical exercise. It is considered one of the critical factors that influence people’s initiation and maintenance of physical exercise behaviour⁴. Multiple studies have shown that individuals’ exercise self-efficacy has a decisive effect on promoting physical activity, and empirical research has confirmed a significant association between exercise self-efficacy and levels of physical activity. A previous study revealed that lower levels of physical activity are correlated with lower expectations of exercise self-efficacy⁷.

Social cognitive theory (SCT) posits that self-efficacy is a crucial psychological mechanism that motivates individuals’ engagement in physical activity (PA)⁶. Self-efficacy is defined as individuals’ beliefs about their ability to perform physical tasks in various environments⁸. Physical tasks and their level of difficulty directly influence behavioural decision-making, such that individuals with high self-efficacy tend to perceive movement barriers as surmountable challenges and maintain physical activity by adapting their effort strategies⁹. Moreover, ESE levels can be significantly elevated by interventions (e.g., goal setting, positive feedback), which in turn improve PA adherence⁵. Notably, according to previous research, teachers with high exercise self-efficacy report more positive emotional experiences, greater enjoyment, lower perceived fatigue, and more pleasure in exercise. They are more likely to persist in physical activities because they believe that they can overcome obstacles and difficulties and achieve success. Regular physical activity can directly improve individuals’ assessment of their exercise ability and self-confidence through improvements in physical fitness (such as increased muscle strength) and the optimization of health indicators (such as weight loss), thereby increasing exercise self-efficacy¹⁰. This finding has been replicated in a study of healthy female adolescents, where a significant relationship was found between physical activity and self-efficacy; moreover, positive exercise experiences were shown to enhance an individual’s exercise self-efficacy¹¹. Therefore, there is a significant bidirectional dynamic interaction between exercise self-efficacy and physical activity, and this bidirectional facilitation ultimately forms a positive feedback loop of “efficacy enhancement-exercise behaviour reinforcement-physiological and psychological gains”.

Additionally, there are controversies regarding the path of exercise self-efficacy (ESE) and the selection and validation of mediator variables, which warrant further investigation. Some studies have explored ways to increase exercise self-efficacy and promote physical activity. Research has shown that individuals’ exercise self-efficacy can be improved by modifying exercise modalities and providing positive feedback and support, which can increase physical activity levels⁵. These findings suggest that exercise self-efficacy affects physical activity both directly and indirectly through other psychological factors.

As social modernization accelerates, the college student population is characterized by increased independence and decreased social frequency. Studies have shown that social-detachment-induced loneliness has become a key predictor of depression risk and has a significant negative association with physical activity levels^12,15,16. Loneliness is the subjective experience of an unpleasant emotional state that arises when an individual perceives a deficiency in social relationships or social connections¹⁷. Individuals with high levels of loneliness often experience weak social support networks and a lack of resources for exercising, which leads to reduced willingness to exercise and, consequently, a lower frequency of physical activity¹³. Research suggests that group exercise, individualized counselling, education about physical activity, and social support during physical activity (as opposed to other social activities) are practical intervention components that can improve loneliness through physical activity. Another study showed that loneliness reduces physical activity levels in widowed individuals¹⁴. Loneliness and physical activity exhibit a dynamic bidirectional effect: chronically low levels of physical activity lead to reduced physical fitness and weaken individuals’ confidence in their athletic ability, which in turn may lead to reduced exercise self-efficacy. Therefore, the relationship between loneliness, physical activity, and exercise self-efficacy remains to be explored.

Body image refers to individuals’ perceptions, emotions, and evaluations of their own body and is influenced by a variety of factors, including psychological, social, and cultural elements. It encompasses not only physical appearance but also the awareness of bodily functions, sensations, and health¹⁸. As the prevalence of obesity continues to increase globally^20,21, individuals’ dissatisfaction with their body image has significantly increased, which in turn has induced a wide range of body image dysfunctions (including body image distortions and anxiety). Research suggests that regular physical activity may improve body image through multiple pathways. Regular physical activity significantly increases individuals’ satisfaction with their physical appearance through functional gains such as improved posture and muscle strength¹⁰. At the same time, a positive body image encourages individuals to participate in sports activities more frequently, thereby forming a virtuous cycle of body image improvement²². As such, body image perception also influences an individual’s exercise self-efficacy—those with negative body image perceptions tend to have lower levels of physical activity participation, which in turn further weakens their exercise self-efficacy.

In contrast, individuals who are satisfied with their body image are typically more confident in their athletic abilities and may have greater exercise self-efficacy¹⁹. Research has shown that both body image and exercise self-efficacy have significant effects on exercise participation. Because the relationship between exercise self-efficacy and body image has been explored less frequently in previous research, the intrinsic link between them requires further investigation.

In summary, there are associations between physical activity and exercise self-efficacy, body image, and loneliness; however, the underlying mechanisms have not been fully clarified. Body image and loneliness mediate the association between physical activity and exercise self-efficacy, and their mechanisms of action require further exploration. In addition, the sociocultural context and demographic characteristics influence this process. Studies have focused primarily on specific regions and have paid less attention to the role of body image and loneliness. The present study aims to address this gap by investigating the relationship between physical activity and exercise self-efficacy as well as the mediating effects of loneliness and body image among college students in northeast, North, central, East, and South China.

In this study, we propose the following hypotheses. The null hypothesis, H0, assumes that there is no significant relationship between the physical activity of college students and their exercise self-efficacy. The alternative hypothesis H1 suggests that there is a substantial relationship between the two. Specifically, H1a postulates a positive association between physical activity levels and exercise self-efficacy, whereas H1b suggests a negative correlation., The alternative hypothesis H2 states that the physical activity of college students affects their exercise self-efficacy through a mediating process involving body image and loneliness. The sub-hypotheses are as follows: H2a: Body image mediates the relationship between physical activity and exercise self-efficacy; H2b: Loneliness plays a mediating role in the relationship between physical activity and exercise self-efficacy; H3: There are differences in the mediating effects of loneliness and body image on the impact of physical activity on exercise self-efficacy. By testing these hypotheses, we aim to explore the mechanisms through which physical activity influences the exercise self-efficacy of college students and to construct a relevant model (Fig. 1).

Fig. 1.

Conceptual mediation model path diagram. PA physical activity, ESE exercise self-efficacy, BI body image, UCLA loneliness.

Methods

Participants and process

A cross-sectional study of university students in northeast, North, central, East, and south China was conducted in April 2025 using a convenience sampling method. The study adhered to the Declaration of Helsinki’s ethical principles for medical research involving human subjects as well as all relevant laws, regulations, and ethical codes to ensure the morality and legality of the study. After identifying the topic and direction of the study, we carefully selected the applicable scales and edited them into Questionnaire Star software to facilitate online data collection. To ensure the quality of the survey, we provided professional training to all participants to familiarize them with the questionnaire’s content and the methodological process. After receiving consent and support from the schools, we distributed the questionnaire to students in specific classes through the online platform. The questionnaire covered several dimensions, including basic sociodemographic information, physical activity level, exercise self-efficacy, body image, and loneliness, to comprehensively assess students’ physical and mental health. Throughout the study, we strictly followed the STROBE checklist for cross-sectional studies to ensure the transparency of the research and the completeness of the report. These measures were designed to improve the reliability and validity of the study. A total of 1236 participants started the questionnaire, with 638 (51.6%) providing complete responses. These completed questionnaires were then processed according to the following steps: (1) samples with missing values in their responses were excluded. (2) removing responses with homogeneity and regularity, such as all responses to a question that were almost the same or that showed a simple increasing or decreasing sequence; and (3) removing uncharacteristically rapid completion. A total of 134 questionnaires were excluded, resulting in 1,102 valid questionnaires with a validity rate of 89.17%. Of these, 598 participants (54.2%) were male, 505 (45.8%) were female, and the mean age of the participants was 19.38 years (SD = 1.255).

Measures

Physical activity

We used the Physical Activity Rating Scale (PARS-3), which was revised by Prof. Liang’s team, to assess students’ physical activity levels. The PARS-3 uses a 5-point Likert scale, which is widely used in China, and considers several aspects, including the intensity, duration, and frequency of exercise, to measure students’ physical activity levels comprehensively. Total individual PA is calculated as PA = intensity (1–5) × duration (1–5) × frequency (0–4). Thus, overall PA was determined to be low (≤ 19), moderate (20–42), or high (≥ 43). In this study, the Cronbach’s α coefficient for the PARS-3 was 0.75²³.

Exercise self-efficacy

The Exercise Self-Efficacy Scale, developed by Bandura, was used to measure exercise self-efficacy. Shin et al. (2001) used this scale with Korean adult patients with chronic diseases, and a panel of experts assessed the scale as having good face validity, with a Cronbach’s α of 0.94 and a reliability coefficient of 0.77 at the 2-week retest. Tung et al. (2011) translated this scale into Chinese. After back-translation and expert panel assessment, good face validity was demonstrated, with a Cronbach’s α of 0.96. Pei et al. (2017) showed that, after back-translation and evaluation by an expert group, the scale had good surface validity, with a Cronbach’s α of 0.96. A reliability test conducted by Pei et al. revealed that Cronbach’s α was 0.962, indicating that the scale was stable and had good structural validity. This scale has also been used with maintenance haemodialysis patients, individuals with cancer, individuals with diabetes, and other stroke risk groups. In the present study, the Exercise Self-Efficacy Scale had a Cronbach’s α coefficient of 0.939. The scale comprises 18 items that measure an individual’s confidence in engaging in regular physical activity in various situations. A 10-point scale was used, ranging from 0 (very unconfident and completely unable to engage in physical activity) to 50 (fair confidence and moderate certainty that one can engage in physical activity) and 100 (high confidence and certainty that one can persist in physical activity). The exercise self-efficacy score of the study participants was equal to the mean score of 18 items ranging from 0 to 100, and higher scores on this questionnaire indicated higher levels of exercise self-efficacy^24,25.

Body image

We used the Chinese version of the Physical Self-Esteem Scale (PSPP) to assess students’ level of physical self-esteem²⁶. This scale employs a 4-point Likert scale for each item. It consists of one main scale (a sense of physical self-worth) and four subscales (motor skills, physical fitness, physical condition, and physical attractiveness) with a total of 25 questions (including 13 reverse-scored questions). The higher the final score is, the higher the level of physical self-esteem. In this study, the Cronbach’s α coefficient for this scale was 0.841.

Loneliness

We also used the UCLA Loneliness Scale (3rd ed., 1988), a self-assessment instrument designed to measure the degree to which an individual feels lonely. The scale was originally developed in 1978, designed by Russell in 1978, and subsequently revised into a third, more college-applicable version that is designed to assess individuals’ social connectedness and closeness to others. The UCLA-20 consists of 20 items for which respondents select the option that best suits their situation based on their affective state and social interactions on a scale from 1 (strongly disagree) to 4 (strongly agree). By measuring an individual’s experience of emotional and social aspects, the UCLA-20 can provide quantitative information about feelings of loneliness, which can contribute to mental health research and the development of interventions. The third edition was developed and validated for college students, nurses, teachers, and older adult populations with internal consistency alpha coefficients ranging from 0.89 to 0.94 and retest reliabilities ranging from 0.62 to 0.73²⁷. In this study, the Cronbach’s α coefficient for the scale was 0.691.

Statistical analysis

First, we screened and processed the collected data. We subsequently conducted descriptive statistical analysis using SPSS version 27 software to make initial generalizations and characterizations of the data. In this process, we paid special attention to the skewness and kurtosis indicators of the data to determine the normality of the data distribution. Independent-samples t tests were used to analyze sex differences, Cronbach’s α coefficients were used to assess the internal consistency of the scales, and Pearson’s correlation analyses were employed to explore the direct correlations between the variables. We then conducted mediation analyses using the PROCESS 4.2 plug-in in SPSS software to analyze the mediation effects. We diagnosed multicollinearity in the model through the variance inflation factor (VIF) and tolerance values. To ensure the robustness and accuracy of the analytical results, we employed the bootstrap method with 5,000 resampling iterations to construct 95% confidence intervals (CIs). The significance of the mediating effect was determined by whether the CI contained zero. If the CI did not contain zero, the corresponding effect was considered statistically significant. In constructing the model, we controlled for sex and age as variables. All variables were standardized before they were included in the mediation effect model.

Results

Descriptive statistics and correlation analysis

Table 1 shows the demographic variables of this study and the scores of sexes on physical activity (PA), exercise self-efficacy (ESE), loneliness (UCLA), and body image (PSPP). The total sample size was 1102 (54.2% males and 45.8% females), with a mean age of 19.38 years (SD = 1.2). Independent sample t tests revealed that males scored significantly lower than females did in terms of physical activity (PA: 12.30 vs. 23.83, p < 0.001) and loneliness (UCLA: 43.94 vs. 43.05, p < 0.001), and exercise self-efficacy (ESE: 98.01 vs. 112.84, p = 0.047) was also significantly lower than that of females; there was no significant sex difference in body image (PSPP: 66.98 vs. 69.94, p = 0.882). The skewness and kurtosis of each variable met Kline’s criteria for a normal distribution (skewness < 3, kurtosis < ± 10), indicating that the data can be used directly for subsequent analysis.

Table 1.

Descriptive statistics and sex differences for various variables.

Variable	Sample (n = 1102)	Male (n = 598)	Female (n = 504)	P value	Skew	Kurt
Age, M (SD)	19.38 (1.2)
Male, n%	54.2%
PA, M (SD)	18.56 (18.87)	12.30 (13.31)	23.83 (21.13)	P < 0.001	1.586	2.72
ESE, M (SD)	106.05 (32.01)	98.01 (28.29)	112.84 (33.38)	P = 0.047	− 0.038	0.856
UCLA, M (SD)	43.46 (9.61)	43.94 (8.89)	43.05 (10.16)	P < 0.001	− 0.006	0.164
PSPP, M (SD)	68.59 (14.74)	66.98 (14.40)	69.94 (14.89)	P = 0.882	0.352	− 0.378

Table 2 shows the results of the Pearson correlation analysis. PA was significantly positively correlated with height (r = 0.118, p < 0.01) and weight (r = 0.175, p < 0.01), as were ESE (r = 0.379, p < 0.01) and PSPP (r = 0.383, p < 0.01). PA was significantly negatively correlated with UCLA (r=-0.126, p < 0.01). ESE was significantly negatively correlated with UCLA (r=-0.213, p < 0.01) and significantly positively correlated with PSPP (r = 0.383, p < 0.01). ESE was significantly positively correlated with height (r = 0.150, p < 0.01) and weight (r = 0.101, p < 0.01). Additionally, PSPP was significantly negatively correlated with UCLA (r=-0.330, p < 0.01).

Table 2.

Pearson correlation analysis for study outcomes.

	Age	Sex	Height	Weight	PA	ESE	UCLA	PSPP
Age	1
Sex	0.061*	1
Height	0.011	0.364**	1
Weight	0.121**	0.353**	0.194**	1
PA	0.023	0.305**	0.118**	0.175**	1
ESE	0.010	0.231**	0.150**	0.101**	0.379**	1
UCLA	0.059	− 0.046	− 0.027	0.018	− 0.126**	− 0.213**	1
PSPP	0.004	0.100**	0.100**	− 0.016	0.383**	0.383**	− 0.330**	1

PA physical activity, UCLA loneliness, ESE exercise self-efficacy, PSPP body image.

**p < 0.01, *P < 0.05.

Mediation analysis

For parallel multiple mediation analysis, Model 6 from the PROCESS macro version 4.2, developed by Hayes, was employed. Physical activity level (PA) served as the independent variable, while loneliness (assessed using the UCLA Loneliness Scale) and body image (assessed using the Physical Self-Perception Profile, PSPP) were specified as the parallel mediator variables. Exercise self-efficacy (ESE) was designated as the dependent variable. To control for potential confounding factors influencing health-related behaviors and psychological states, gender and age were included as covariates in the model. A standard 95% confidence interval was established, and bootstrap resampling with 5,000 samples was performed to enhance the accuracy of the analysis²⁸.

The results of the regression analysis (Table 3) showed adjusted R² values of 0.019, 0.147, and 0.233 for the three models, respectively. The collinearity diagnostics indicated that the variance inflation factors (VIFs) for physical activity, loneliness, and body image were 1.172, 1.122, and 1.129, respectively, all of which were less than 3.

Table 3.

Regression estimates for the parallel mediation model.

Outcome variable	Model 1 UCLA			Mode 2 PSPP			Model 3 ESE
	β	SE	t	β	SE	t	β	SE	t
PA	− 0.124	0.016	− 3.953***	0.383	0.022	13.286***	0.231	0.050	7.735***
UCLA							− 0.095	0.094	− 3.414***
PSPP							0.249	0.065	8.295***
Sex	− 0.062	0.229	2.983*				0.130	1.786	4.704***
Age
R²	0.019			0.147			0.233
F	7.432			63.170			66.930

***p < 0.001, **p < 0.001, *p < 0.05.

Model 1 revealed that physical activity significantly and negatively predicted loneliness (β=-0.124, p < 0.001), whereas age was a significant positive predictor (β = 0.062, p = 0.037); gender showed no significant association with loneliness. The model explained 1.99% of the variance (R²=0.019, F = 7.432, p = 0.0001), indicating that its prediction of loneliness was statistically significant.

Model 2, regressing body image on the predictors, showed that physical activity significantly and positively predicted body image (β = 0.383, p < 0.001), whereas neither age nor gender exhibited a significant effect. The model accounted for 14.72% of the variance (R²=0.147, F = 63.17, p < 0.001).

In Model 3, regressing exercise self-efficacy on the predictors, physical activity positively predicted exercise self-efficacy (β = 0.231, p < 0.001), whereas loneliness exerted a negative effect (β=-0.095, p = 0.0007) and body image exerted a positive effect (β = 0.249, p < 0.001). Age did not significantly predict exercise self-efficacy, but gender showed a significant positive association (β = 0.130, p < 0.001). The model explained 23.39% of the variance (R²=0.233, F = 66.93, p < 0.001).

Table 4; Fig. 2 summarize the mediation results. The total effect of physical activity on exercise self-efficacy was β = 0.340 (95% CI = 0.480 to 0.674), of which 67.9% was direct (β = 0.231, 95% CI = 0.293 to 0.492). The combined indirect effects accounted for 32.1% of the total effect (β = 0.108, 95% CI = 0.082 to 0.136), supporting H2. Specifically, the pathway via loneliness yielded an effect of β = 0.011 (95% CI = 0.002 to 0.024), corresponding to 3.5% of the total effect, whereas the pathway via body image produced β = 0.097 (95% CI = 0.071 to 0.124), accounting for 28.5%, thereby confirming H2a and H2b. The contrast between mediators (C1) was β=-0.0851 (95% CI=-0.116 to-0.056), supporting H3. Thus, PA not only directly enhances exercise self-efficacy but also exerts significant indirect effects.

Table 4.

Effect sizes of the overall mediation effects of the association between PA (X) and ESE (Y).

Trials	β	SE	LLCI	ULCI	Proportion (%)
Aggregate effect	0.340	0.049	0.480	0.674	100%
Direct effect	0.231	0.050	0.293	0.492	67.9%
Intermediary effect	0.108	0.013	0.082	0.136	32.1%
PA-UCLA-ESE	0.011	0.005	0.002	0.024	3.5%
PA-PSPP-ESE	0.097	0.013	0.071	0.124	28.5%

PA physical activity, UCLA loneliness, ESE exercise self-efficacy, PSPP body image.

Fig. 2.

Influence of the parallel mediation model of body image and loneliness on physical activity and exercise self-efficacy.

Discussion

The present study examined the mediating effects of loneliness and body image on the relationship between physical activity and exercise self-efficacy through parallel mediation analysis. Results demonstrated a statistically significant direct positive association between physical activity and exercise self-efficacy, consistent with prior empirical evidence and thus supporting Hypothesis 1a (H1a). Furthermore, the analysis revealed significant dual mediation pathways through loneliness and body image, underscoring the critical intermediary roles of these variables. These findings substantiate Hypotheses 2 and 3 (H2, H3).

This study found a significant direct positive relationship between physical activity and exercise self-efficacy, indicating that individuals with higher physical activity levels exhibit stronger exercise self-efficacy. This finding is consistent with previous research²⁹. Through regular physical activity (e.g., completing a running programme or strength training), individuals accumulate “success stories” that directly increase their confidence in their physical abilities, which also increases exercise self-efficacy. Regular physical activity directly improves individuals’ fitness levels. When people experience improvements in cardiorespiratory fitness, muscle strength, and endurance through exercise, they have more confidence in their ability to sustain and complete a variety of physical activities, which increases their exercise self-efficacy⁵. Another study of patients with spinal cord injuries revealed that participation in a tele-exercise intervention significantly increased patients’ exercise self-efficacy³⁰. In addition, people with high exercise self-efficacy are predicted to be more physically active. This has been validated in a study of physical activity and self-efficacy in older adults, where self-efficacy in the early years (Year 1) significantly predicted physical activity levels in the later years (Year 5)³¹. This means that exercise self-efficacy and physical activity form a positive cycle in which individuals who are more physically active are more athletic. More athletic individuals tend to have greater exercise self-efficacy and are therefore more inclined to exercise and increase their physical activity. In this study, exercise self-efficacy was confirmed as an important mediating variable that linked exercise capacity and physical activity. Nonetheless, the relationship between exercise self-efficacy and physical activity is complex and variable, which makes it important to examine the connection between these two factors. Additionally, sex and age may influence the relationship between physical activity and exercise self-efficacy. Therefore, when examining the direct impact of physical activity on exercise self-efficacy, it is essential to acknowledge that this influence may not be immediately apparent due to the interference of many accompanying factors. Considering this, universities should consider the characteristics of different types of physical activities when adopting them as strategies to increase exercise self-efficacy. This will help to identify more appropriate starting points and methods for research, which will more effectively promote the improvement of exercise self-efficacy.

The study establishes a significant association between physical activity and exercise self-efficacy, while further revealing body image as a mediator influencing exercise self-efficacy. This finding highlights the importance of cultivating body awareness. In the overall mediation analysis, the mediating effect of body image was found to be particularly significant: body image was significantly and positively correlated with exercise self-efficacy. This finding is consistent with previous research. Research has shown that when individuals have a positive view of their bodies, they are more likely to believe in their ability to improve their physical state and achieve their health goals³². Bandura suggested that individuals’ beliefs about their ability to complete a task successfully (i.e., self-efficacy) influence their behavioural choices and level of persistence³³. A positive body image can increase self-confidence and self-esteem, which in turn can increase self-efficacy. When individuals are satisfied with their body size and fitness, they are more likely to participate in physical activities and believe that they can accomplish these tasks³⁴. At the same time, higher levels of exercise self-efficacy enable individuals to overcome the challenges of an exercise programme, thereby influencing their physical self-perceptions^35,36. In the present study, we found that the associations between body image and age or sex were not significant. This differs from the results of previous studies. Silva et al. (2018) noted that body image is negatively correlated with sex, with female participants exhibiting higher rates of eating disorders and greater concerns about body image³⁷. This may be related to the characteristics of the study sample. Our study population consisted of a group of college students with a relatively narrow age range and similar educational backgrounds and upbringings, which may have limited the manifestation of age and sex differences in body image. In addition, this study revealed the positive effects of physical activity on exercise self-efficacy, a process primarily realized through improved body image. This finding provides important theoretical support for the design and implementation of health promotion interventions. Encouraging individuals to participate in physical activity can improve physical function and health status directly and can improve exercise self-efficacy indirectly through enhanced body image, which can encourage individuals to participate in health behaviours more actively, forming a virtuous cycle^37,38.

The present study reveals the complex associations among physical activity, loneliness, and exercise self-efficacy and elucidates the mechanisms of action through path analysis of the mediation model. These findings suggest that loneliness not only negatively predicts exercise self-efficacy but also plays a significant mediating role in the relationship between physical activity and exercise self-efficacy. Specifically, physical activity can indirectly improve exercise self-efficacy by reducing loneliness. These findings suggest that loneliness plays a significant role in the relationship between physical activity and exercise self-efficacy. Its mediating effect indicates that increased physical activity not only directly improves physical functioning and health status but also significantly increases individuals’ exercise self-efficacy by reducing loneliness. This finding offers new insights into the mental health benefits of physical activity and highlights the importance of considering individuals’ psychological state when promoting physical activity.

Our study further demonstrates the interrelationship among physical activity, exercise self-efficacy, and loneliness. Although the mediating effect of loneliness in this investigation was relatively small yet statistically significant, extant research has established a significant negative association between physical activity and loneliness, indicating that higher levels of physical activity correspond to lower levels of loneliness^39,40. Additionally, some studies have shown that the relationship between loneliness and physical activity is moderated by affective valence (AV). One study reported that physical activity was more strongly associated with reduced loneliness for individuals with higher affective valence. In contrast, individuals with lower affective valence may experience a weaker or even negative relationship between physical activity and loneliness³⁹. This moderating effect may partially explain the small mediating impact of loneliness in the present study. However, despite the small effect size, the negative influence of loneliness remained, and this influence had a significant negative impact on exercise self-efficacy. Specifically, individuals with higher levels of loneliness tend to have fewer opportunities to engage in physical activity, often due to factors such as a lack of athletic peers or weaker socialization skills, which results in lower levels of physical activity. This lower level of physical activity further undermines their confidence in their athletic ability, thereby reducing exercise self-efficacy. Conversely, individuals with lower levels of exercise self-efficacy are more likely to be less physically active, creating a vicious cycle in which loneliness leads to lower levels of physical activity, which in turn lowers exercise self-efficacy; low levels of exercise self-efficacy inhibit participation in physical activity, ultimately exacerbating loneliness. In addition, the present study reveals that loneliness is related to age, which is like previous research that suggests a nonlinear relationship between loneliness and age, such as a U-shaped relationship whereby adolescents and older adults are more likely to feel lonely. Middle-aged adults (aged 30–60 years) are generally the least lonely⁴¹. In summary, this study highlights the complex role of loneliness in the relationship between physical activity and exercise self-efficacy. Although previous research has recognized the impact of loneliness on sports participation, its mediating role between physical activity and exercise self-efficacy has not been explored. The findings of this study suggest that when interventions are designed, it is important to focus not only on physical activity but also on individuals’ emotional states and social environments to break this vicious cycle and improve overall physical and mental health.

In this study, loneliness and body image served as parallel mediators between physical activity and exercise self-efficacy. The mediating effect of body image was more pronounced, which aligns with H3. In contrast, the impact of loneliness was less apparent. This difference may stem from the distinct pathways through which they operate: the modulation of loneliness often involves psychological factors, making it more subjective⁴², whereas individuals with moderate levels of loneliness might also engage in exercise spontaneously. Conversely, the influence of body image is more direct and evident. For example, individuals may avoid physical activity due to a lack of confidence in their body shape, which in turn affects their exercise self-efficacy⁴³. Consequently, the mediating effect of loneliness is relatively lower. Although the impact of loneliness was minor in this study, it still poses potential negative risks.

In the process of improving exercise self-efficacy through physical activity, it is important to focus on the positive role of body image and remain vigilant about the vicious cycle that loneliness might trigger, especially its potential threat to mental health, such as tendencies towards depression. When designing intervention measures, it is essential to foster a positive body image and create a supportive social environment to improve individuals’ exercise self-efficacy and mental health levels comprehensively.

Research limitations and future perspectives

While this study provides valuable insights into the underlying mechanisms linking physical activity and exercise self-efficacy among university students, several limitations warrant attention in future research. First, the cross-sectional design employed here can only reveal correlations between variables and cannot infer causality. Although we examined the relationships among physical activity, loneliness, body image, and exercise self-efficacy using mediation models, the causal direction of these relationships remains undetermined. Future research should adopt longitudinal designs, incorporating multi-wave measurements and long-term follow-ups, to verify causal links between the variables. Second, constrained by practical limitations, the measurement of physical activity relied on self-report questionnaires (e.g., PARS-3), which are susceptible to recall bias. Future studies should employ objective tools, such as accelerometers, to enhance measurement accuracy and reliability. Moreover, the study sample comprised a relatively homogeneous university student population with a limited age range and diversity. This homogeneity may obscure potential moderating effects of demographic variables such as age and gender. Concurrently, the research focused solely on loneliness and body image as mediating variables, neglecting other potential mediating factors such as exercise motivation, emotion regulation ability, exercise identity, anxiety, and psychological resilience. This narrow scope limits the generalizability of the findings. Future research should refine study designs, measurement instruments, and analytical approaches to provide a more comprehensive and in-depth understanding of the complex interplay inherent in this relationship.

Conclusion

Employing parallel mediation analysis, this study demonstrated a significant direct positive association between university students’ physical activity levels and exercise self-efficacy, as well as significant mediating effects of loneliness and body image. Physical activity indirectly enhanced exercise self-efficacy by alleviating loneliness and improving body image; body image exhibited a more pronounced mediating effect. These findings provide a theoretical basis for understanding the relationship between the two and emphasize the importance of considering individuals’ psychological states when promoting physical activity. By improving body image and reducing loneliness, the exercise self-efficacy of college students can be effectively improved, which can motivate them to engage in healthy behaviours and create a virtuous cycle actively. The findings of this study have profound practical implications for public health and physical education. The research results highlight the importance of physical activity in improving exercise self-efficacy, which can provide a basis for policymakers to design policies and programmes that promote physical activity. For example, by increasing the number of public sports facilities and providing community sports programmes and activities, people can be encouraged to participate more actively in physical activity. Moreover, physical education teachers and coaches can use these findings to design curricula that focus not only on skill and physical training but also on improving students’ exercise self-efficacy and body image. This approach can strengthen students’ positive attitudes and participation in physical activities. Additionally, understanding the impact of loneliness and body image on exercise self-efficacy can contribute to the design of courses, especially for individuals who may avoid physical activity due to feelings of loneliness or negative body image. For example, group therapy or social skills training can help these individuals build confidence and social connections. Intervention measures should also focus on fostering positive body image and creating a supportive social environment to comprehensively improve college students’ exercise self-efficacy and mental health levels.

Author contributions

Fangbing Zhou and Jinchao Gao designed the study, conducted data analysis and interpretation, and took primary responsibility for writing the manuscript. Wenlei Wang and Cong Liu organized the research process and conducted the literature search. Chunxue Shao performed the literature screening and assisted in drafting the manuscript. Yuyang Nie contributed resources, methdology, and validation. Jie Wu and Qiu Wentao provided experimental and testing guidance and contributed to revising the manuscript.

Funding

This study was conducted without financial support.

Data availability

The research data supporting this study have been uploaded to the GitHub database. Access link: https://github.com/nie-ship-it/PA-Z/blob/8f4d7cedc146b48f6a40c18f7de3e3ca81c39fda/%E6%9 C%80%E5%85%A8%E9%9D%A2%E4%B8%94%E6%9 C%80%E4%BC%98%E6%9 C%80%E6%96%B0%E6%95%B0%E6%8D%AE.sav.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

Ethical approval for this study was obtained from the Institutional Review Board (IRB) of the School of Psychology, Beijing Normal University BNU, under protocol number BNU202506090152. Informed consent was obtained from all participants.