Physical exercise moderates the mediating effect of sleep quality between internet addiction and college students’ academic burnout

Abstract

Internet addiction is a significant contributing factor to academic burnout among college students, though the underlying psychological and behavioral mechanisms warrant further investigation. This study investigated the mediating role of sleep quality between internet addiction and college students’ academic burnout, as well as the moderating role of physical exercise on this mediation pathway. A study was conducted to assess internet addiction, sleep quality, academic burnout, and physical exercise among college students. Descriptive statistics and correlation analyses were performed on these variables, and a moderated mediation model was developed. Internet addiction was positively correlated with sleep quality and academic burnout, and sleep quality was positively associated with academic burnout. Additionally, physical exercise demonstrated negative associations with internet addiction, sleep quality, and academic burnout. Furthermore, physical exercise moderated the relationship between internet addiction and sleep quality. This study provides further insights into the psychological mechanisms underlying the relationship between internet addiction and academic burnout among college students. Sleep quality serves as a mediating factor, while physical exercise act as a moderating factor in the relationship between internet addiction and sleep quality. This study highlights the role of sleep quality and physical exercise in the relationship between internet addiction and academic burnout, offering novel perspectives and strategies for addressing these challenges among college students.

Introduction

The rapid development of internet technology in recent years has made internet addiction a significant social problem, negatively impacting college students’ psychological well-being and learning efficacy¹. Research shows that when college students overuse digital media (e.g., social networks, online games, short videos), they often experience cognitive deficits, emotional instability, and poor impulse control, resulting in academic burnout². Academic burnout refers to students’ negative attitudes and behaviors resulting from prolonged exposure to academic stress or loss of learning motivation. It manifests as emotional exhaustion, cynicism, and reduced personal accomplishment³, which significantly impair students’ learning motivation and academic performance⁴. Currently, academic burnout has become a prevalent phenomenon in universities, with more than half of students experiencing academic burnout⁵. Academic burnout not only significantly diminishes individuals’ self-esteem and academic self-efficacy but also reduces their learning engagement, ultimately impairing subjective well-being and overall quality of life⁶. It is noteworthy that the prevalence of internet addiction among Chinese college students has shown a consistent upward trend⁷, with the consequent issue of academic burnout becoming increasingly prominent. Against this background, an in-depth exploration of the intrinsic relationship and underlying mechanisms between internet addiction and academic burnout among college students is not only conducive to improving the mental health status of college students but also has important practical significance for enhancing the quality of talent cultivation for college students.

Internet addiction and academic burnout

Internet addiction refers to an intense sense of dependence on the internet in the absence of addictive substances, accompanied by a loss of control over usage impulses⁸. It primarily encompasses compulsive online activities such as social networking, gambling, gaming, cybersex, and shopping⁹. Internet addiction demonstrates significant associations with academic impairments, including increased academic stress¹⁰ and declining grade point averages¹¹. Furthermore, internet addiction heightens the risk of psychological and social dysfunction, manifesting as depression, anxiety, and impairments in academic performance, occupational functioning, and interpersonal relationships¹². Conservation of resources theory provides a robust framework for understanding these relationships. The theory posits that individuals strive to protect their valued resources (e.g., time, energy, self-esteem), and stress occurs when resources are lost or threatened¹³. Critically, conservation of resources theory introduces the concept of resource loss spirals, where initial loss leads to further losses, creating a vicious cycle. In the context of internet addiction, college students may initially use the internet as a substitute coping strategy to escape academic pressure or compensate for depleted psychological resources¹⁴. However, excessive and compulsive internet use consumes time, attention, and cognitive energy without replenishing lost resources. Instead, it exacerbates resource depletion by displacing sleep, reducing study time, and increasing guilt or anxiety. Thus, internet addiction traps students in a resource loss spiral: academic stress → escapist internet use → further resource loss → heightened stress and burnout. Therefore, we propose Hypothesis 1: Internet addiction positively predicts academic burnout among college students.

The mediating effect of sleep quality

Sleep quality is a core indicator for assessing an individual’s sleep status, encompassing multiple dimensions such as difficulty falling asleep, sleep maintenance disturbances, frequency of nighttime awakenings, and morning refreshments¹⁵. Research indicates a significant relationship between sleep quality and internet addiction¹⁶. Sleep problems and poor sleep quality were positively related to high internet use among college students¹⁷. One systematic review and meta-analysis confirmed a significant negative correlation between internet addiction and sleep quality among college students¹⁸. Furthermore, a negative association was found between the degree of internet addiction and sleep quality among college students¹⁹. These findings suggest that internet addiction may lead to sleep problems, while poor sleep quality may also result in excessive internet use. Sleep quality also demonstrates a robust association with academic burnout. High-quality sleep quality exerts positive effects on academic performance²⁰, whereas excessive smartphone dependence adversely impacts learning outcomes and exacerbates academic burnout severity²¹. College students with better sleep quality generally exhibit superior academic adaptation and performance²². From the perspective of conservation of resources theory, sleep quality functions as a key resource reservoir. Sleep is a fundamental recovery resource that replenishes energy, restores self-regulatory capacity, and consolidates memory²³. According to the theory, resource loss is more impactful than resource gain, and individuals must invest resources to prevent further loss. Internet addiction depletes sleep quality through multiple mechanisms: late-night screen time displaces sleep hours, blue light disrupts circadian rhythms, and cognitive arousal delays sleep onset. Poor sleep quality, in turn, represents a critical resource deficit that reduces students’ ability to cope with academic demands, impairs emotion regulation, and accelerates depletion of remaining cognitive resources. This creates a mediated resource loss pathway: internet addiction → deterioration of sleep quality → increased vulnerability to academic burnout. Sleep quality thus acts as the “resource bridge” through which internet addiction propagates its negative effects. Consequently, deteriorated sleep quality may serve as a potential risk factor for academic burnout among college students. Studies demonstrate that smartphone addiction significantly impairs students’ self-regulatory capacity and time-management efficiency, consequently triggering academic procrastination and predisposing them to burnout syndrome²⁴. Based on the aforementioned research, this study proposes Hypothesis 2: Sleep quality mediates the relationship between internet addiction and academic burnout among college students.

The moderating effect of physical exercise

The effect of internet addiction on academic burnout may be moderated by behavioral interventions such as physical exercise. Physical exercise is any bodily movement that is planned, structured, repetitive, and purposefully focused on the improvement or maintenance of one or more components of physical fitness²⁵. Regular physical exercise can help regulate circadian rhythms and promote normal melatonin secretion²⁶, thereby counteracting the disruption of biological rhythms caused by nighttime blue light exposure in individuals with internet addiction²⁷. Meanwhile, physical exercise has been demonstrated to effectively alleviate cognitive load and psychological stress induced by Internet use²⁸, reduce bedtime procrastination, and improve sleep quality²⁹. Furthermore, physical exercise can enhance self-regulatory capacity, enabling individuals with internet addiction to better manage their online time and thereby reducing the negative impacts of excessive Internet use on both academic performance and sleep quality^30,31. Conservation of resources theory provides a powerful explanation for why physical exercise acts as a moderator. According to the theory, individuals can invest resources to prevent future loss or initiate gain spirals. Physical exercise serves as a resource investment strategy that buffers the resource loss spiral triggered by internet addiction. Specifically, exercise promotes resource replenishment by improving sleep quality through regulating circadian rhythms and reducing stress, directly counteracting the sleep-depleting effect of internet addiction. It also facilitates resource generation by enhancing self-regulatory capacity and psychological resilience—“resource caravans” that help students control internet use and protect sleep time. By protecting sleep quality from the detrimental effects of internet addiction, exercise weakens the mediating pathway. In conservation of resources terms, exercise acts as a resource gain strategy that disrupts the loss spiral: students who exercise can maintain or even improve sleep quality, reducing the translation of internet addiction into academic burnout. Thus, physical exercise moderates the relationship between internet addiction and sleep quality by providing healthy resource replenishment that competes with maladaptive internet use, thereby disrupting the indirect pathway to academic burnout. Based on this, the present study proposes Hypothesis 3: Physical exercise moderates the relationship between internet addiction and sleep quality among college students.

Under the guidance of the conservation of resources theory, this study attempts to construct a moderated mediation model with internet addiction as the independent variable, academic burnout as the dependent variable, sleep quality as the mediating variable, and physical exercise as the moderating variable. The research aims to further elucidate the underlying psychological and behavioral mechanisms between internet addiction and academic burnout among college students.

The proposed model is illustrated in Fig. 1. In this proposed model, the three hypotheses form a clear hierarchical structure. Specifically, H1 establishes the direct predictive relationship between internet addiction and academic burnout, serving as the total effect in the model. H2 proposes a mediation mechanism, in which sleep quality acts as an intervening variable that transmits the effect of internet addiction to academic burnout. H3 further extends this mediation model by proposing a moderated mediation effect, wherein physical exercise moderates the relationship between internet addiction and sleep quality, thereby influencing the strength of the indirect pathway. Collectively, these hypotheses move from a direct effect (H1) to a mediation model (H2) and finally to a moderated mediation model (H3), offering a progressively deeper understanding of how and under what conditions internet addiction leads to academic burnout among college students.

Fig. 1.

Moderated mediation model.

Participants and methods

Participants

The survey was conducted among college students from six universities in Guangdong Province using a stratified cluster random sampling method from October to December 2025. Participants were first stratified by academic discipline (Engineering and Technology, Natural Sciences, Humanities and Literature, Social Sciences and Management, and Arts), and cluster sampling was then conducted by academic year (Year 1 through Year 4) within each disciplinary stratum. A total of 1,060 questionnaires were collected in this survey. After excluding 53 invalid responses (including incomplete, contradictory, or patterned answers), 1,007 valid questionnaires were retained, yielding an effective response rate of 95.0%. To ensure the sample size was adequate to detect the hypothesized effect, an a priori power analysis was conducted. Using G*Power software, the analysis was configured with an effect size of f² = 0.15, α = 0.05, and power (1 − β) = 0.80. The choice of a medium effect size follows Cohen’s guidelines (f² ≥ 0.15) for multiple regression and structural equation modeling³², a standard consistently adopted in behavioral and educational research^33,34. Statistical power analysis was performed using GPower 3.1. Available from: https://www.psychologie.hhu.de/arbeitsgruppen/allgemeine-psychologie-und-arbeitspsychologie/gpower? Prior studies on internet addiction, sleep quality, and academic burnout have also anticipated medium effect sizes. In addition, following the rule of thumb that sample size should be 5 to 10 times the number of items, the 62 items in this study required a minimum of 620 participants. The actual sample size of 1,007 meets and exceeds this criterion. All participants engaged in this study voluntarily after signing informed consent forms. A comprehensive pre-survey briefing covered the research objectives, data protection measures, and specific applications of the collected data. The study protocol was approved by the Academic Ethics Committee of Zhaoqing University in accordance with the Declaration of Helsinki.

Measurement tools

Internet addiction scale

The Revised Chinese Internet Addiction Scale (CIAS-R), revised by Bai and Fan³⁵ based on Young’s Internet Addiction Test (IAT)⁸, was utilized to assess Internet addiction among college students. The scale consists of 19 items across four dimensions: compulsive internet use and withdrawal symptoms, internet addiction tolerance, interpersonal and health problems, and time management issues. A 4-point Likert scale was used, ranging from 1 (strongly disagree) to 4 (strongly agree). The total score ranges from 19 to 76, with higher scores reflecting more severe Internet addiction tendencies. This measurement tool has been widely verified and applied in a population of Chinese college students³¹. In this study, the scale’s Cronbach’s α was 0.920, and the coefficients for its four dimensions were 0.870, 0.926, 0.936, and 0.920, respectively. The confirmatory factor analysis revealed acceptable model fit: χ²/df = 4.361; GFI = 0.941; AGFI = 0.918; IFI = 0.977; TLI = 0.971; CFI = 0.977; RMSEA = 0.057.

Academic burnout scale

The Chinese version of the Learning Burnout Scale (LBS-C), originally developed by Lian et al.³⁶ based on the Maslach Burnout Inventory-Student Survey³⁷, was employed to assess college students’ academic burnout. The scale consists of 20 items categorized into three dimensions: emotional exhaustion, improper behavior, and low accomplishment. Items were rated on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Higher total scores indicate more severe academic burnout. This measurement tool has been widely verified and applied in a population of Chinese college students³⁸. In this study, the full scale achieved a Cronbach’s α of 0.942, and its three subscales obtained coefficients of 0.947, 0.918, and 0.949. The confirmatory factor analysis demonstrated a good model fit: χ²/df = 3.588; GFI = 0.945; AGFI = 0.931; IFI = 0.976; TLI = 0.972; CFI = 0.976; RMSEA = 0.050.

Pittsburgh sleep quality Index

The Pittsburgh sleep quality Index (PSQI) was developed by Buysse et al.³⁹ and revised by Liu et al.⁴⁰ to assess the sleep quality of college students. It consists of 18 items measuring seven distinct components: subjective sleep quality, sleep latency, sleep efficiency, sleep duration, sleep disturbances, use of sleep medications, and daytime dysfunction. The scale consists of 4 open-ended questions and 14 multiple-choice items, with the multiple-choice questions scored from 1 to 3. The total score ranges from 0 to 21. The higher the score, the poorer the sleep quality. The Cronbach’s α coefficient of this scale was 0.776.

Physical activity rating scale

Physical exercise was assessed using the physical activity rating scale (PARS-3) developed by Hashimoto⁴¹ and compiled by Liang⁴². The scale consisted of three items: intensity, time, and frequency. Amount of physical exercise = intensity × time × frequency. Each item is rated on a five-point scale, with intensity and frequency scored from 1 to 5, and time scored from 0 to 4. The higher score represents a higher level of physical exercise. This measurement tool has been widely verified and applied in a population of Chinese college students³⁸. In this study, Cronbach’s α for the scale was 0.809.

Statistical analysis

The data analysis was conducted using SPSS version 26.0 and the PROCESS macro (version 4.1) for mediation and moderation analysis. Harman’s single-factor test was conducted to assess common method bias. Descriptive statistical analyses were carried out for variables. Pearson correlation analysis was then employed to assess the correlations among variables. Hayes’ PROCESS macro (version 4.1, Model 4) was utilized to test the mediating effect of sleep quality between Internet addiction and academic burnout. Further, Hayes’ PROCESS macro (Version 4.1, Model 7) was employed to examine the moderating effect of physical exercise on the mediating role of sleep quality in the relationship between Internet addiction and academic burnout. Additionally, the study employed a Bootstrap method with 5,000 resamples to obtain the 95% confidence interval (95% CI), with a non-inclusion of zero in the interval indicating significance. To ensure the robustness and replicability of the bootstrap results, we performed 5,000 bootstrap resamples with a random seed fixed at 12,345. For associations involving categorical variables, the analytical approach was tailored to the variable type: point-biserial correlations were used for binary variables (e.g., gender), and ANOVA was conducted for unordered categorical variables (e.g., family socioeconomic status). Pearson correlation analysis was reported solely for continuous-continuous pairs.

Results

Demographic characteristics of participants

Participants comprised 498 males (49.5%) and 509 females (50.5%). The distribution across academic years was as follows: 223 freshmen (22.2%), 392 sophomores (38.9%), 240 juniors (23.8%), and 152 seniors (15.1%). Descriptive statistics for key continuous variables were as follows: age (20.00 ± 1.12), family socioeconomic status (5.44 ± 2.88), self-rated health (3.68 ± 1.13), neuroticism (6.05 ± 3.02), and deprehssive symptoms (2.10 ± 1.59).

Common method bias test

Common method bias refers to systematic bias or error that occurs when all variables are influenced by the same measurement method or source, typically assessed using Harman’s single-factor test. In this study, the variance explained by the first factor was only 32.23%, which is below the 40% threshold recommended by scholars. Therefore, it can be concluded that the survey results do not exhibit severe common method bias.

Correlation analysis

Descriptive statistics were calculated for internet addiction, sleep quality, academic burnout, and physical exercise. Pearson correlation analysis was then performed to examine the bivariate linear relationships among these variables. As shown in Table 1, internet addiction is positively correlated with sleep quality (r = 0.269, p < 0.01) and academic burnout (r = 0.301, p < 0.01). Sleep quality is positively correlated with academic burnout (r = 0.424, p < 0.01). Additionally, physical exercise is negatively correlated with internet addiction (r = − 0.265, p < 0.01), sleep quality (r = − 0.222, p < 0.01) and academic burnout (r = − 0.211, p < 0.01).

Table 1.

Correlation and descriptive statistics.

Variable	1	2	3	4
1 Internet addiction	1
2 Sleep quality	0.269**	1
3 Physical exercise	− 0.265**	− 0.222**	1
4 Academic burnout	0.301**	0.424**	− 0.211**	1
M	44.159	5.921	28.105	64.879
SD	14.366	3.405	7.105	17.168

**p < 0.01. M, mean; SD, standard deviation. Sleep quality represents the total PSQI score.

Diagnosis of multicollinearity

To prevent the impact of multicollinearity on the regression results, variance inflation factor (VIF) tests were conducted on the independent variables in this study. The results indicated that the VIF values for all independent variables ranged from 1.009 to 1.422, all well below the critical threshold of 5. This suggests that there is no severe multicollinearity issue among the variables, confirming their appropriateness for subsequent regression analysis.

Mediation test

A simple mediation analysis was conducted using Model 4 of the PROCESS v4.1 macro developed by Hayes⁴³, with sleep quality as the mediator, internet addiction as the independent variable, and academic burnout as the dependent variable. Gender, age, family socioeconomic status, self-rated health, neuroticism, and depressive symptoms were included as control variables. The results in Table 2 and Fig. 1 show that internet addiction has a positive effect on academic burnout (β = 0.277, t = 9.228, p < 0.001). After introducing sleep quality into the regression model, internet addiction also has a positive effect on academic burnout (β = 0.192, t = 6.568, p < 0.001). Furthermore, internet addiction demonstrated a positive effect on sleep quality (β = 0.227, t = 8.351, p < 0.001), and sleep quality shows a positive effect on academic burnout (β = 0.373, t = 11.375, p < 0.001) (Fig. 2).

Table 2.

Regression analysis.

Regression analysis (N = 1,007)		Goodness-of-fit measures		Statistical significance of the coefficients
Outcome variables	Control variables/predictor variables	R2	F(df)	β	SE	t	P	BC 95% CI Lower	BC 95% CI Upper
Academic burnout		0.130	14.905*** (10)
	Gender			0.062	1.029	2.061	0.040	0.003	0.121
	Age			− 0.009	0.457	− 0.286	0.775	− 0.067	0.050
	Family socioeconomic status			0.002	0.177	0.073	0.942	− 0.056	0.061
	Self-rated health			0.015	0.450	0.504	0.615	− 0.043	0.073
	Neuroticism			0.131	0.172	4.339	0.000	0.072	0.190
	Deprsssive symptoms			0.096	0.329	3.153	0.002	0.036	0.156
	Internet addiction			0.277	0.036	9.228	0.000	0.218	0.336
Sleep quality		0.282	39.127***(10)
	Gender			− 0.049	0.186	− 1.788	0.074	− 0.102	0.005
	Age			0.006	0.082	0.214	0.831	− 0.047	0.059
	Family socioeconomic status			− 0.008	0.032	− 0.281	0.779	− 0.060	0.045
	Self-rated health status			− 0.022	0.081	− 0.825	0.410	− 0.075	0.031
	Neuroticism			0.196	0.031	7.122	0.000	0.141	0.248
	Deprsssive symptoms			0.345	0.059	12.468	0.000	0.290	0.399
	Internet addiction			0.227	0.007	8.351	0.000	0.174	0.280
Academic burnout		0.230	27.059***(11)
	Gender			0.080	0.970	2.830	0.005	0.025	0.136
	Age			− 0.011	0.430	− 0.381	0.703	− 0.066	0.044
	Family socioeconomic status			0.005	0.167	0.179	0.858	− 0.050	0.060
	Self-rated health status			0.023	0.424	0.832	0.406	− 0.032	0.078
	Neuroticism			0.058	0.166	1.993	0.047	0.001	0.115
	Deprsssive symptoms			− 0.033	0.333	− 1.063	0.288	− 0.093	0.028
	Internet addiction			0.192	0.035	6.568	0.000	0.135	0.249
	Sleep quality			0.373	0.166	11.375	0.000	0.310	0.439

df, degrees of freedom; SE, standard error; BC 95% CI, bias-corrected 95% bootstrap confidence interval. β is the standardized coefficient.

Fig. 2.

Mediation model test. Note: ***p < 0.001.

The mediating role of sleep quality between internet addiction and academic burnout was further examined using the Bootstrap method. As shown in Table 3, the total effect of internet addiction on academic burnout is 0.277 (95% CI [0.218, 0.336]), with the confidence interval not including 0, indicating a statistically significant effect and supporting Hypothesis 1. Internet addiction shows a significant direct effect on academic burnout (β = 0.192, 95% CI [0.130, 0.250]), accounting for 69.29% of the total effect. Simultaneously, the indirect effect through sleep quality was also significant (β = 0.085, 95% CI [0.059, 0.114]), representing 30.71% of the total effect. These results demonstrate that sleep quality serves as a mediator in the relationship between internet addiction and academic burnout, supporting Hypothesis 2. The convergence of both mediation testing methods confirms that sleep quality plays a significant partial mediating role between internet addiction and academic burnout.

Table 3.

Bootstrap-based mediation effect test.

Effect	β	Boot SE	95% CI	Ratio of effect
Direct effect	0.192	0.030	[0.130, 0.250]	69.29%
Indirect effect	0.085	0.014	[0.059, 0.114]	30.71%
Total effect	0.277	0.030	[0.218, 0.336]

5000 bootstrap resamples. All coefficients reported are standardized effects (β). Boot SE = bootstrap standard error; 95% CI = 95% bootstrap confidence interval.

Moderated mediation test

Using Model 7 from the PROCESS v4.1 macro developed by Hayes⁴³, we analyzed the moderating effect of physical exercise on the relationship between internet addiction and sleep quality, with gender, age, family socioeconomic status, self-rated health status, neuroticism, and depressive symptoms as control variables. To reduce non-essential multicollinearity and improve result interpretation, all variables were standardized, and standardized interaction terms between the independent variable and moderator were computed. The results are presented in Table 4 and Fig. 3. Internet addiction shows a positive effect on sleep quality (β = 0.177, t = 6.269, p < 0.001), while physical exercise demonstrates a negative effect on sleep quality (β = − 0.150, t = − 5.136, p < 0.001). Importantly, the interaction term between internet addiction and physical exercise also exhibited a negative effect on sleep quality (β = − 0.077, t = − 2.744, p < 0.01). These findings indicate that physical exercise significantly moderates the relationship between internet addiction and sleep quality, supporting Hypothesis 3.

Table 4.

Moderated mediation analysis.

Regression analysis (N = 1,007)		Goodness-of-fit measures		Statistical significance of the coefficients
Outcome variables	Control variables/predictor variables	R2	F(df)	β	SE	t	P	BC 95% CI Lower	BC 95% CI Upper
Sleep quality		0.303	35.979***(12)
	Gender			− 0.027	0.189	− 0.964	0.335	− 0.081	0.028
	Age			0.016	0.082	0.581	0.561	− 0.037	0.068
	Family socioeconomic status			− 0.002	0.032	− 0.070	0.944	− 0.054	0.050
	Self-rated health status			− 0.017	0.080	− 0.629	0.529	− 0.069	0.036
	Neuroticism			0.182	0.031	6.726	0.000	0.129	0.235
	Depresssive symptoms			0.341	0.059	12.498	0.000	0.288	0.395
	Internet addiction (A)			0.177	0.007	6.269	0.000	0.122	0.233
	Physical exercise (B)			− 0.150	0.004	− 5.136	0.000	− 0.207	− 0.092
	A × B			− 0.077	0.000	− 2.744	0.006	− 0.131	− 0.022
Academic burnout		0.230	27.059***(11)
	Gender			0.080	0.970	2.830	0.005	0.025	0.136
	Age			− 0.011	0.430	− 0.381	0.703	− 0.066	0.044
	Family socioeconomic status			0.005	0.167	0.179	0.858	− 0.050	0.060
	Self-rated health status			0.023	0.424	0.832	0.406	− 0.032	0.078
	Neuroticism			0.058	0.166	1.993	0.047	0.001	0.115
	Depresssive symptoms			− 0.033	0.333	− 1.063	0.288	− 0.093	0.028
	Internet addiction			0.192	0.035	6.568	0.000	0.135	0.249
	Sleep quality			0.373	0.166	11.375	0.000	0.310	0.439

df, degrees of freedom; SE, standard error; BC 95% CI, bias-corrected 95% bootstrap confidence interval. β is the standardized coefficient. p < 0.001.

Fig. 3.

Statistical diagram (**p < 0.01; ***p < 0.001).

The mediating role of sleep quality between internet addiction and academic burnout was further examined using the Bootstrap method across different levels of physical exercise. The results (see Table 5) showed that when the level of physical exercise was low (M − 1SD), the mediation effect of sleep quality was 0.095 (95% CI [0.062, 0.133]), and the confidence interval did not include zero, indicating a significant mediation effect. When the level of physical exercise was high (M + 1SD), the mediation effect decreased to 0.038 (95% CI [0.004, 0.074]), and the confidence interval still did not include zero, suggesting that the mediation effect remained significant but was diminished. The moderated mediation index was − 0.029 (95% CI [− 0.054, − 0.005]), and the confidence interval did not include zero, indicating that physical exercise had a significant moderating effect on the mediation role of sleep quality. In conclusion, higher levels of physical exercise weaken the mediating effect of sleep quality between internet addiction and academic burnout, suggesting a buffering role of physical exercise on this indirect pathway.

Table 5.

The mediating role of sleep quality between internet addiction and academic burnout at different physical exercise levels.

Physical exercise	β	Boot SE	Boot LLCI	Boot ULCI
M-1SD	0.095	0.018	0.062	0.133
M	0.066	0.013	0.042	0.094
M + 1SD	0.038	0.018	0.004	0.074
Index	− 0.029	0.012	− 0.054	− 0.005

M, mean; SD, standard deviation. The bias-corrected percentile bootstrap method was used, with 5,000 resamples, to estimate the confidence intervals. All coefficients are standardized effects (β). CI, confidence interval.

To further examine the moderating effect of physical exercise on the relationship between internet addiction and sleep quality, a simple slope analysis was performed by categorizing physical exercise and sleep quality into groups at ± 1 SD from the mean. Further simple slope analysis (see Table 6 and Fig. 4) revealed that compared to low levels of physical exercise (simple slope = 0.254, t = 6.942, p < 0.001), the positive effect of internet addiction on sleep quality gradually decreased among college students with moderate levels of physical exercise (simple slope = 0.177, t = 6.269, p < 0.001) and high levels physical exercise (simple slope = 0.101, t = 2.381, p < 0.05). These results indicate that as physical exercise levels increase, the effect of internet addiction on sleep quality progressively weakens.

Table 6.

The effect of internet addiction on sleep quality across different levels of physical exercise.

Physical exercise	β	SE	t	p	LLCI	ULCI
M-1SD	0.254	0.037	6.942	0.000	0.182	0.325
M	0.177	0.028	6.269	0.000	0.122	0.233
M + 1SD	0.101	0.043	2.381	0.018	0.018	0.185

M, mean; SD, standard deviation. The bias-corrected percentile bootstrap method was used, with 5,000 resamples, to estimate the confidence intervals. All coefficients reported are standardized effects (β). CI, confidence interval.

Fig. 4.

Simple slope plot. Note: PE, physical exercise. All data are presented as standardized coefficients.

Discussion

This study investigated the relationship between internet addiction and academic burnout among college students, examining the mediating role of sleep quality and the moderating role of physical exercise. Firstly, correlational analysis revealed a significant positive association between internet addiction and academic burnout among college students. Secondly, mediation analysis demonstrated that sleep quality played a mediating role between internet addiction and academic burnout among college students. Thirdly, the moderated mediation analysis revealed that physical exercise significantly moderated the relationship between internet addiction and sleep quality, thereby attenuating the negative effect of internet addiction on academic burnout. Based on the aforementioned findings, this study further elucidates the psychological and behavioral mechanisms underlying the relationship between internet addiction and academic burnout among college students.

The effect of internet addiction on academic burnout

This study found that internet addiction is positively correlated with poor sleep quality among college students, supporting Hypothesis 1 and aligning with previous research^15,44. Internet addiction could serve as a potential predictor of academic burnout among college students¹⁵. Conservation of Resources Theory posits that individuals experience burnout when they undergo persistent loss of psychological resources (e.g., time, attention, and emotion regulation capacity)¹³. Internet addiction exacerbates academic burnout among college students by depleting psychological capital through three pathways: time displacement, cognitive resource depletion, and diminished social support⁴⁵. The study revealed that college students with internet addiction exhibit diminished self-control, predisposing them to a vicious cycle of "late-night smartphone overuse → next-day academic inefficiency"^46,47. From a neurobiological perspective, internet addiction induces significant structural alterations in the brain, manifesting as reduced gray matter volume in specific regions and abnormal white matter integrity, with some changes demonstrating duration-dependent correlations^48,49. Furthermore, neurobiological alterations collectively form the biological basis for the impact of internet addiction on academic burnout through multiple pathways that impair motivational regulation (dopaminergic pathways), impair cognitive resources (prefrontal cortex deficits), exacerbate emotional depletion (HPA-amygdala system hyperactivity), and decrease neuroplasticity (hippocampal and white matter deterioration)⁵⁰. In summary, internet addiction predicts and exacerbates academic burnout among college students through multiple pathways, including psychological resource depletion, diminished self-regulation, reduced academic self-efficacy, and neurobiological mechanisms.

The mediating effect of sleep quality

This study found that sleep quality mediates the relationship between internet addiction and academic burnout among college students, confirming Hypothesis 2. Consistent with prior research²⁰, the present study revealed a significant positive correlation between internet addiction and poor sleep quality among college students. Firstly, excessive internet use (particularly at night) can delay sleep onset, reduce total sleep duration, and disrupt circadian rhythms, leading to increased insomnia symptoms⁵¹. Individuals with internet addiction tend to allocate excessive time to online activities, particularly during nighttime. This excessive internet use directly displaces sleep time, thereby impairing sleep quality⁵². Studies found that college students with internet addiction commonly exhibit bedtime procrastination and poorer sleep quality⁵³. Secondly, the psychological stimulation from online activities (such as gaming and social media) can increase brain arousal and suppress melatonin secretion, thereby prolonging sleep latency⁵⁴. The blue light emitted by electronic devices suppresses melatonin secretion, delays circadian rhythms, and consequently impairs sleep quality⁵⁵. Furthermore, internet addiction is often accompanied by anxiety and depressive symptoms, which can further impair sleep maintenance capacity, leading to nocturnal awakenings and early morning awakenings^56,57.

This study also found that poor sleep quality was associated with academic burnout among college students, consistent with previous research¹⁷. Firstly, sleep plays an important role in learning processes and memory consolidation^58,59. Both sleep quality and sleep duration are closely associated with students’ learning ability and academic performance⁶⁰. Sleep loss is typically linked to poorer learning capacity, whereas sleep optimization can enhance academic achievement⁶⁰. Sleep deprivation impairs cognitive performance (e.g., attention and memory), leading to reduced learning efficiency. This compels students to invest more time in completing academic tasks, consequently exacerbating emotional exhaustion and fatigue⁶¹. Furthermore, sleep disturbances disrupt emotional regulation, exacerbate mood disorders, and trigger anxiety and depressive symptoms, thereby further diminishing academic motivation⁶². Moreover, chronic poor sleep quality diminishes self-efficacy, reducing the motivation and confidence required to cope with daily challenges, ultimately impairing academic performance⁶³. In conclusion, behavioral interventions (e.g., reducing screen time) and cognitive interventions (e.g., decreasing psychological dependence on the internet) may effectively improve sleep quality, thereby alleviating academic burnout.

The moderating effect of physical exercise

This study found that physical exercise moderated the relationship between internet addiction and sleep quality, thereby validating Hypothesis 3. Notably, all three levels of physical exercise (low, moderate, and high) demonstrated significant moderating effects, albeit with distinct patterns and practical implications.

Physical exercise can improve sleep quality without notable adverse effects^64,65. From a psychological mechanism perspective, physical exercise effectively alleviates anxiety and depressive symptoms associated with Internet addiction by enhancing self-efficacy and emotion regulation capacity, thereby reducing bedtime procrastination^66,67. Exercise-induced release of endogenous opioid peptides not only enhances positive emotional states but also directly modulates the sleep–wake cycle⁶⁸. Physiologically, physical exercise enhances sleep quality by boosting melatonin to regulate circadian rhythms and offset blue light effects⁶⁹, and activating the endogenous opioid system to modulate sleep–wake cycles⁶⁸. Research indicates that physical exercise can reduce Internet addiction levels⁷⁰. Indeed, physical exercise can not only alleviate Internet addiction but also mitigate the negative effects associated with Internet addiction^70,71.

Elaborating on the practical significance of different exercise levels, low-level physical exercise (e.g., occasional walking or stretching, < 3 times per week, < 30 min per session) primarily offers a low-threshold entry point for physically inactive or severely addicted students. Even such minimal activity can initiate the disruption of sedentary behavior and the vicious cycle of “addiction-poor sleep”—for instance, by briefly reducing evening screen time and alleviating immediate state anxiety⁶⁶. The moderation effect at this level is significant but relatively modest, suggesting that while any amount of exercise is beneficial, merely meeting minimal standards may be insufficient to fully counteract the negative consequences of internet addiction. Medium-level physical exercise (e.g., 3–5 times per week, 30–60 min per session of moderate-intensity activities such as jogging or swimming) appears to produce optimal moderation effects. This level is realistically achievable for most college students and can be integrated into campus routines (e.g., as credit-bearing physical education courses or extracurricular sports clubs). It likely facilitates more consistent circadian rhythm entrainment and sustained elevation of brain-derived neurotrophic factors, thereby enhancing executive control and self-regulatory capacity over internet use behaviors⁷³. High-level physical exercise (e.g., daily vigorous training, > 60 min per session) demonstrated the strongest moderating effect, although its feasibility for the general student population is limited due to time commitment and risk of overtraining. For student athletes or exercise enthusiasts, however, this high level provides the most robust protection, possibly via more profound modulation of the hypothalamic–pituitary–adrenal (HPA) axis function and greater neuroplasticity in the prefrontal cortex and hippocampus^72,73. From a neurobiological perspective, exercise can alleviate internet addiction by modulating neurobiological mechanisms in both the central and autonomic nervous systems⁷². Physical exercise enhances neuroplasticity in the prefrontal cortex and hippocampus, thereby strengthening executive control functions and improving self-regulatory capacity over internet use behaviors⁷³. Regular physical exercise also modulates HPA axis function, reduces cortisol levels, and alleviates stress responses associated with internet addiction⁷². Physical exercise also enhances the secretion of brain-derived neurotrophic factors, improves emotional regulation capacity, and reduces anxiety and depressive symptoms associated with internet addiction, thereby promoting better sleep quality⁷⁴.

Furthermore, physical exercise is also associated with reduced academic burnout³⁸. Physical exercise (both aerobic and strength) was effective in reducing academic burnout levels in college students⁷⁵. College students who engage in regular physical exercise exhibit lower levels of academic burnout, which may be attributed to improved sleep quality, enhanced self-regulation, and reduced physiological susceptibility to stress^38,76. Research indicates that higher levels of exercise participation correlate with lower degrees of academic burnout in college students³⁸. In summary, physical exercise may reduce academic burnout levels among college students by modulating the impact of internet addiction on sleep quality. From a practical standpoint, these findings suggest that campus-based exercise interventions should be tiered: promoting simple, accessible activities (e.g., standing stretches or brief walks) to engage sedentary students; encouraging moderate, structured physical activity through physical education and recreational sports for the general population; and providing specialized support for student athletes to maintain high-intensity training without overuse injuries. Such a stratified approach can maximize the protective role of physical exercise in mitigating the adverse effects of internet addiction on sleep quality and, consequently, academic burnout.

Limitations

First, this study employed a cross-sectional design. Given the ambiguity of temporal sequence, single-source bias, and cultural/contextual specificity issues, it is recommended that future research adopt longitudinal or experimental designs for validation. Second, the sample in this study was limited to college students from a single province in China. Since internet usage patterns, academic stress, sleep habits, and physical exercise practices may vary significantly across different regions, types of institutions, or cultural backgrounds, future studies should expand the sample size and enhance its diversity and representativeness. Third, the research data collection primarily relied on self-reported questionnaires, which may introduce social desirability bias, recall bias, and subjective bias. Future research could consider incorporating more objective measurement methods (e.g., device-recorded screen time, fitness trackers monitoring sleep and activity, academic performance records, etc.) to supplement or validate self-reported data. Fourth, although we implemented procedural remedies and Harman’s single-factor test did not indicate severe common method bias, we acknowledge the inherent limitations of cross-sectional, self-report data and the diagnostic constraints of Harman’s test. Future research could employ longitudinal designs or incorporate marker variables to more robustly control for this potential bias. Finally, certain psychosocial confounders were not measured. Other potential confounding variables, such as social support, academic motivation, and disciplinary characteristics, should be adequately controlled or investigated in future study. Moreover, additional mediating or moderating mechanisms (e.g., self-control ability, coping strategies, time management skills) should be further explored.

Conclusions

This study investigated the relationship between college students’ internet addiction and academic burnout, revealing sleep quality as a mediating factor and physical exercise as a moderator. Specifically, sleep quality mediated the relationship between internet addiction and academic burnout. The mediating effect of sleep quality was significantly moderated by physical exercise, with higher levels of physical exercise attenuated the adverse effects of internet addiction on sleep quality. Furthermore, higher levels of physical exercise attenuated the negative impact of internet addiction on sleep quality. These findings elucidate the effect of internet addiction on academic burnout among college students and its underlying mechanisms, underscoring the moderating role of physical exercise. This study highlights the distinctive role of physical exercise in mitigating the effect of internet addiction on academic burnout, offering novel perspectives and strategies for addressing these challenges among college students.

Author contributions

Qingkun Feng and Yanying Liu wrote the main manuscript text and performed data analysis, Yao Tong interpreted the results, and Guihua Huang distributed the questionnaires and collected the data. All authors reviewed the manuscript and approved the submitted version.

Funding

This study was supported by the Guangdong Philosophy and Social Sciences Planning Project (Grant No. GD25YDXZTY01 and GD26YDXZ16), the Teaching Quality Construction Project for Undergraduate Colleges and Universities in Guangdong Province (Grant No.: Yue Higher Education Letter [2024] No. 9-298), and the Zhaoqing University university-level Scientific Research Fund Key Project (ZD202402).

Data availability

The data and materials used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki. Ethical approval was obtained from the Academic Ethics Committee of Zhaoqing University (Approval ID: 2025040; Date of Approval: July 9, 2025). Informed consent was obtained from all individual participants included in the study. All participants were aged 18 years or older.