The influence of physical activity on internet addiction among Chinese college students: the mediating role of self-esteem and the moderating role of gender

The sampling period for this study was from June 1st to 10th, 2022. Due to the impact of the COVID-19 pandemic in most provinces of China during that time, a convenience sampling method was adopted to select three provinces that were not affected by the pandemic. Simple random sampling was then used to select five universities: Zhengzhou University, Beijing Normal University, Inner Mongolia University of Science and Technology, Henan University of Engineering, and China University of Mining and Technology (Beijing). This study employed a cluster random sampling method, with classes as the unit of sampling, to select 950 college students [19.96 ± 2.69 years old, 497 male students (20.02 ± 3.17 years old) and 358 female students (19.66 ± 1.79 years old ) ] as the research subjects. All participants were between 18 and 24 years old.

In this study, the test was implemented collectively as a class, with two experienced postgraduates serving as the primary examiners for each class. Prior to the implementation of the test, special training (For instance, how to connect with the respective universities, follow the established steps, maintain proper etiquette and courtesy, introduce ourselves when conducting the survey, outline the contents of the questionnaire, explain each item, provide instructions for filling out the questionnaire, time the process, conduct a final check of the completed questionnaires, and ensure confidentiality throughout the entire process.) was given to all examiners. Adolescents participated voluntarily after obtaining the consent of school leaders and teachers in charge. Measurements were conducted anonymously in the classroom, and before carrying out the survey, the test subjects were asked to fill out the questionnaires truthfully followed by instructions according to their real circumstances. All questionnaires were collected on the spot, and the investigator answered any questions asked by the subjects. In the process of administering the survey, the counselor was shunned and the questionnaires were filled out anonymously, without involving personal privacy such as student number. The document was also informed to clearly state that this survey is anonymous, the results of the survey are only used for scientific research, and will not pose any risk to the daily life of the participants, and that the participants’ participation in the study is completely voluntary. Participants were asked to confirm that they had read the document and agreed to participate in the study before continuing the survey. The test was completed in one sitting and took approximately 10 min, and all questionnaires were returned on the spot, with a total of 950 questionnaires distributed. After screening, 855 were valid, resulting in a questionnaire validity rate of 90%. The data screening process is shown in Fig. 2.This study was conducted in compliance with local laws and regulations and institutional requirements. For another, the study was approved by the Ethics Committee of Zhengzhou University (Approval No. ZZUIRB2022-JCYXY0029), with informed consent from all participants. Still, ensuring the voluntary participation and informed consent of all participants during the investigation should be seriously performed.

The scale was developed by the Japanese scholar, Mr. K.Y. Hashimoto [31] and revised by Mr. Liang Deqing et al. [32]. The scale has three dimensions, namely, activity intensity, activity time and activity frequency, with one item for each dimension and a 5-point Likert scale, where the higher the score, the greater the physical activity. The scale is scored by the formula “Intensity × (Time-1) × Frequency = Total Physical Activity Score”, with scores less than or Eq. 19 corresponding to a low level of physical activity, scores ranging from 20 to 42 indicating a moderate level of physical activity, and scores 43 representing a high level of physical activity. The reliability of this scale is high, with a re-test reliability of 0.82, and the total clombach α is 0.774.

The scale was compiled by Rosenberg in 1965 [14], introduced and translated by Ji Yifu and Yu Xin [33], and widely used in China. There are 10 items, with 1, 2, 4, 6, 7, and 8 as positively scored questions, and the rest as negatively scored questions. Adopting a 5-point Likert scale(By adding a neutral option between the options of a four-point Likert scale, the original four-point Likert scale is transformed into a five-point Likert scale.), the scale has a score range of 10–40, with higher scores indicating higher levels of self-esteem in individuals. The scale mainly measures overall self-esteem, which is simple to understand, easy to score, and is one of the most widely used self-esteem scales. The scale demonstrates good validity, with a retest reliability of 0.72. The total Cronbacha’s alpha of the scale is 0.89.

The scale was developed by Shuhui Chen et al. (2003) [34], with a total of twenty-six questions in five dimensions, namely: COMPULSIVENESS (α = 0.886), WITHDRAWAL SYMPTOMS (α = 0.93), TOLERANCE (α = 0.911), INTERPERSONAL HEALTH PROBLEMS (α = 0.895), and TIME MANAGEMENT PROBLEMS ( α = 0.908). The scale uses a 5-point Likert scale(By adding a neutral option between the options of a four-point Likert scale, the original four-point Likert scale is transformed into a five-point Likert scale.) ranging from “very non-compliant” to “very compliant,” with a higher total score indicating a more severe degree of Internet addiction; If the total score is ≥ 58, the individual is preliminarily screened as a potential Internet addict; if the total score is score is ≥ 68, the individual is diagnosed as an Internet addict. The scale has high reliability and validity, with a retest reliability of 0.83 and a Cronbach’s coefficient of 0.976 for the total scale.

In this study, three indicateors of self-esteem, physical activity and Internet addiction were subjected to normality test using SPSS 21.0, results showed that p-values were greater than 0.05, indicating a normal distribution. Then, descriptive statistics, reliability test and Pearson correlation analysis [35] were performed. After the research variables were standardized, mediation effect analysis was performed through the standardized hierarchical multiple regression equations, taking the amount of physical activity as an independent variable, the self-esteem as the mediator variable, and Internet addiction as the dependent variable, and the incremental changes in R² and F values in the results were used to assess the main effects of the study variables. Finally, after standardizing each research variable, the significance of the mediation effect test was evaluated by bias-corrected nonparametric percentile Bootstrap method (5000 repetitive samples) using the macro program process in SPSS [36], and the absence of a confidence interval of 0 indicates significance (P < 0.05) [37]. Finally, the equation model conducted moderated mediation effect analysis using physical activity as the independent variable, gender as the moderator variable, self-esteem as the mediator variable, and Internet addiction as the dependent variable.

Data collection in this study was completed by self-assessment, which may introduce common methodological bias. Harman’s one-way test was used to factor analyze all items involved in this study. The results showed that a total of 11 factors with eigenvalues greater than 1 were extracted from the exploratory factor analysis, and the amount of variance explained by the first factor was 26.7%, much lower than the critical value of 40%. This indicateed that the data of this study were not affected by common method bias.

By taking gender as a grouping variable, the amount of physical activity, intensity of physical activity, frequency of physical activity, duration of physical activity, Internet addiction, compulsivity, withdrawal symptoms, tolerance, interpersonal health problems, and time management problems as test variables, the results of the independent samples t-test are shown in Table 1. In terms of physical activity, physical activity intensity, and physical activity frequency, the scores of men were significantly greater than those of women (all p < 0.01). In terms of physical activity time, the difference between genders was not significant (P > 0.05). In terms of Internet addiction, withdrawal symptoms, interpersonal health problems, and time management problems, the scores of female students were all significantly greater than those of male students (P < 0.01). Nonetheless, none of the differences between genders were significant (P > 0.05) in terms of compulsion and tolerance.

Using physical activity, self-esteem, and Internet addiction as variables, and age and grade as control variables, a biased correlation bivariate two-sided test shows (Table 2) that physical activity is significantly and positively correlated with self-esteem (r = 0.26, P < 0.01), and significantly and negatively correlated with Internet addiction (r=-0.23, P < 0.01), while self-esteem was significantly and negatively correlated with Internet addiction (r=- 0.22, P < 0.01). The significant correlation between the main variables suggests that mediation effects can be further inspected [27, 38].

Using physical activity as the independent variable X, Internet addiction as the dependent variable Y, self-esteem as the mediator variable M, and other factors as covariates, mediation effect regression analysis was conducted through Model 4 in the Process macro program. The test results in Table 3 showed that the negative predictive effect of physical activity on Internet addiction among college students was still significant (B = -0.14, with a 95% confidence interval of [-0.20, − 0.09], P < 0.01). In addition, the amount of physical activity can significantly and positively predict the degree of self-esteem (B = 0.22, 95% confidence interval of [0.17, 0.26], P < 0.01), and the degree of self-esteem can significantly and negatively predict the extent of Internet addiction (B = -0.15, 95% confidence interval of [-0.21, -0.09], P < 0.01). According to the mediating effect test criterion proposed by Wen Zhonglin [35] as well as the results of the study, it can be indicated that self-esteem is a mediating variable of physical activity which has effect on college students’ Internet addiction, playing a mediating role between the two.

As can be seen from Table 4, the direct path effect value of “physical activity → Internet addiction” is −0.04 and its 95% confidence interval is [−0.06, −0.02], which does not contain 0, indicating that the path effect is significant. Thus physical activity can directly affect college students’ Internet addiction, while the indirect path effect value of “physical activity → self-esteem → Internet addiction” is −0.15, and its 95% confidence interval is [−0.20, −0.10] which excludes 0, indicating that the path is also significant. The total indirect effect value is −0.19, and its 95% confidence interval [−0.21, −0.10] does not include 0, indicating that this path is also significant. Given the above, physical activity can not only directly reduce college students’ Internet addiction, but also indirectly affect their Internet addiction by moderating their self-esteem, where self-esteem plays a mediating role between physical activity and college students’ Internet addiction, with the mediating effect accounting for 78.95% of the totality, as shown in Fig. 3.

With physical activity as the independent variable X, Internet addiction as the dependent variable Y, self-esteem mediator variable M, gender as the moderator variable Z, and other factors as covariates, model 5 in the Process macro program was used to conduct the moderated mediating effect, and the results showed that the product of physical activity and gender significantly predicted Internet addiction (β = 0.181, t = 2.99, P < 0.01), suggesting that gender plays a moderating role between physical activity and internet addiction, as shown in Table 5.

To further explore the moderating role of gender in the relationship between physical activity and internet addiction among college students, a simple slope analysis was conducted. The results indicate that gender significantly moderates the relationship between physical activity and internet addiction, suggesting that the relationship pattern differs between male and female groups. When the level of physical activity is low, male students tend to have slightly higher levels of internet addiction than females. However, as the level of physical activity increases, the internet addiction of male students (simple slope = −0.17, t = −5.08, P < 0.001) decreases at a faster rate compared to female students (simple slope = −0.10, t = −2.6, P < 0.05), indicating that male students’ internet addiction significantly decreases in comparison to females when the level of physical activity is higher. The results are shown in Fig. 4.

The present study shows that male college students generally score higher than female students in physical activity, which is consistent with the findings of Aubert et al. [39]. This result further validates the previous findings, all of which concluded that males are more inclined to participate in physical activities compared to females. Social role theory suggests that people play different roles in different social situations [40]. Traditionally, men are expected to play more energetic, adventurous, and competitive roles, while women are expected to stay quiet, refined, and soft. This role expectation may make men more inclined to participate in physical activities in order to fulfill the social expectations placed on them.

The present study finds that the scores of female students are significantly greater than those of male students in terms of Internet addiction, which is contrary to the study of Chen et al. [41], whose study also compared the differences in Internet addiction between genders using Chinese college students as the subject of the study, and all results showed that male students had significantly higher scores of Internet addiction than females. However, the present study is in line with the findings of Aylaz et al. (2016), which revealed that female high school students in Turkey had significantly higher Internet addiction scores than male students [42]. Additionally, Tang et al. (2014) showed that the prevalence of social network addiction among males and females was 27.8% and 37.3% respectively, with a significantly higher prevalence among females [43]. Reasons for this situation may be that, on the one hand, according to the statistical report on China’s Internet development, female Internet users accounted for 39.6% of the total number of Internet users in the country in 2004, indicating that the number of female Internet users was significantly lower than that of males in the early stage of the development of the Internet. However, by 2020, female Internet users accounted for 49% of total Internet users, and the proportion of male and female Internet users tended to be the same, which indicates that in the past 15 years, the number of female Internet users has been gradually increasing, with females engaging in online activities more frequently. This may be an important reason for the yearly rise of female Internet addiction symptoms. On the other hand, some studies have found that gender is a key factor explaining why individuals become addicted to the Internet in different ways [44]. Females have a greater preference for online interpersonal interactions and online shopping consumption than males [45]. In the past 15 years, China has seen a rise in online shopping consumption and social media platforms. Chinese women spend a lot of time interacting with friends and family on newly emerging social applications such as WeChat or TikTok [46], and also make online purchases through e-commerce applications such as Amazon and Alibaba [46]. These various online behaviors may lead to the increasing likelihood of Internet addiction among female college students. Additionally, this could be closely related to the personality traits of male students. Research has found that compared to females, males tend to be more open-minded and optimistic, while females are more prone to experiencing depression [47, 48]. According to the theory of needs satisfaction, when adolescents’ psychological needs are not met in their real lives, they are more likely to seek happiness, comfort, and fulfillment in the virtual world of the internet, leading to a vicious cycle of addiction [49]. Therefore, when male students have higher levels of physical activity, and they are aware of their interests and hobbies, they are more likely to resist the temptations of online gaming and other aspects of the internet, resulting in a lower risk of problematic internet behavior compared to females, which is consistent with previous research.

The results of this study showed that physical activity negatively predicts Internet addiction. This aligns with a meta-analysis conducted by Wu Jin [12, 13], which includes a large number of RCTs of exercise intervention for Internet addiction. The final results indicate that physical activity intervention can effectively reduce the level of Internet addiction in adolescents, which not only improves the physical condition of addicts, but also helps to enhance self-esteem, hone the will, and resistance to Internet addiction. Both the Internet and physical activity provide participants with a sense of interaction and entertainment. The difference lies in the fact that the participatory use of the Internet immerses individuals in the virtual world it brings, whereas physical activity allows participants to experience the real world by guiding and building their physical and mental well-being. One of the theoretical underpinnings of exercise interventions is that engaging in physical activity occupies the time spent on the Internet, while at the same time subconsciously brings positive physiological and psychological changes to the participant. Therefore, adolescents who are physically active on a regular basis usually have good physical and mental health, and healthier Internet use behaviors [50].

In this study, self-esteem is significantly positively correlated with physical activity and significantly negatively correlated with Internet addiction. Furthermore, self-esteem partially mediates the relationship between physical activity and Internet addiction among college students, accounting for 78.95% of the total effect. This suggests that physical activity can reduce college students’ Internet addiction index by improving individual self-esteem, in which self-esteem exists as a mediating factor. The positive effects of physical activity on the body and mind have been confirmed [51], and individuals who often engage in physical activity have higher level of self-esteem and positive psychological emotions and states [52], while those with low self-esteem are often prone to fall into doubt about their own abilities and may turn to the Internet as a way of hiding from the reality and seeking psychological compensation [53]. Physical activity has the effect of shaping the human body, improving mental image, and promoting the health of the individual, which may have a positive effect on individual’s well-being, and thus enhance self-esteem. This finding is also consistent with the results of Sun Chao’s study [17]. For another, in Kumer’s study, a survey on self-esteem and Internet addiction among 200 college students, found that those with low self-esteem have a higher degree of Internet addiction, and those with high self-esteem have a lower tendency to Internet addiction, indicating a significant negative correlation between self-esteem and Internet addiction [20]. Therefore, after self-esteem is raised, individuals may have more time and energy to present themselves and participate more in activities such as socializing, which further reduces the symptoms of Internet addiction. If individuals are less involved in physical activities, their health status and physical shape may be unsatisfactory, affecting the level of their self-esteem. Therefore, they may reduce their social interaction and find self-satisfaction in the virtual network, leading to the increase of Internet addiction.

First, in this study, the measures of physical activity, self-esteem, and Internet addiction were taken from the self-assessment reports of college students, which may result in a slight deviation from the actual situation. However, the study tried to ensure standardized testing procedures, and removed as many invalid questionnaires as possible after the tests were completed, minimizing the bias and maintaining the reliability of the findings. Second, sampling bias may exist in the study. The random distribution of the questionnaire resulted in a biased sample in terms of gender, with more males and fewer females, as well as uneven differences in place of origin, grade, age, etc. Consequently, a sampling bias that may have had some impact on the external validity of the results. Thirdly, this study only used cross-sectional research without longitudinal tracking to explore the changing mechanism of influencing factors on Internet addiction among college students, which may be considered as a future research direction. Additionally, a small R2 value may be due to insufficient sample size or other factors. However, the model’s p-value is less than 0.01, indicating that the model is highly significant. Subsequent research will aim to expand the sample size for further investigation.