Ego depletion theory assumes that a person’s capacity for self-control is limited. When individuals engage in self-control behaviors and do not immediately replenish this resource, it may be temporarily depleted [17]. Once exhausted, it impairs performance on self-control tasks [18]. Owing to brain resources being depleted and taking time to recover, a failure of self-regulation may occur. Therefore, ego depletion is used to describe the self-regulation failure and impaired task performance observed during this recovery period [19]. When faced with a conflict between dual motivations, individuals may experience self-depletion. Consequently, they must suppress one motivation in order to satisfy the other, either by overriding their desire to act to do something (inhibitory response) or not to do something (amplification response).

Ego depletion is one of the most reliably replicated phenomena in social psychology. Its relevance extends across many areas due to the importance of self-control for various desirable behaviors. This has led to the theory’ s application in decision-making, planning, and initiative [20]. Moreover, ego depletion influences how individuals learn and what content they absorb [21]. When self-control is depleted, it can hinder a successful academic career [17]. In summary, this theory should adequately explain the correlation between TikTok addiction and SEL. Therefore, this study explored the literature through the lens of ego depletion theory, which informs the research hypotheses and hypothesized model.

The direction of the relationship between addictive social media use and social-psychological issues remains unclear. Moreover, there is no direct evidence linking TikTok addiction to SEL. However, there are still indications from the results of past related studies. For example, Xiao et al. suggested that excessive use of the internet may lead to the development of problems [22]. According to their argument, changes in adolescents’ social media interactions and time spent online are driving the increase in psychosocial issues. Dhingra and Parashar also highlighted a growing body of evidence showing that social media can enhance social connections, provide emotional support during difficult times, and serve as a source for discovering valuable information and learning [10]. It can also illustrate the acquisition of perspectives. However, they noted that there is only a thin line between using and abusing the media. It is reasonable to believe that richer and denser social interactions lead to more information and emotional support. This also enhances the internalization of social world diversity. However, both users and media are diverse and are influenced by usage practices [23].

Based on the above literature, the hypotheses about the two types of TikTok addiction, short video addiction and live content addiction, and the five dimensions of SEL are as follows:

Hypothesis 1 (H1): Video content addiction is negatively correlated with self-awareness;

Hypothesis 2 (H2): Video content addiction is negatively correlated with self-management;

Hypothesis 3 (H3): Video content addiction is negatively correlated with relationship skills;

Hypothesis 4 (H4): Video content addiction is negatively correlated with responsible decision-making;

Hypothesis 5 (H5): Video content addiction is negatively correlated with social awareness;

Hypothesis 6 (H6): Video content addiction is negatively correlated with self-awareness;

Hypothesis 7 (H7): Live streaming content addiction is negatively correlated with self-management;

Hypothesis 8 (H8): Live streaming content addiction is negatively correlated with relationship skills;

Hypothesis 9 (H9): Live streaming content addiction is negatively correlated with responsible decision-making;

Hypothesis 10 (H10): Live streaming content addiction is negatively correlated with social awareness.

Short video addiction (SVA) is a potential behavioral addiction that leads to negative health and social consequences [46]. Based on ego depletion theory and relevant literature, although TikTok serves social functions, addicted students may prioritize its entertainment aspects. This shift can diminish its socio-emotional benefits. Moreover, students engrossed in entertainment may reduce their engagement in learning essential competencies, including self-awareness, self-management, social awareness, relationship skills, and responsible decision-making. The research model is shown in Fig. 1.

After obtaining ethics approval from the academic committee of the co-authors’ institution (the Academic Committee of the School of Education, Guangxi University of Foreign Languages; Approval number: GXUFL-SE-24016), we conducted data collection using a cross-sectional design. We administered an online survey on the Wenjuanxing platform and employed a snowball sampling method to recruit students from Chinese vocational colleges. The survey period spanned from 6 December 2024, to 3 January 2025. A total of 665 questionnaires were collected. After excluding 102 responses completed in less than 3 min (with the standard derived from the average time of completing the answers), 563 valid responses remained. This resulted in an effective response rate of 84.66%. Participants had an average age of 20.20 years (SD = 1.36). Additional participant details are provided in Table 1.

The questionnaire used in this study was adapted from an established instrument with proven reliability and validity. We translated and revised the instrument by replacing the vague term “short video” in the Short Video Addiction Scale with the more specific labels “short-video content” and “live-streaming content.” We also expanded several items on the Social-Emotional Learning Scale to reflect the Chinese context. Two experts in Internet addiction research and three educators then reviewed the revised questionnaire draft to ensure wording accuracy and confirm its cultural appropriateness. The questionnaire used a 5-point Likert scale, where 1 = “strongly disagree,” 2 = “disagree,” 3 = “neutral,” 4 = “agree,” and 5 = “strongly agree.”

After excluding invalid questionnaires with excessively short completion times, we conducted a series of analyses using SPSS 23 (IBM Corp., Armonk, NY, USA). First, we tested for common method bias to ensure that no systematic error influenced our findings. Next, we examined the reliability and validity of each scale to confirm the integrity of all variables. We calculated descriptive statistics—means, standard deviations, and so on—and evaluated the correlations among variables. To validate our measurement and structural models, we employed AMOS 26 (IBM Corp., Armonk, NY, USA), assessed factor loadings and fit indices, and confirmed that the model met acceptable criteria. Finally, we performed path analysis to evaluate the ten research hypotheses and the overall hypothesized framework.

This study used principal component analysis with Varimax rotation. Seven factors emerged with eigenvalues exceeding 1, and the cumulative variance explained was 25.30% after rotation. According to Podsakoff and Organ, if a single factor explains less than 50% of the variance in exploratory factor analysis, severe common method bias is unlikely [49].

Before running the structural equation model, we first examined the fit of the measurement model. Hair et al. and Kenny et al. recommended that the chi-square/degrees of freedom (χ²/df) should be below 5, the root mean square error of approximation (RMSEA) should be below 0.10, and both goodness of fit index (GFI) and adjusted goodness of fit index (AGFI) should be above 0.80, while factor loadings should exceed 0.5 [50,51]. Table 2 shows the results. According to the criteria, we removed items as follows: the video content addiction scale was reduced from 10 to eight items; the live streaming content addiction scale was also reduced from 10 to eight items; the self-awareness scale was reduced from eight to six items; the self-management scale was reduced from seven to six items; the relationship skills scale was reduced from eight to six items; and the social awareness scale was also reduced from eight to six items.

Assessing discriminant validity is a widely accepted prerequisite for examining relationships among latent variables [52]. According to the Fornell-Larcker criterion, discriminant validity is established if the correlations among variables (or dimensions) are lower than the square root of the AVE. As shown in Table 3, all variables in this study demonstrated adequate discriminant validity.

We evaluated the overall model fit using AMOS. Hair et al. and Abedi et al. recommend that the χ²/df value should be below 5, the RMSEA below 0.10, and the indices GFI, AGFI, NFI, NNFI, CFI, IFI, and RFI above 0.80, while PNFI and PGFI should exceed 0.50 [50,53]. In this study, the fit indices were: χ² = 1627.78, df = 1002, χ²/df = 1.63, RMSEA = 0.03, GFI = 0.89, AGFI = 0.88, NFI = 0.93, NNFI = 0.97, CFI = 0.97, IFI = 0.93, RFI = 0.93, PNFI = 0.87, and PGFI = 0.79.

We verified the model using AMOS. The results revealed that: Video content addiction was negatively correlated with self-awareness (β = −0.38, p < 0.001); video content addiction was negatively correlated with self-management (β = −0.48, p < 0.001); video content addiction was negatively correlated with relationship skills (β = −0.39, p < 0.001); video content addiction was negatively correlated with responsible decision-making (β = −0.42, p < 0.001); video content addiction was negatively correlated with social awareness (β = −0.35, p < 0.001); live streaming content addiction was negatively correlated with self-awareness (β = −0.44, p < 0.001); live streaming content addiction was negatively correlated with self-management (β = −0.30, p < 0.001); live streaming content addiction was negatively correlated with relationship skills (β = −0.31, p < 0.001); live streaming content addiction was negatively correlated with responsible decision-making (β = −0.32, p < 0.001); and live streaming content addiction was negatively correlated with social awareness (β = −0.38, p < 0.001).

Cohen’s f² allows researchers to assess the local effect size of a single predictor in a multiple regression model [54]. We calculated f ² using Cohen’s formula, f ² = (R²_AB−R²_A/1−R²_AB). Conventionally, f ² values of 0.02, 0.15, and 0.35 indicate small, medium, and large effects, respectively [55]. Furthermore, video and live streaming addiction together explained 62% of the variance in self-awareness (f ² = 0.54), 56% in self-management (f ² = 0.48), 48% in relationship skills (f ² = 0.54), 50% in responsible decision-making (f ² = 0.48), and 49% in social awareness (f ² = 0.48). As shown in Fig. 2, TikTok addiction explained the greatest variance in self-awareness and the least in social awareness. These findings suggest that TikTok addiction may significantly influence social-emotional learning, especially self-awareness and relationship skills.

This study found that both short video and live streaming addiction were negatively correlated with self-awareness, it also means that the assumptions of H1 and H6 are supported. In other words, the more severe a user’s addictive behavior on TikTok, the poorer the development of their self-awareness. TikTok’s rich and varied content readily immerses students in video viewing. They often become so absorbed that they neglect self-reflection and self-monitoring. Over time, this neglect undermines their self-awareness. Irudayasamy stated that addictive behaviors are considered destructive and their negative impact on the emotional and social well-being of individuals cannot be ignored [6]. Cedrún and Civila reported that over 50% of TikTok users are influenced by the platform’s content, absorbing its values and changing their personal views [9]. This finding reveals the subtle influence of short video platforms on users’ thought processes. Furthermore, Shasizadeh et al. demonstrated a significant negative relationship between smartphone addiction and students’ emotional self-awareness [24]. Relevant studies have also shown that social media addiction exerts profound neurological and psychological effects, harming students’ cognitive abilities and potentially causing long-term negative consequences for their emotional development [6].

This study indicated that both video content addiction and live streaming content addiction were negatively correlated with self-management, it also means that the assumptions of H2 and H7 are supported. In other words, the more pronounced the addictive behaviors of TikTok users, the less effective their self-management development becomes. This is because addicted TikTok users often struggle to regulate their short-video app usage, and this self-control failure further impairs their self-management capacity. This finding is consistent with Yilmaz and Karaoglan Yilmaz, who reported a moderate negative correlation between problematic internet use and both self-control and self-management [27]. Moreover, Ye et al. indicated that short video addiction is often a consequence of trait self-regulation failure [8]. Therefore, addicts tend to exhibit higher levels of impulsivity, risky decision-making, and biased reasoning compared to non-addicts [29]. Meanwhile, through ego depletion theory, Mu et al. proposed that applications such as short video platforms may distract users, thereby diminishing their self-control capacity [28]. Collectively, these findings underscore the negative impact of addictive behaviors on individuals’ self-management and self-control abilities.

This study confirmed that both video content addiction and live streaming addiction were negatively correlated with relationship skills, it also means that the assumptions of H3 and H8 are supported. In other words, the more pronounced the addictive behaviors exhibited by TikTok users, the poorer their ability to develop effective relationship skills. This is because excessive reliance on TikTok for social interaction leads students to neglect face-to-face communication in real life, thereby undermining their interpersonal skills. This finding is supported by Camilleri, who noted that interactions conducted via social media networks often lack the authenticity and satisfaction inherent in face-to-face exchanges [32]. Furthermore, Liao also mentioned that the social model promoted by short video platforms caters to users’ tendencies to evade the high level of interpersonal skills required in real-life interactions [33]. Meanwhile, Mu et al. demonstrated that short videos allow users to immerse themselves in the platform and avoid awkward situations caused by social phobia [28]. These findings underscore the character of short video platforms as tools for evading genuine interpersonal engagement, suggesting that short video addiction may exacerbate users’ social difficulties [33].

The results of this study indicated that both video content addiction and live streaming addiction were negatively correlated with responsible decision-making, it also means that the assumptions of H4 and H9 are supported. This result is due to the fact that problematic use of TikTok itself can be seen as a form of poor self-management. When TikTok users exhibit severe addictive behaviors, they struggle to develop responsible decision-making skills. Liao noted that the platform continuously alters users’ viewing behavior through its design mechanisms [33]. This process reinforces dependency and weakens critical thinking. Consequently, addictive behaviors adversely affect decision-making [35]. Bechara further showed that addicted individuals often deny or fail to recognize their problems [37]. When facing risks that could lead to long-term negative consequences, they tend to choose immediate rewards over future benefits. Racine et al. highlighted that addiction impairs an individual’s freedom to choose among different options [36]. We attribute this impairment to weakened neural mechanisms for reflection and decision-making, which shift behavior from self-directed control to automated, sensation-driven responses [35].

This study confirmed that both video content addiction and live streaming addiction were negatively correlated with social awareness, it also means that the assumptions of H5 and H10 are supported. In other words, as TikTok users’ addictive behaviors intensify, their ability to develop social awareness declines. Excessive use of TikTok for entertainment or personal interests traps student users in an information cocoon. As a result, they overlook social events, public issues, and others’ needs, which limits their understanding of social diversity and complexity. Educators and experts can use TikTok in engaging ways to raise public awareness of important social issues [56]. However, social media use still poses significant risks [42]. Despite the fact that responsible social media use contributes to increased social awareness and concern for others, research has found a correlation between increased frequency of short video use and increased social isolation among adolescents [45]. As short videos gain more influence, people shift their social interactions toward the integration of virtual reality. This transition raises concerns about potential socio-psychological issues [43]. In addition, Hsieh et al. supported these concerns by finding that Internet addiction is negatively associated with social awareness [44]. Their study indicated that excessive reliance on the Internet may weaken individuals’ social cognition and interpersonal skills.

Social media continually innovates and offers people the chance to better understand the world around them. It enables people to socialize, build connections, share information, and learn. Short-video platforms such as TikTok prioritize user experience. Their product mechanisms and functionalities deliver the commercial results that companies expect. However, evidence based on ego depletion theory raises serious concerns. We must carefully consider how to effectively reduce TikTok addiction or problematic short video use. This study distinguished two main types of TikTok addiction based on short videos. This classification allowed us to explore TikTok addiction issues in greater detail. In addition, the instrument developed in this study can also help to examine the rapidly growing issue of live streaming addiction.

Furthermore, TikTok’s video quality and creator competence vary widely. The platform should strengthen its AI screening and human review processes to prevent gradual exposure to distorted worldviews, values, employment outlooks, and career perspectives that may harm users’ physical and mental development. At the same time, the platform should also provide alert notices for excessive use and enforce usage restrictions for minors to prevent students from becoming addicted to TikTok and other short video platforms. Moreover, managing leisure time effectively is crucial. We must cultivate students’ self-leadership, self-management skills, and media literacy. In addition, according to self-depletion theory, teaching self-regulation strategies is an effective measure to prevent addiction. Student users should be trained to accumulate self-resources, enabling better self-monitoring and control over short video consumption habits.

SEL is considered one of the keys to success in a student’s career. Teachers and parents should actively encourage responsible decision-making and self-management. They should help students develop social awareness and relationship skills. This support would enable students to build and maintain healthy interpersonal interactions. Improving self-awareness lets students monitor and reflect on their thoughts and actions. Over time, this process will shape them into global citizens with strong social and emotional competencies. We must also pay attention to the way in which students use social media and the content they watch to ensure that these practices do not undermine the effectiveness of their SEL.

SEL can help students use short videos more effectively by guiding them through five key steps. First, it guides them to reflect on and record their motivations and emotions when they watch short videos. Second, it teaches them alternative emotion-regulation strategies for coping with anxiety or boredom instead of immediately opening video apps. Third, it helps them recognize that face-to-face interactions provide a deeper emotional exchange than virtual content. Fourth, it educates students about the risks of excessive viewing and trains them to plan their screen time responsibly. Finally, it encourages them to align their viewing choices with their interests and values by selecting knowledge-based content, which naturally reduces time spent on purely entertaining videos.

Our study provides evidence that TikTok addiction negatively affects learning. However, several limitations warrant further investigation.

First, our reliance on self-report surveys lacks objective data for comparison. Future studies should incorporate objective measures (such as app-recorded usage time and physiological indicators) to validate our results. At the same time, the cross-sectional survey method collects data at a single time point, preventing observation of variable changes over time. Consequently, it cannot determine the temporal sequence of variables, hindering causal inferences between TikTok addiction and SEL. We should therefore employ longitudinal and experimental designs that collect data at multiple time points to clarify the directionality of these associations so as to yield more robust conclusions. Additionally, researchers must remain vigilant about potential sample attrition in longitudinal studies. Therefore, beyond increasing sample size, it is essential to recruit participants with a high willingness to participate. Although we confirmed that both types of TikTok addiction negatively correlate with SEL, the underlying mechanisms remain unclear. Future research should employ qualitative methods such as interviews or network ethnography to further explore these causes.

Second, to reduce social desirability bias, we encouraged students to repost the survey on social media to boost data collection. This practical approach may introduce selection bias and limit our findings’ generalizability to all Chinese vocational college students. Future research should adopt diverse sampling strategies so that sample characteristics better reflect the target population and strengthen the inferential power of the results.

Third, emerging social media addiction, such as TikTok addiction, adversely affects students’ health, well-being, academic performance, and non-cognitive skill development. TikTok’ s influence continues to expand. It affects not only adolescents but also young children and older adults to varying degrees. Future studies should examine these differences across age groups and regions.

Fourth, in addition to the mechanism of TikTok, the content is also an important factor in causing addiction. Future research can classify content by themes such as short videos, micro-dramas, or live streams to determine which types most easily trigger addictive behaviors. Moreover, although our study explored live streaming addiction, this emerging phenomenon can also bring about the information cocoon effect, high levels of adhesion, overuse, irrational rewards, impulsive consumption, and other problems. There is still a lack of research on the causes and effects of live content addiction. Therefore, more evidence and data are urgently needed to raise awareness of the potential dangers of excessive live stream viewing. It can be explored as a topic in subsequent studies. Clearly, the overall impact of short video platforms, including TikTok, has not been fully examined. More research is needed to understand both their benefits and drawbacks.

Fifth, this study focused on individual-level links between addiction and SEL. External factors such as institutional policies, educational support systems or platform governance may moderate these links. Future research can apply ecosystem theory to explore how these factors buffer or amplify TikTok addiction’s negative impact on students’ SEL development. The interplay between addiction and platform use also warrants further study. For instance, TikTok’s ease of use, convenience and familiarity may reinforce addictive behavior. Thus, future work should consider users’ prior technological experience as a key factor.