How balanced time perspective affects short-video addiction: a chain mediation model based on trait anxiety and ego depletion

Abstract

Background

Previous studies have identified balanced time perspective as a protective factor against short-video addiction among adolescents. However, the underlying psychological mechanisms linking these two constructs remain insufficiently explored. Guided by the cognition-emotion-behavior theory, this study aimed to investigate the mediating mechanisms through which balanced time perspective influences short-video addiction.

Methods

Two studies were conducted: a cross-sectional study (Study 1, N = 689; 54.3% male, mean age = 14.40 years, SD = 2.11) and a two-wave longitudinal study (Study 2, N = 470; 50.0% male, mean age = 14.23 years, SD = 1.01). In Study 1, a multiple mediation model was employed to examine the mediating role of trait anxiety and ego depletion in the relationship between balanced time perspective and short-video addiction. In Study 2, a cross-lagged model was used to verify the longitudinal mediating chain involving balanced time perspective, trait anxiety, ego depletion, and short-video addiction.

Results

Findings from both studies consistently supported the proposed hypotheses. Study 1 found a serial mediation effect of trait anxiety and ego depletion (total indirect effect = -0.2159, 95% CI [-0.2771, -0.1578], 78.52% of total effect). Study 2 confirmed the longitudinal chain.

Conclusions

This study reveals a cognition-emotion-resource-behavior mechanism underlying the association between balanced time perspective and short-video addiction. It enriches the existing research on short-video addiction prevention from the perspective of time psychology and provides theoretical implications for developing targeted intervention strategies to reduce adolescent short-video addiction.

Introduction

Against the backdrop of the digital wave, short videos have rapidly permeated people’s daily lives due to their fragmented, highly immersive, and interactive features [1], thereby giving rise to typical addictive behavioral patterns. As a new form of Internet addiction, short-video addiction refers to a series of psychological and behavioral manifestations in which individuals gradually develop a chronic or periodic state of indulgence through long-term and habitual use of short-video platforms such as Douyin and Kuaishou, accompanied by persistent and intense craving and behavioral dependence [2]. Short-video addiction exerts multifaceted negative impacts on the development of adolescents, including hindering academic achievement [3], exacerbating depression and anxiety [4], and impairing sleep quality [5]. The problem of addiction caused by excessive short-video use has increasingly become a major concern in the field of public health. Therefore, revealing the psychological mechanisms of short-video addiction and clarifying its influencing factors and mediating pathways are of great theoretical and practical significance for scientifically understanding short-video addiction and further developing effective intervention strategies.

Among the numerous psychological constructs that explain addictive behaviors, time perspective is regarded as a core cognitive framework. Time perspective (TP) refers to the cognitive, experiential, and behavioral tendencies towards the past, present, and future [6]. Balanced Time Perspective (BTP), defined as an ideal state, enables individuals to flexibly switch among different time orientations according to situational demands, thereby achieving optimal psychological functioning [7]. A person with a balanced perspective on time usually possesses flexible time focus [8]. Zimbardo distinguished five time orientations in the Zimbardo Time Perspective Inventory (ZTPI) [9], and the relationship between these five time orientations and addictive behaviors.

Based on the self-regulation theory and the limited resource model, this study proposes that trait anxiety and ego depletion may serve as two key mediating variables connecting the imbalance of time perspective with short-video addiction. As a stable personality trait, trait anxiety refers to an individual’s tendency to perceive threats and experience tension and anxiety in different situations [10]. People with an imbalanced time perspective—especially those with a weak future orientation and unclear planning, or those with a strong fatalistic present orientation and a belief that effort is futile—are more likely to be in a long-term state of high anxiety due to a lack of goals and a sense of control [11]. According to the self-consumption theory, all self-control activities (including emotion regulation and impulse inhibition) share the same limited pool of cognitive resources [12]. Long-term high trait anxiety means that individuals need to continuously consume psychological resources to regulate their internal emotions, which further leads to ego depletion, that is, a temporary decline in self-control ability [13, 14]. More specifically, ego depletion is a temporary state in which an individual’s performance on subsequent self-control tasks declines due to the consumption of limited psychological energy resources from previous self-control behaviors. People in a state of ego depletion have significantly reduced ability to resist immediate temptations (such as continuously scrolling to find the next recommended video) and adhere to preset goals (such as stopping use on time) [15].

This study adopts the I-PACE model [16] as the core theoretical framework. The I-PACE model explains specific Internet-use disorders through a multi-step path: “individual predisposing factors → affective/cognitive responses → reduced executive functioning → addictive behavior.” Within this framework, balanced time perspective serves as a cognitive predisposing factor, trait anxiety as an affective response, and ego depletion as reduced executive functioning, ultimately leading to short-video addiction. This chain mediation logic is further supported by Ellis’s ABC theory [17, 18] (cognition-emotion-behavior interaction) and Baumeister’s strength model of self-control, which are integrated as complementary explanations. Together, they form a coherent theoretical basis, guiding the chain mediation hypothesis of this study.

A possible chain pathway is: unbalanced time perspective → higher trait anxiety → greater Ego depletion → stronger short-video addiction tendency. As a low-threshold and highly immediate-rewarding stimulus, short videos can easily become a tool for individuals with Ego depletion to quickly relieve negative emotions or fill inner emptiness, thereby increasing habitual use and the risk of addiction [19].

Internet addiction and short-video addiction

From the perspective of the conceptual lineage, short-video addiction is one of the specific forms of generalized internet addiction. However, there are significant differences between the two: generalized internet addiction encompasses various behaviors such as web browsing, online gaming, social interaction, and information search [20], and the phenomenon of short-video addiction focuses on algorithm-driven, short-duration, and high-density consumption of video content [21]. From the perspective of diagnostic criteria, both share the core characteristics of behavioral addiction, including tolerance, withdrawal reactions, and functional impairment [22, 23]. However, from a mechanism perspective, the unique algorithmic recommendations and infinite scrolling design of short-video platforms have created an intermittent reinforcing addiction path [24], this gives time-related variables a special explanatory role in the phenomenon of short-video addiction. In recent years, the academic community has gradually shifted from the ‘generalized internet addiction’ to the study of ‘specific problematic use’ of specific online activities [24]. This study precisely focuses on the addictive mechanism of the specific behavior of short videos under this academic trend.

The mediating role of trait anxiety between balanced time perspective and short-video addiction

Trait anxiety, as a stable negative emotional tendency, may serve as a key emotional pathway connecting the temporal perspective and addictive behaviors. Davis [25]’s cognitive-behavioral model indicates that distant predisposing factors (such as personality traits) influence an individual’s emotional experience, thereby leading to pathological internet use (PIU). Numerous studies have shown that anxiety is an important predictor of internet addiction [26, 27]. Individuals with high anxiety tend to seek emotional relief through virtual environments, thereby increasing the risk of addiction [28, 29]. At the same time, temporal perspective imbalance (such as extreme fatalism now) is often accompanied by inefficient emotional regulation strategies, leading to an increase in anxiety levels [30].

Therefore, we propose the hypothesis that trait anxiety plays a mediating role in balancing the temporal perspective and short-video addiction.

The mediating role of ego depletion between balanced time perspective and short-video addiction

Ego depletion theory provides a cognitive-energetic explanation for this relationship. According to this theory, individuals have limited self-regulatory resources. After these resources are expended in prior self-control efforts, individuals enter a state of “Ego depletion,” resulting in reduced subsequent self-control capacity [31, 32]. Unbalanced time perspective (e.g., a strong present-hedonistic orientation) may make it more difficult for individuals to recruit self-control resources to resist the temptation of short videos during goal conflicts (e.g., needing to study instead of engaging in entertainment), thereby accelerating Ego depletion [33]. Individuals in a depleted state show significantly impaired ability to control impulsive behaviors (such as compulsive short-video scrolling), which in turn increases addictive behaviors [15]. Importantly, high trait anxiety may also intensify Ego depletion due to the persistent internal emotional consumption it entails [34]. This suggests that trait anxiety and Ego depletion may function sequentially, jointly forming a complete mediating chain from cognitive tendency to behavioral dyscontrol.

In summary, the following hypothesis is proposed: Ego depletion mediates the inhibitory effect of balanced time perspective on short-video addiction.

Chain mediation model

Our analysis indicates that trait anxiety and ego depletion may play a mediating role between the balance of time perspective and short-video addiction, but whether they are serial or parallel remains to be clarified. The dual-system self-control model indicates that individual decisions are jointly guided by the hot system (emotionally driven, preferring immediate gratification) and the cold system (cognitively driven, preferring delayed rewards) [35]. The final choice depends on the system used when pursuing the goal [36]. Individuals with high trait anxiety have higher emotional arousal, consume a large amount of self-control resources, inhibit the function of the cold system, and lead to ego depletion [13]. At the same time, anxious individuals are prone to negative emotional reactions, and their ability to inhibit interference is weakened, further exacerbating resource consumption. Therefore, trait anxiety may exacerbate ego depletion [37]. The two systems usually work collaboratively: anxious individuals’ hot system is easily activated, and they have a strong preference for immediate rewards [38], long-term negative emotions continuously consume self-regulation resources, making teenagers unable to control their phone usage impulses and more prone to addiction.

Based on this, this study has constructed a comprehensive theoretical model (as shown in Fig. 1), and proposes the hypothesis that trait anxiety and ego depletion play sequential mediating roles in the path of balancing the time perspective to inhibit short-video addiction.

Fig. 1.

Conceptual model of the effects of balanced time perspective on short video addiction mediated by trait anxiety and ego depletion

The present study

With the global popularity of short-video platforms, the risk of addiction has become an important topic in psychology. Previous studies have shown that past negative experiences, present hedonism or fatalism, and future time perspectives are risk factors for addictive behaviors, while future-oriented time perspectives have a protective effect [39–42]. An imbalance in time perspectives is closely related to various problem behaviors such as substance abuse and internet game addiction [43], but the specific relationship between balanced time perspectives and short-video addiction and its psychological mechanism still lack research [44]. Research has indeed shown that different dimensions of time perspective have distinct relationships with addictive behaviors: past negative and present fatalistic orientations are positive predictors of Internet addiction, while future orientation serves as a negative predictor. Based on these insights, the present study moves beyond examining single dimensions to investigate the role of balanced time perspective (BTP) [45]—a composite indicator that captures the adaptive integration of multiple temporal orientations [46].

Although trait anxiety and ego depletion have been confirmed to be related to addiction, the mediating roles of these two factors in the chain path of “balanced time perspective → trait anxiety → ego depletion → short-video addiction” have not been verified. Therefore, Study 1 adopts a cross-sectional mediation model to explore the direct and indirect effects of balanced time perspectives on short-video addiction; Study 2 constructs a cross-lagged panel model (CLPM) to further test the stability of this chain mediation. This study deepens the theoretical understanding of the mechanism of short-video addiction from the perspective of the interaction between time cognition and self-regulation resources, and provides empirical evidence for multi-dimensional interventions such as cognitive restructuring (adjusting time perspectives), emotion management (reducing trait anxiety), and resource recovery (alleviating ego depletion).

Considering Study 1 and Study 2, we propose the following hypotheses:

• H1: Balanced time perspective is associated with short-video addiction, mediated by trait anxiety and Ego depletion.

• H2: Trait anxiety and Ego depletion exert a cross-sectional and longitudinal serial mediating effect in the relationship between balanced time perspective and short-video addiction.

Study1

Methods

Participants and procedures

Cluster sampling was adopted to select 20 classes from junior and senior high schools in Jingzhou City, Hubei Province. Before collecting data, researchers fully explained the purpose of the study and obtained the oral consent of the students and their guardians. Questionnaires were distributed offline, and college students who had received specialized training served as the main testers. The tests were conducted in classes. Before the formal test, the main testers read the instructions aloud, using a unified set of instructions. They informed the participants of the purpose of the questionnaire survey, emphasized that the results would be kept strictly confidential and used only for academic research, and would not be used for commercial purposes. It took about 20 min for the participants to complete the questionnaire. The questionnaires were immediately collected after all participants had finished. A total of 730 students completed the questionnaire. Eventually, 689 valid questionnaires were retrieved (94.38%). Among them, there were 374 male students (54.3%) and 315 female students (45.7%). The average age of the participants was 14.40 years (SD = 2.11 years).

Assessment instruments

Balanced time perspective

Balanced time perspective was assessed using the Time Perspective Inventory (TPI) developed by Li and Lyu [47]. The scale consists of 28 items, scored on a 5-point Likert scale from 1 (“strongly disagree”) to 5 (“strongly agree”), with a total score ranging from 28 to 140. The scale has a 7-factor structure: Past Positive (items 1–4), Past Negative (items 5–8), Mindful Present (items 9–12), Present Hedonistic (items 13–16), Present Excessively Task-oriented (items 17–20), Future Positive (items 21–24), and Future Negative (items 25–28). The balanced time perspective score was calculated as the sum of standardized positive-oriented subscales (Past Positive, Mindful Present, Present Hedonistic, Future Positive) minus the sum of standardized negative-oriented subscales (Past Negative, Present Excessively Task-oriented, Future Negative). Higher BTP values indicate a more adaptive and balanced time perspective. In this study, the Cronbach’s α for the full scale was 0.71, and the subscale alphas ranged from 0.73 to 0.93, consistent with previous validation studies.

Short video addiction

Adapted from Young’s [48] Internet Addiction Scale, there are eight items to assess participants’ perceived addiction level to short video use. The original scale was modified by replacing the Internet with short videos. A sample item is “How often do you find that you need to spend more and more time using short videos to get the same level of satisfaction?” The scale ranges from “Never” to “Always” on a 5-point scale ranges from 8 to 40, with a higher average score indicating a higher tendency towards short video addiction. The Cronbach’s alpha coefficient of this questionnaire is 0.769. An exploratory factor analysis was conducted on the adapted scale (KMO = 0.881, Bartlett’s test of sphericity p < 0.001), and a single factor explained 48.36% of the variance. The item loadings ranged from 0.471 to 0.849. The confirmatory factor analysis showed standardized loadings ranging from 0.543 to 0.736 (all p < 0.001), with model fit indices being CFI = 0.896, RMSEA = 0.117, and SRMR = 0.055. The composite reliability CR = 0.903, and the average variance extracted AVE = 0.542. These results indicate that the scale has acceptable structural validity. The adaptation method of this study referred to existing research [49, 50], further supporting the rationality of the scale adaptation.

Trait anxiety

The trait anxiety sub-scale of the State-Trait Anxiety Inventory (STAI-T) developed by Spielberger et al. was used. It consists of 20 descriptive questions, with questions 1, 3, 4, 6, 7, 10, 13, 14, 16, and 19 being scored in reverse order. It uses Likert 4-point scoring, ranging from “1—completely not” to “4—very obvious”, ranging from 20 to 80.The higher the score on this scale, the higher the trait anxiety level of the subject. The Cronbach’s alpha coefficient of this questionnaire is 0.867.

Ego depletion

Developed by Johnson et al. [51], assessing individuals’ perceived depletion of self-control resources in daily life. Although ego depletion is traditionally considered a state-like construct, chronic negative emotions (e.g., trait anxiety) can produce sustained resource depletion, making it measurable in cross-sectional designs. This approach has been widely adopted in recent studies [52]. The scale consists of 10 items(e.g., “I feel exhausted,” “My willpower is gone,” “I feel mentally drained”), rated on a 5-point Likert scale from 1 (strongly disagree) to 5 (strongly agree), with total scores ranging from 10 to 50 (higher scores indicate greater ego depletion). In this study, the scale demonstrated excellent internal consistency. The Cronbach’s alpha coefficient of this questionnaire is 0.942, and subscale alphas ranged from 0.73 to 0.91.

Data analysis

The data were subjected to descriptive statistics and correlation analysis using SPSS 27.0. The significance of the mediating effect was tested using the bias-corrected bootstrap method.

Results

Common method bias test

Using the Harman single-factor test, a factor analysis was conducted on all items of the balanced time perspective, short-video addiction, trait anxiety, and ego depletion variables. One factor was extracted [53]. The results showed that the explanatory rate of the first factor variable was 22.78%, which was much less than 40%. Therefore, there was no problem of common method bias in this study.

Descriptive statistics

Descriptive statistics and Pearson correlation analysis were conducted for each variable. As shown in Table 1, balanced time perspective was significantly negatively correlated with short-video addiction, trait anxiety, and ego depletion, while trait anxiety and ego depletion were significantly positively correlated with short-video addiction, and there was a significant positive correlation between trait anxiety and ego depletion.

Table 1.

Results of descriptive statistics and correlations among variables in study 1

	1	2	3	4	5	6
1.BTP	1
2.SVA	−0.287**	1
3.TA	−0.598**	0.345**	1
4.ED	−0.466**	0.376**	0.526**	1
5.age	−0.084*	0.163**	0.085*	0.06	1
6.gender	−0.053	0.132**	0.108**	0.078*	0.077*	1
M	30.34	2.44	46.26	12.20	14.40	1.46
SD	13.48	2.29	9.49	5.85	2.11	0.52

BTP balanced time perspective, TA trait anxiety, ED ego depletion, SVA short-video addiction, M mean, SD standard deviation

*p < 0.05,**p < 0.01,***p < 0.001, The following is the same

Analysis of mediating effects

All pairwise correlations among balanced time perspective, short-video addiction, trait anxiety, and ego depletion were significant. This met the statistical requirements for further mediation effect analysis between balanced time perspective and short-video addiction [54]. Based on the correlation analysis, the mediating roles of trait anxiety and ego depletion in the relationship between balanced time perspective and short-video addiction were further examined. Age was significantly correlated with balanced time perspective, short-video addiction, trait anxiety, and ego depletion, and gender was significantly correlated with short-video addiction, trait anxiety, and ego depletion; both were included as control variables in the subsequent regression analysis.

After controlling for the effects of age and gender, Model 6 of the SPSS macro PROCESS provided by Hayes (2013) was adopted, with balanced time perspective as the independent variable, short-video addiction as the dependent variable, and trait anxiety and ego depletion as chain mediators. The regression results (see Table 2) showed that balanced time perspective had a significant negative effect on trait anxiety, explaining 37.4% of its variance (R² = 0.374, F(3, 678) = 135.03, p < 0.001; β = –0.60, t = –19.64, p < 0.001). Trait anxiety, in turn, had a significant positive effect on ego depletion (R² = 0.314, F(4, 677) = 77.53, p < 0.001; β = 0.37, t = 9.29, p < 0.001). When predicting short-video addiction, the model accounted for 20.1% of its variance (R² = 0.201, F(5, 676) = 33.95, p < 0.001). Ego depletion positively predicted short-video addiction (β = 0.26, t = 6.20, p < 0.001), and trait anxiety also showed a positive effect (β = 0.16, t = 3.44, p < 0.001). However, the direct effect of balanced time perspective on short-video addiction was not significant (β = –0.06, t = –1.33, p = 0.184), indicating that its influence was fully mediated through the sequential path via trait anxiety and ego depletion. The total effect of balanced time perspective on short-video addiction was significant (r = –0.287, p < 0.001), consistent with the overall protective role of balanced time perspective.

Table 2.

Regression analysis of variable relationships

Outcome	Predictor	β	SE	t	p
TA	BTP	−0.599	0.030	−19.638	<.001
	Age	0.029	0.014	0.949	.343
	Gender	0.076	0.059	2.489	.013
R2 =.374,F(3, 678) = 135.026,p <.001
ED	BTP	−0.245	0.040	−6.110	<.001
	TA	0.374	0.040	9.293	<.001
	Age	0.002	0.015	0.062	.950
	Gender	0.026	0.062	0.802	.423
R2 =.314,F(4, 677) = 77.528,p <.001
SVA	BTP	−0.059	0.038	−1.330	.184
	TA	0.159	0.040	3.444	<.001
	ED	0.257	0.036	6.198	<.001
	Age	0.125	0.014	3.603	<.001
	Gender	0.075	0.058	2.159	.031
R2 =.201,F(5, 676) = 33.948,p <.001

All the variables in the model have been standardized, and the same applies hereinafter

BTP balanced time perspective, TA trait anxiety, ED ego depletion, SVA short-video addiction

Taken together, a significant chain mediating effect can be predicted: balanced time perspective reduced short-video addiction by first lowering trait anxiety, which in turn decreased ego depletion. The significance and effect size of this path demonstrated the important mediating roles of trait anxiety and ego depletion, which formed a sequential chain in the influence of balanced time perspective on addiction. The findings supported the chain mediation model of “trait anxiety → ego depletion → short-video addiction” and suggested that intervention strategies could focus on improving individuals’ balanced time perspective, alleviating anxiety, and reducing ego depletion to lower the risk of short-video addiction. Notably, the magnitudes of the paths differed: the direct inhibitory effect of balanced time perspective on trait anxiety was the strongest, followed by its direct inhibitory effect on ego depletion; both direct and indirect paths contributed significantly to addiction.

The mediation effect was tested using the Bootstrap method. The results showed that the direct effect of balanced time perspective on short-video addiction was not significant. The path through trait anxiety was significant, with an indirect effect of − 0.10 (95% CI [− 0.15, − 0.04]). The path through ego depletion was significant, with an indirect effect of − 0.06 (95% CI [− 0.10, − 0.03]). The chain mediation path through trait anxiety and ego depletion was also significant, with an indirect effect of − 0.06 (95% CI [− 0.08, − 0.03]). The total indirect effect was − 0.22 (95% CI [− 0.28, − 0.16]). These results supported the chain mediating role of trait anxiety and ego depletion in the relationship between balanced time perspective and short-video addiction (Table 3).

Table 3.

Analysis of the mediating effect of trait anxiety and ego depletion

Predicted path	Effect	BootSE	BooLLCl	BooULCl	Relative mediating effect
Total indirect effect	−0.2159	0.0305	−0.2771	−0.1578	78.52%
Balanced time perspective → Trait anxiety → Addiction to short videos	−0.0953	0.0281	−0.1537	−0.0436	34.63%
Balancing time perspective → Ego depletion → Addiction to short videos	−0.0630	0.0164	−0.0977	−0.0332	22.88%
Balanced time perspective → Trait anxiety → Ego depletion → Short-video addiction	−0.0577	0.0128	−0.0837	−0.0345	20.97%

The Boot standard error, as well as the lower and upper limits of BootCI, respectively refer to the standard error of the indirect effect estimated by the percentile Bootstrap method with bias correction, and the lower and upper limits of the 95% confidence interval; all values are rounded to four decimal places

Study2

Study 1 confirmed the correlation between balanced time perspective and short-video addiction, as well as the mediating role of trait anxiety and ego depletion. However, these results were all obtained using questionnaires at the same time. Therefore, we cannot rule out the influence of participants’ temporary state characteristics, such as emotions that occurred at that time or specific events. Hence, Study 2 constructed a cross-lagged model to examine the longitudinal associations among balanced time perspective, trait anxiety, ego depletion, and short-video addiction, further confirming the causal and sequential relationships between the variables.

Methods

Participants and procedures

The same offline, class-based survey procedure as in Study 1 was used. After obtaining the consent of the participants, a re-test was conducted on 689 participants six month after the end of Study 1 (from March 2025 to April 2025). The method for excluding abnormal data was the same as that in Study 1. After data cleaning, valid data were obtained from 470 participants (235 males (50.00%), 235 females (50.00%)), with a retention rate of 68.21%. The average age of the participants was 14.23 years (SD = 1.01).

Assessment instruments

All the assessment tools are consistent with Study 1. For the 730 valid questionnaires in the T1 sample, the Cronbach’s α coefficients for each questionnaire were 0.709 (balanced time perspective), 0.867 (trait anxiety), 0.942 (ego depletion), and 0.769 (short-video addiction). The Cronbach’s α coefficients for the T2 sample were 0.956 (balanced time perspective), 0.869 (trait anxiety), 0.832 (ego depletion), and 0.882 (short-video addiction).

Longitudinal measurement comparability– Because each construct was assessed using a single total score (scale sum or mean) rather than multiple item-level indicators, the traditional item-level longitudinal invariance tests (configural, metric, and scalar invariance) could not be performed. To support cross-time comparability of the constructs, test–retest correlations of the total scores were examined (see Table 4). As shown in the diagonal of Table 4, all test–retest correlations were moderate-to-high stability and statistically significant, ranging from 0.463 to 0.674 (all p < 0.001). Following established guidelines [55], the observed correlations provide evidence of substantial temporal stability. Thus, the constructs can be meaningfully compared across the two measurement waves.

Table 4.

Descriptive statistical results and correlations among variables in the longitudinal study

	1	2	3	4	5	6	7	8	9	10
T1
1.Balanced time perspective	1
2.Trait anxiety	-.680**	1
3.Ego depletion	-.468**	.488**	1
4. short-video addiction	-.199**	.177**	.226**	1
T2
5.Balanced time perspective	.674**	-.495**	-.360**	-.235**	1
6.Trait anxiety	-.572**	.610**	.451**	.316**	-.640**	1
7.Ego depletion	-.372**	.360**	.569**	.352**	-.503**	.558**	1
8. short-video addiction	-.107*	.162**	.215**	.463**	-.176**	.247**	.269**	1
9.Age	-.206**	.218**	.172**	.137**	-.213**	.212**	.157**	.130**	1
10.Gender	.024	.033	.055	.059	-.066	.111*	.115*	.083	-.094*	1
M	30.13	49.54	12.12	2.5	29.6	45.56	11.78	2.52	14.23	1.50
SD	14.52	10	5.9	2.55	15.79	10.18	5.85	3	1.01	0.50

^*p < 0.05, ^**p < 0.01 (two-tailed)

Data analysis

Two models were employed to determine whether trait anxiety and ego depletion longitudinally and sequentially mediated the relationship between balanced time perspective and short-video addiction. Similar to Study 1, age and gender were controlled as covariates. Descriptive statistics and correlation analyses were conducted using SPSS 27.0, and longitudinal CLPM was constructed using Mplus 8.0. The longitudinal mediating pathway was also tested by using bootstraping with 1000 bootstrap samples.

Results

Descriptive statistics

Table 4 presents the mean values, standard deviations, and correlations of the variables at T1 and T2. Balanced time perspective (BTP) was significantly negatively correlated with ego depletion (ED) and short-video addiction (SVA) at both T1 and T2 (rs ranging from –0.503 to –0.176, all p < 0.01). Trait anxiety (TA) was positively correlated with ED and SVA (rs = 0.162–0.610, p < 0.01). ED was positively correlated with SVA at T1 (r = 0.226) and T2 (r = 0.269, both p < 0.01). Test–retest correlations were high (rs = 0.463–0.674, p < 0.001). Among the covariates, age was negatively correlated with balanced time perspective at T1 (r = –0.206,p < 0.01) and T2 (r = –0.213, p < 0.01), and positively correlated with short-video addiction at T1 (r = 0.137,p < 0.01) and T2 (r = 0.130, p < 0.01). The remaining correlations were not significant.

Longitudinal model analysis

We constructed a cross-lagged panel model (CLPM, Model 1) including two waves of data (T1 and T2) for balanced time perspective, trait anxiety, ego depletion, and short-video addiction (Table 5).

Table 5.

Balancing temporal insight in the chain mediation model for the addictive effect of short videos

Predicted path	Effect	BootSE	BootLLCI	BootULCI	Relative mediation effect
Total indirect effect	−0.038*	0.014	−0.069	−0.014	100%
BTP (T1) → TA (T2) → SVA (T2)	−0.001	0.001	−0.003	0.001	2.6%
BTP (T1) → ED (T2) → SVA (T2)	−0.022*	0.010	−0.044	−0.005	57.9%
BTP (T1) → TA (T2) → ED (T2) → SVA (T2)	−0.006*	0.002	−0.011	−0.002	15.8%

Abbreviations: BTP balanced time perspective, TA trait anxiety, ED ego depletion, SVA short-video addiction

*p <.05 (bias-corrected bootstrap 95% CI does not contain zero)

First, we controlled for the autoregressive effects of each variable (i.e., prediction from Time 1 to Time 2). The results showed that balanced time perspective had high temporal stability (β = 0.532, p < 0.001), while trait anxiety (β = 0.499, p < 0.001), ego depletion (β = 0.444, p < 0.001), and short-video addiction (β = 0.384, p < 0.001) showed moderate temporal stability. After removing non-significant paths related to control variables, the final cross-lagged model demonstrated good fit: χ² = 34.184, df = 6, p < 0.001; CFI = 0.973, TLI = 0.902; RMSEA = 0.101; SRMR = 0.043.

As shown in Fig. 2, after controlling for age and gender, balanced time perspective at T1 significantly negatively predicted trait anxiety at T2 (β = − 0.183, 95% CI [− 0.276, − 0.079]), supporting the long-term inhibitory effect of time perspective on anxiety and the hypothesis that temporal cognitive coordination reduces perceived uncertainty. Trait anxiety at T1 significantly positively predicted ego depletion at T2 (β = 0.189, 95% CI [0.081, 0.304]), consistent with the cumulative effect of emotional distress on mental resource consumption as proposed by self-control resource theory. Ego depletion at T1 significantly positively predicted short-video addiction at T2 (β = 0.177, 95% CI [0.123, 0.360]), supporting the hypothesis that resource depletion directly leads to behavioral dyscontrol and verifying that addictive behavior results from failures in self-control and regulation.

Fig. 2.

Longitudinal prediction model of the influence of balanced time perspective on short-video addiction

Meanwhile, trait anxiety at T1 significantly negatively predicted balanced time perspective at T2 (β = − 0.532, 95% CI [− 0.332, − 0.136]), indicating a reciprocal relationship between balanced time perspective and trait anxiety.

Together, these results reveal a clear longitudinal predictive chain among the four core variables, supporting the theoretical pathway from cognition (time perspective) to emotion (trait anxiety), then to self-control resources (ego depletion), and finally to behavior (short-video addiction).

The regression coefficients and standard errors of the significant paths in the cross-lagged model are presented in the figure below.

To make the expression more concise, we define a1 as the path coefficient from T1 balanced time perspective to T2 trait anxiety; the regression path coefficient of the trait anxiety at T2 on the ego depletion at T2 is named d1; the path regression coefficient of the ego depletion at T2 on the short-video addiction at T2 is b1; the direct effect of the balanced time perspective at T1 on the short-video addiction at T2, the path coefficient is named c. We explore the chain mediating effect of trait anxiety and ego depletion at different time points (“a1*d1*b1”) [37]. All three models control age and gender, and the confidence interval estimation with bias correction is conducted based on the Bootstrap method (2000 sampling). The results show that the mediating path of a1*b1*d1 is all significant, supporting the chain mediation (Effect = −0.008, 95%CI(−0.014, −0.002)).

In order to further test our hypothesis and obtain more convincing evidence, we added another semi-vertically chained mediating model (M2, as shown in Fig. 3), which takes the balanced time perspective at T1 as the independent variable, the short-video addiction at T2 as the dependent variable, and incorporates trait anxiety and ego depletion at T2 as sequential mediating variables. The model produced a good fit: χ²(2) = 11.375, p = 0.003; CFI = 0.969, TLI = 0.907, RMSEA = 0.100 (90% CI: 0.049–0.160), SRMR = 0.035. Although the RMSEA is slightly higher than the traditional standard, CFI and SRMR indicate that the model is acceptable.

Fig. 3.

A conceptual model of the influence of balanced time perspective on short-video addiction: the mediating roles of trait anxiety and ego depletion

The path analysis results showed that, after controlling for age and gender, the balanced time perspective at T1 significantly negatively predicted the trait anxiety at T2 (β = −0.469, p < 0.001, 95% CI [−0.542, −0.392]), which supported the theoretical expectation that “balanced time cognition can reduce an individual’s long-term anxiety susceptibility”; the trait anxiety at T2 significantly positively predicted ego depletion at the same time point (β = 0.402, p < 0.001, 95% CI [0.295, 0.493]), confirming that high anxiety traits would simultaneously consume psychological resources, consistent with the “immediate resource depletion due to anxiety” hypothesis; ego depletion at T2 also significantly positively predicted short-video addiction at T2 (β = 0.150, p = 0.010, 95% CI [0.043, 0.273]), indicating that the depletion of self-control resources would directly increase the pursuit of immediate rewards, thereby increasing the risk of addiction. Additionally, the direct effect of balanced time perspective on short-video addiction was also significant (β = −0.158, p = 0.009), indicating that in addition to the mediating path, it also has an independent protective effect on its own.

The saturated model provides an ideal foundation for analysis. However, the weak explanatory power of Y2 (R² = 0.086) suggests that the theoretical model may need to incorporate other paths, such as moderating variables, to comprehensively explain the mechanism.

Discussion

Short-video addiction has emerged as a pressing developmental concern in recent years, yet the psychological mechanisms that explain why some adolescents develop addictive usage patterns while others do not remain inadequately understood. Time perspective is a stable individual difference linked to mental health and behavior, yet its mechanism in addiction remains unclear. Study 1 (cross-sectional) examined the serial mediation of trait anxiety (TA) and ego depletion (ED) between balanced time perspective (BTP) and short-video addiction (SVA); Study 2 (longitudinal cross-lagged model) tested temporal ordering.Both studies consistently showed that BTP negatively predicted SVA through TA and ED. Higher BTP was associated with lower TA and ED, leading to lower SVA risk; conversely, elevated TA weakened BTP, increased ED, and aggravated addiction. Study 1’s mediation analysis revealed complete mediation (direct effect non-significant after including TA and ED), supporting a cognitive-affective-resource cascade: BTP reduces threat-oriented appraisals, lowers chronic anxiety [56], preserves self-control resources, and thereby prevents impulsive reward seeking.

These findings integrate Davis’s (2001) cognitive-behavioral model and Baumeister’s strength model. BTP acts as a distal protective factor whose influence is fully transmitted via emotional and executive pathways [25]. Study 2’s longitudinal results verified this chain and revealed a bidirectional feedback loop: T1 TA negatively predicted T2 BTP, suggesting that high anxiety narrows cognitive focus [57], undermining temporal balance [58]. Thus, a protective cycle (BTP → ↓TA → ↓ED → ↓SVA) and a risk cycle(↑TA → ↓BTP → ↑ED → ↑SVA) operate dynamically.

The serial mediation model provides mechanism-based targets for intervention. First, to enhance balanced time perspective, practitioners can implement time perspective training [59] (e.g., life-line drawing, future letter writing, mindful present-focus), which may reduce trait anxiety by fostering a coherent temporal self. Second, to break the anxiety-to-depletion pathway,emotion regulation skills(e.g., cognitive reappraisal, brief mindfulness exercises) should be taught, as they help prevent the rapid consumption of self-control resources. Third,resource restoration activities(e.g., 2-min mindful breathing, brief nature exposure) can replenish executive resources, reducing impulsive short-video use. A staged intervention—emotion regulation first, followed by time perspective restructuring, then resource support—may be particularly effective for adolescents. This is consistent with a recently developed prevention program for adolescent addictive behaviors, which implemented a staged intervention model, starting with guided self-help and progressing to individual intervention if needed [60].

At the same time, several limitations should be noted. First, all data were self-reported, which may introduce bias. Future studies should include peer/parent ratings, device usage logs, or physiological indicators. Second, the six-month follow-up is relatively short; longer intervals are needed. Third, the sample was limited to Chinese middle school students, limiting generalizability to other age groups or cultural contexts. Cross-cultural and age-diverse samples are needed. Fourth, the study focused on protective factors predicting negative outcomes but did not examine how positive factors promote adaptive outcomes or how risk factors affect diverse trajectories. Fifth, measurement invariance for trait anxiety across time points was not formally established. Finally, the adapted short-video addiction scale showed suboptimal CFA fit (CFI = 0.896, RMSEA = 0.117), likely due to large sample size and low item loadings. Future research should refine the items and validate the scale in independent samples. Notably, the six-month test–retest correlation of trait anxiety was low (r = 0.174, p < 0.001), even though the scale demonstrated high internal consistency (T1 α = 0.867, T2 α = 0.869). This may reflect short-term variability in anxiety among students rather than poor measurement quality. Additionally, the use of a student-only sample limits the generalizability of the results. Future studies should include more diverse populations and longer follow-up periods to further explore the temporal stability of trait anxiety in different contexts.

Conclusion

This study provides both cross-sectional and longitudinal research to clarify the internal mechanism by which the ability to perceive balance in time influences the addiction to short videos. It confirms the chain-mediated mechanism of trait anxiety and ego depletion in the inhibitory effect of the ability to perceive balance in time on the addiction to short videos. Adolescents with the ability to perceive balance in time have lower levels of anxiety and ego depletion, ultimately indicating a lower level of addiction to short videos. This result reveals the dynamic influence mechanism of cognition—emotion—resources—behavior, and deepens the research on the perspective of time perception in the addiction to short videos.

Abbreviations

BTP

Balanced Time Perspective

CLPM

Cross-Lagged Panel Model

ED

Ego Depletion

PF

Present Fatalistic

PH

Present Hedonistic

PN

Past Negative

PP

Past Positive

STAI-T

State-Trait Anxiety Inventory

SVA

Short-video Addiction

TA

Trait Anxiety

PIU

Pathological Internet Use

ZTPI

Zimbardo Time Perspective Inventory

Authors’ contributions

Z.Y.was involved in conceptualization,drafting and revising the manuscript,data processing and analysis,and result visualization. G.X. was responsible for project funding,project management,participation in discussions, and manuscript revision. F.H. participated in data collection,manuscript review.

Funding

1. General Project of Education under the National Social Science Fund:

“Research on Identification and Intervention of Short Video Addiction among Left-Behind Children Based on Multimodal Data and Machine Learning”(BBA250048).

2. The General Projects for Philosophy and Social Science Research of Department of Education of Hubei Province in 2023 (23Y105)

3. The General Projects for Social Science Foundation of Yangtze University in 2022 (2022csy09)

4. The project for Hubei Provincial Research Center for Grassroots Governance of Hubei Province Key Research Base for Humanities and Social Sciences in Universities in 2024.5.the key projects of education science plan of Hubei Province in 2025 (2025GA026）

5. The key projects of education science plan of Hubei Province in 2025 (2025GA026）

6.Achievements of General Project of Hubei Provincial Social Science Fund (Post-Funding Project) in 2025 (HBSKJJ20253425）

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This survey was approved by the ethics committee of psychological research at Yangtze University (Approval No:CE5S-20260424).Informed consent was obtained from both participants and their parents, and their anonymity and confidentiality were strictly protected throughout the study.And the study was conducted in accordance with the Declaration of Helsinki (as revised in 2013).

Consent for publication

All authors have approved the manuscript for publication. Consent for publication is not applicable as the study does not involve any identifying images or personal/clinical details of participants that compromise anonymity.

Competing interests

The authors declare no competing interests.