Screen time, including watching television or movies, playing video games, visiting social media sites, and video chatting, is core to adolescents’ daily lives and is perhaps their most prevalent behavior. A survey of 114,072 Chinese school children revealed that 34.7 % of them were exposed to more than 2h of screen time daily (Chen et al., 2021). Excessive screen time has been verified to be related to future poorer cognitive performance, physical problems, and mental health problems (Chen et al., 2021, Tang et al., 2021). Previous theories and studies have revealed that screen time is significantly correlated with both depressive symptoms and sleep difficulties (Tang et al., 2021). On the one hand, the upward social comparison hypothesis postulates that people comparing themselves with those they perceive to be in a more favorable position (Boers et al., 2019). For this hypothesis, the repeatedly exposure to screen content may lead to negative social comparison, cognitive biases and physiological activation, which in turn contribute to a range of psychological and physical issues (Boers et al., 2019, Cain and Gradisar, 2010, Nesi and Prinstein, 2015). On the other hand, the compensatory internet use theory posits that individuals are likely to seek support online as a means of alleviating their distressing feelings (Kardefelt-Winther, 2014), suggesting that depressive symptoms and sleep difficulties may be linked to excessive screen time. The above theories suggest that screen time may be reciprocally related to depressive symptoms and sleep problems over time. More screen time may lead to depressive symptoms and sleep problems, which in turn increase screen time. Therefore, it is vital to investigate the reciprocal associations among screen time, depressive symptoms and sleep in a longitudinal framework. More importantly, early adolescence is a period of significant biological, social, cognitive, and emotional changes (Rudolph, 2014). Adolescents become more sensitive to social information and rewards, such as novel content in electronic screen activity (Nesi, 2020), and are more susceptible to emotional and behavioral problems (Rudolph, 2014). As a result, early adolescence is a particularly important period for understanding screen time and its associations with depressive symptoms and sleep (i.e., sleep quality and duration).

Despite the rich insights provided by theories and previous studies, important limitations remain. First, while some studies have examined the reciprocal associations between screen time and depressive symptoms as well as sleep, the causative links in these relationships remain inconclusive. Moreover, few studies have focused on how these relationships vary between boys and girls. Addressing this research gap is crucial for developing targeted intervention strategies. Second, the association between screen time and depressive symptoms is likely to be complex and varied, given the heterogeneity of depressive symptoms (Østergaard et al., 2011). The network theory of psychopathology (Borsboom, 2017) postulates that depression is a network system of interrelated symptoms, implying that the association between screen time and depression may vary with different depressive symptoms. Guided by the network theory of psychopathology, this study investigates the longitudinal associations and sex differences among screen time, depressive symptoms, and sleep over time among Chinese early adolescents via cross-lagged panel network (CLPN) models.

Multiple theories and empirical studies have posited associations between screen time and depressive symptoms. On one hand, upward social comparison hypothesis postulates that individuals tend to compare themselves with other who they believe are in a more favorable position (Boers et al., 2019). Screen users are more likely to be frequently exposed to idealized portrayals and images, which they may use as a benchmark for self-comparison. These comparisons often foster cognitive biases and low self-evaluations, which can contribute to the development of depressive symptoms in the long term (Boers et al., 2019, Nesi and Prinstein, 2015). Empirical longitudinal research has also revealed that more screen time increases depressive symptoms one year later (Boers et al., 2019). On the other hand, the compensatory internet use theory (Kardefelt-Winther, 2014) suggests that screen time activity may act as a coping mechanism for negative emotions and social needs that are not satisfied in real life. Specifically, adolescents with depressive symptoms and their associated sadness lead to increased screen time, such as addicting to internet games and short videos, to alleviate distressing feelings (Qu et al., 2024, Zhou et al., 2023). Considering these theories, screen time and depressive symptoms may serve as risk factors for each other, reinforcing a spiral cycle in the long term. However, empirical studies have yielded inconsistent results regarding the longitudinal association between screen time and depressive symptoms across studies (Li et al., 2022, Tang et al., 2021). For example, a recent review of 35 longitudinal studies among youth aged 10–24 years revealed a modest association between screen time and subsequent depressive symptoms, which was stronger than the reverse effect (Tang et al., 2021); whereas some longitudinal studies involving adolescent samples have not reported significant longitudinal associations between screen time and depressive symptoms (Houghton et al., 2018) or effects from depressive symptoms on screen time (Zink et al., 2023).

One potential explanation for these discrepancies is sex differences. Previous studies have examined the associations between screen time and depressive symptoms across sexes. Specifically, a meta-analysis found significant associations between screen time and depressive symptoms among women (Li et al., 2022). Girls are more likely to exhibit social comparison and feedback-seeking behaviors (Nesi & Prinstein, 2015). As screen time increases, girls spend more time engaging in social comparison and self-centered thinking, such as rumination, which can potentially lead to depressive symptoms (Nesi & Prinstein, 2015). Tang et al. (2021) further underscored sex as a strong moderating factor in the associations between screen time and depressive symptoms. Notably, this study highlighted the inconsistencies in the direction of the effect, suggesting the need for further longitudinal research to elucidate the differences in how screen time is associated with depressive symptoms among boys and girls (Tang et al., 2021).

Another source of mixed results may be the heterogeneity of depressive symptoms (Zink et al., 2020). Studies on screen time and depressive symptoms often use categorical depression diagnoses or symptom composite scores as the outcome variables of interest (Tang et al., 2021, Wang et al., 2021). However, depression manifests with a high degree of heterogeneity in terms of clinical presentation, origins, and ensuing impacts (Østergaard et al., 2011). Moreover, the network theory of psychopathology conceptualizes psychopathology (e.g., depression) as a complex and dynamic network in which symptoms function as nodes and causal interactions between these symptoms form links between the nodes (Borsboom, 2017). Within this framework, each depressive symptoms have different weights within the symptom network and maintain distinct relationships with screen time (Zink et al., 2020). Symptom-level analyses are critical for understanding how specific symptoms relate to particular outcomes and thus identifying which symptoms should be prioritized for interventions (Borsboom, 2017). Specifically, network analysis can provide insights into whether depressive symptoms associated with irritability are more strongly connected to subsequent increased screen time, potentially fostering a negative feedback loop between depressive symptoms and screen time that contributes to the emergence and persistence of mental health issues. Certain network analyses have further explored the associations between screen-related variables and depressive symptoms (Lin et al., 2020). Specifically, Lin et al. (2020) constructed cross-sectional networks among early adolescents, revealing a stable positive association between age-inappropriate screen time and various depressive symptoms, whereas other forms of screen time showed varied associations with depressive symptoms (Lin et al., 2020). Nevertheless, this research was conducted only on the basis of cross-sectional data, limiting the ability to infer the temporal and causal relationships between symptoms and investigate whether a central symptom resulted in or was caused by other symptoms (Funkhouser et al., 2021).

Cross-lagged panel network (CLPN) models integrate the network model with the cross-lagged panel model to construct temporal and casual effect network associations between symptoms (Funkhouser et al., 2021). Moreover, we identify central symptoms by calculating expected influence (EI), which locates the symptoms that have the greatest impact in the network. A symptom with high EI shows strong connections with other symptoms in the network (Borsboom & Cramer, 2013). Switching on symptoms with high EI will quickly activate other symptoms throughout the network, which eventually manifests as the emergence and development of disorders. Furthermore, EI indices include out-expected influence (out-EI), in-expected influence (in-EI), and bridge expected influence (bridge EI). Out-EI determines the influence of symptoms on predicting all other symptoms and in-EI signifies the susceptivity of symptoms being predicted by other symptoms (Funkhouser et al., 2021). Identifying central symptoms with high out-EI or in-EI provides directional information and yields valuable information for psychological interventions that target antecedent symptoms (i.e., symptoms with high out-EI) rather than those without evidence of downstream effects (i.e., symptoms with high in-EI). Bridge EI identifies bridge symptoms that are more likely to related to symptoms of another disorder (Cramer et al., 2010). Bridge symptoms are more likely to activate symptoms of another disorder, thereby creating a vicious cycle and lead to the development of comorbid mental issues (Cramer et al., 2010). Identifying bridge symptoms and associations can help to understand the co-morbid mechanisms, providing new insight to more precise and efficient intervention. CLPN models have been used to examine the associations between problematic screen time and adjustment outcomes. For example, Qu et al. (2024) used CLPN models to investigate short video addiction and depressive symptoms among Chinese adolescents; their findings indicate that the depressive symptom ‘sad mood’ was a bridge symptom that motivated adolescents to seek ‘mood modification’ through short video platforms. Simultaneously, the addiction symptom of ‘conflict’ was a symptom with the highest out-EI and bridge EI, indicating that the activation of ‘conflict’ increases the risk of co-occurrence between depression and short video addiction.

In addition, the impact of screen time may vary between weekdays and weekends. Specifically, prior studies have indicated that average levels of screen time are typically higher on weekends than on weekdays (Sanz-Martín et al., 2022, Zink et al., 2023). Further longitudinal research has also demonstrated that greater screen time on weekends is positively associated with withdrawn/depressed symptoms one year later in females but not with anxious/depressed symptoms or somatic complaints (Zink et al., 2023). Therefore, our study employs CLPN models to further explore how the longitudinal relationships between screen time and various depressive symptoms may differ by sex and specific periods of the week (i.e., weekdays and weekends).

In addition to increases in screen time, the recorded incidence of sleep problems has increased, particularly in adolescence (Pagano et al., 2023). Sleep health in adolescence is multifaceted, encompassing both quantitative aspects (e.g., sleep duration) and qualitative aspects (e.g., sleep quality). Our study focuses on both sleep duration and sleep quality, which are essential and common indicators for assessing sleep health (Vazsonyi et al., 2021). Theoretical and empirical studies have noted the negative effect of excessive screen time on sufficient and high-quality sleep (Brautsch et al., 2023, Cain and Gradisar, 2010). Specifically, the sleep displacement hypothesis argues that prolonged screen time may encroach upon time otherwise allocated for healthy activities (Weigle & Shafi, 2024), such as sufficient and good-quality sleep (Cain and Gradisar, 2010, Kraut et al., 1998). Moreover, screen use exposes adolescents to a vast number of idealized and stimulating information, which may induce greater psychological and physiological arousal and in turn adversely relate to sleep by delaying bedtime and bringing poor sleep quality (Brautsch et al., 2023). Empirically, a substantial body of review and meta-analysis evidence has indicated that prolonged screen time is negatively related to both sleep duration and quality over time (Brautsch et al., 2023, Pagano et al., 2023). Another cross-sectional study revealed that the association between screen time and sleep difficulties was more pronounced on weekdays (Khan et al., 2023).

Furthermore, poor sleep health may create a self-reinforcing vicious circle that further increase screen usage (Pagano et al., 2023). In this cycle, screen use can serve as a common compensatory behavior for individuals experiencing sleep difficulties (Pagano et al., 2023). The youths who have sleep problems may engage in electronic screen activities to regulate negative feelings that come with not being able to fall asleep or sleep well. Taken together, a vicious cycle between screen time and sleep can take place, as prolonged screen time drives sleep problem, which in turn leads to excessive screen use as a means to cope with sleep-related issues. However, among the limited longitudinal studies, three studies have investigated the longitudinal association between sleep and screen time among adolescents. While Barlett et al. (2012) identified a negative link from sleep duration to later excessive screen time, neither Poulain et al. (2019) nor Zink et al. (2023) reported a significant effect of sleep duration on screen time. A meta-analysis focusing on adolescents aged 10–19 years revealed a nonsignificant effect (r = -0.13) from sleep to screen time (Pagano et al., 2023).

The physiological and psychological changes that occur during adolescence increase the risk of sleep problems, with this effect being particularly pronounced in girls. Specifically, research has shown that girls in adolescence generally experience poorer sleep quality, shorter sleep duration and more insomnia symptoms than boys do (Crockett et al., 2020). Prior studies have also examined the association between screen time and sleep across sexes. A recent study using representative data from adolescents in 38 countries revealed stronger positive associations between screen time across all device types and sleep difficulties among girls (Khan et al., 2023). In addition, girls are more vulnerable to depressive symptoms in early adolescence (Crockett et al., 2020). Consequently, the interplay among prolonged screen time, sleep, and depressive symptoms may be pronounced in adolescent girls. However, no research has focused on potential sex differences in a network analysis of screen time, depressive symptoms, and sleep. Considering the mixed findings regarding sex differences in previous studies, the present study aims to examine these differences in screen time, depressive symptoms, and sleep via network analysis.

While extensive literature suggests a complex interplay among screen time, depressive symptoms, and sleep in adolescence, the directionality and sex-specific differences in the longitudinal associations among these factors at the symptom level in early adolescents remain poorly understood. Thus, this study aims to examine the potential differences across sex in the longitudinal associations among screen time, depressive symptoms, and sleep using a cross-lagged panel network modeling approach. Additionally, our study focused on identifying central symptoms and bridge symptoms in the network, which could illuminate potential targets for early prevention and intervention.