📄 social_science

Cyber Victimization, Smoking, and Non-Suicidal Self-Injury Among Chinese Adolescents: A Three-Wave Random-Intercept Cross-Lagged Panel Study

This three-wave study of Chinese adolescents reveals that while stable between-person vulnerabilities largely explain the co-occurrence of cyber victimization, smoking, and non-suicidal self-injury (NSSI), there is partial evidence for time-specific, transactional within-person pathways where smoking may mediate the bidirectional relationship between cyber victimization and NSSI.

Original authors: zhen wang, yuanyuan chen

Published 2026-08-15

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: zhen wang, yuanyuan chen

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine your brain is a high-tech control tower, constantly monitoring the weather of your social life. Sometimes, the sky is clear; other times, a storm rolls in. In the world of psychology, scientists are obsessed with understanding what happens when that storm hits a teenager's life. One particularly nasty storm is cyber victimization—that's the digital version of being bullied, where mean comments, rumors, or exclusion happen online, often feeling inescapable because they live on screens you can't easily turn off. Another storm is non-suicidal self-injury (NSSI), which is when a young person hurts themselves on purpose (like cutting or burning) not to end their life, but to try to stop overwhelming emotional pain or to feel something else when they feel numb.

Now, imagine there's a third element in this mix: smoking. You might wonder, how do these three things connect? Does getting bullied online make someone start smoking, which then leads to self-harm? Or does self-harm lead to smoking, which then makes someone more likely to get bullied? For a long time, researchers have looked at these as separate lines on a graph, but they often missed a crucial detail: the difference between who you are and what you are doing right now. Think of it like the difference between a person's permanent height (their "trait") and how much they jump up and down on a trampoline at a specific moment (their "state"). Most studies mixed these up, but this new research wants to see if the jumping of one behavior causes the jumping of another, moment by moment, or if it's just that some people are simply more prone to all of these storms in general.

This study, conducted with 1,835 teenagers in China, acts like a high-speed camera trying to capture the exact moment these behaviors interact. The researchers followed these students over three time points, checking in every six months. They used a fancy statistical tool called a "Random-Intercept Cross-Lagged Panel Model" (RI-CLPM), which is essentially a way to separate the "permanent personality" factors from the "temporary mood" factors.

Here is what the study found, and it's a bit more complicated than a simple "A causes B" story. First, at the "permanent" level, the data showed a strong link: teenagers who generally experienced more cyber bullying also tended to generally report more self-harm and smoking across the board. It's like saying, "If your control tower is usually in a stormy zone, you're likely to see all three types of bad weather."

However, when the researchers zoomed in on the "temporary" changes—looking at whether a specific spike in bullying led to a specific spike in smoking or self-harm six months later—the picture became a bit foggy. They found partial evidence for a chain reaction. Specifically, they saw that when a student experienced more cyber victimization than usual at the first check-in, they were more likely to start smoking more than usual six months later. And, in a later interval, when a student smoked more than usual, they were more likely to engage in self-harm more than usual six months after that. This suggests a possible path: Cyber Bullying → Smoking → Self-Harm.

But the story doesn't end there. The researchers also looked at the reverse: could self-harm lead to smoking, which then leads to more bullying? They found some evidence for this too, but the connections were "mixed and time-specific." This means the chain didn't work the same way every time they checked. For instance, the link between bullying and smoking was strong in one six-month gap but disappeared in the next. Similarly, the link between smoking and self-harm showed up in one gap but not the other.

The authors are careful to tell us that this isn't a fixed, unbreakable rule. They didn't prove that bullying always leads to smoking, which always leads to self-harm. Instead, they suggest that for some teenagers, at certain times, these behaviors might feed into each other in a loop. However, a huge chunk of the connection between these three things comes from the "permanent" side of things—meaning that some teens might have underlying vulnerabilities (like difficulty managing emotions or facing other life stresses) that make them susceptible to all three issues simultaneously, rather than one strictly causing the other in a straight line.

In short, this study paints a picture of a complex dance rather than a straight line. While smoking might act as a bridge between cyber bullying and self-harm for some teens at some times, the researchers warn that we shouldn't assume this happens the same way for everyone or at every moment. The most consistent finding is that these three issues often travel together because of deeper, stable factors in a teenager's life, and the specific "cause-and-effect" steps between them are tricky and change depending on the timing.

Technical Summary: Cyber Victimization, Smoking, and Non-Suicidal Self-Injury Among Chinese Adolescents

Problem Statement
Non-suicidal self-injury (NSSI) is a prevalent adolescent mental health concern, often linked to interpersonal stressors such as cyber victimization. While empirical work has established associations between cyber victimization and NSSI, the specific temporal pathways connecting these variables remain unclear. Existing longitudinal studies often utilize conventional cross-lagged or mediation models that conflate stable between-person vulnerabilities with within-person temporal changes. Furthermore, the role of smoking as a potential behavioral pathway linking cyber victimization to NSSI, and the possibility of a reverse pathway (NSSI leading to cyber victimization via smoking), has not been rigorously tested using a design that separates trait-like stability from state-like fluctuations, particularly within the Chinese adolescent context.

Methodology
This study employed a three-wave longitudinal design with six-month intervals, assessing 1,835 adolescents (55.9% male; M age = 12.32) from two junior high schools in Guangdong and Shandong provinces, China. Data were collected at the fall of Grade 7 (T1), spring of Grade 7 (T2), and fall of Grade 8 (T3).

Measures: Cyber victimization, smoking, and NSSI were assessed via self-report. Smoking was measured using a single ordinal item adapted from prior Chinese research, modeled as a continuous indicator of involvement.
Analytic Approach: The study utilized Random-Intercept Cross-Lagged Panel Models (RI-CLPM). This framework was selected to disentangle stable between-person differences (trait-like mean levels) from within-person temporal associations (deviations from an individual's own mean).
Controls: All models controlled for gender, age, and family socioeconomic status (SES).
Procedure: Analyses were conducted in R using the lavaan package. Missing data were handled via Full Information Maximum Likelihood (FIML). Model fit was evaluated using $\chi^2$ , CFI, TLI, and RMSEA. The analysis proceeded in two steps: (1) a bivariate RI-CLPM examining cyber victimization and NSSI, and (2) a three-variable RI-CLPM testing smoking as a mediator in both hypothesized (cyber victimization $\rightarrow$ smoking $\rightarrow$ NSSI) and reverse (NSSI $\rightarrow$ smoking $\rightarrow$ cyber victimization) directions. Longitudinal indirect effects were tested using bias-corrected bootstrap confidence intervals (5,000 resamples).

Key Results

Between-Person Level: Strong, stable covariation was observed. Adolescents with higher trait-like levels of cyber victimization consistently reported higher trait-like levels of NSSI and smoking across waves. In the mediation model, the random intercepts for all three constructs were positively correlated ( $r$ values ranging from 0.18 to 0.50, all $p < .05$ ).
Within-Person Level (Cyber Victimization $\rightarrow$ NSSI): A time-specific association was found where higher-than-usual cyber victimization at T2 predicted higher-than-usual NSSI at T3 ( $\beta = 0.57, p = .002$ ). However, cyber victimization at T1 did not significantly predict NSSI at T2.
Mediation Pathways:
- Hypothesized Path: A significant indirect effect was found from T1 cyber victimization to T3 NSSI via T2 smoking (Indirect effect = 0.01, 95% CI [0.003, 0.027]). This was driven by a significant path from T1 cyber victimization to T2 smoking ( $\beta = 0.14$ ) and T2 smoking to T3 NSSI ( $\beta = 0.11$ ). Notably, the component paths were not consistent across adjacent intervals (e.g., T2 cyber victimization did not predict T3 smoking).
- Reverse Path: A significant reverse indirect effect was also observed from T1 NSSI to T3 cyber victimization via T2 smoking (Indirect effect = 0.03, 95% CI [0.002, 0.051]). This path was driven by T1 NSSI predicting T2 smoking ( $\beta = 0.35$ ) and T2 smoking predicting T3 cyber victimization ( $\beta = 0.08$ ).
Mixed Cross-Lagged Paths: The three-variable model revealed a complex, mixed pattern of cross-lagged associations. For instance, while T1 NSSI predicted T2 cyber victimization, this effect did not replicate in the T2-to-T3 interval. Similarly, bidirectional associations between smoking and NSSI were observed from T2 to T3 but not consistently from T1 to T2.

Significance and Claims
The authors claim that this study provides partial evidence for transactional within-person links among cyber victimization, smoking, and NSSI, while emphasizing that stable between-person vulnerability accounts for a substantial portion of the observed co-occurrence.

Methodological Contribution: The study demonstrates the utility of RI-CLPM in preventing the misinterpretation of stable between-person covariation as directional temporal effects. It highlights that conventional models may overstate the strength of short-term causal mechanisms by failing to control for enduring individual differences.
Theoretical Contribution: The findings support a stress-coping framework where smoking may serve as a behavioral pathway linking cyber victimization to NSSI, and conversely, where NSSI may alter peer environments to increase vulnerability to cyber victimization. However, the authors explicitly caution that these pathways are time-specific rather than uniformly stable across developmental periods.
Practical Implications: The results suggest that prevention efforts should address cyber victimization as a significant stressor and consider smoking as a potential marker for heightened NSSI risk among victimized adolescents. Integrated interventions targeting cyber safety, emotion regulation, and substance use are proposed as potentially more effective than single-behavior approaches.

Limitations and Modesty of Claims
The paper maintains a modest tone regarding its conclusions. The authors explicitly state that the findings do not establish a fixed causal chain. They note that the pathways were not consistently replicated across all adjacent intervals, suggesting that the mechanisms may depend on specific developmental or contextual conditions (e.g., peer network shifts, academic pressure). Limitations acknowledged include the reliance on self-report data, the use of a single ordinal item for smoking, the three-wave design limiting the testing of nonlinear dynamics, and the specific cultural context of the sample. The authors conclude that while the study identifies meaningful within-person dynamics, the strong between-person associations indicate that enduring individual and contextual vulnerabilities are primary drivers of the co-occurrence of these behaviors.

Technical Summary: Cyber Victimization, Smoking, and Non-Suicidal Self-Injury Among Chinese Adolescents

More like this