Unveiling Diverse Trajectories of Internet Addiction and the Influence of Family Environment and Obsessive Beliefs: Multi-Wave Longitudinal Study With Growth Mixed Model.

Background: Recent longitudinal studies have revealed the heterogeneity of the developmental trajectory of internet addiction (IA), which is believed to be due to the influences of interindividual variables. In a social-cognitive framework, family environment (FE) and obsessive beliefs (OBs) are associated with IA severity. However, it remains unclear how these environmental and individual cognition factors interact to influence IA development. Objective: This study aimed to identify the growth trajectories of IA among college students, considering individual differences over time, and explore how FE and OBs contribute to the identified trajectories. Methods: A convenience sample of 3575 first-year college students (female: 65.29% [n=2334], mean age 18.7 [SD 0.9]) was recruited, with longitudinal data collected over 3 waves (2019-2021) and retention rates of 72.4% (n=2585) at T1 and 61.34% (n=2193) at T2. IA trajectories were classified using the latent growth mixture model, and the effects of FE and OBs on the IA intercept and slope were examined by the latent growth curve model. Multivariate logistic regression assessed the predictive effects of FE and OBs on trajectory classification, controlling for sex, residence, and parents' education. Furthermore, structural equation modeling was used to map the road from FE and OBs to follow-up IA. Results: Latent growth mixture model uncovered 4 distinct trajectories: high-risk (5.09%), medium to high-risk (29.85%), medium to low-risk (35.95%), and low-risk (29.11%), while latent growth curve model revealed that both FE and OBs significantly influenced IA initial level (intercept: βFE_cohesion/ conflict=-0.169/-0.191, P<.001; βOBs_responsibility/ control of thoughts=0.129/0.279, P<.05) and development rate (slope: βFE_conflict=0.073, P<.05; βOBs_ control of thoughts=-0.165, P<.001). Furthermore, logistic regression showed that compared with the low-risk group: high-risk students exhibited reduced cohesion (odds ratio [OR] 0.831, 95% CI 0.721-0.957; P<.01), elevated conflict (OR 0.866, 95% CI 0.745-1.006; P<.05), and lower independence (OR 0.841, 95% CI 0.710-0.996; P<.05); medium-high risk showed higher conflict (OR 0.890, 95% CI 0.826-0.959; P<.01) and OBs (ORresponsibility 1.020, 95% CI 1.003-1.037; ORcontrol of thoughts 1.028, 95% CI 1.010-1.045; P<.01); and medium-low risk had increased conflict (OR 0.911, 95% CI 0.841-0.986; P<.05). Moreover, structural equation modeling demonstrated a significant partial mediation effect of OBs on the relationship between FE and follow-up IA (effect T0/ T1/ T2=-0.03/-0.02/ -0.02, P<.001). Conclusions: This study reveals 4 heterogeneous IA trajectories among college students, influenced by both FE and OBs through their effects on the IA initial level and development rate. Notably, FE not only influences IA development directly but also exerts its influence indirectly through the mediation of OBs. These findings highlight the necessity of targeted interventions addressing family environmental risk factors and maladaptive OBs in youth for IA.