Abstract
<p>Advancing research on adolescents’ social media use and wellbeing requires methods that capture the temporal dynamics of both phenomena and their interplay. Traditional longitudinal models, even when identifying the within-person relationship between social media use and an outcome of interest, do not distinguish between effects driven by use that is greater or lesser than an individual’s average. Consequently, it remains unclear whether higher-than-usual versus lower-than-usual social media use shows similar persistence over time (autoregressive effects) and whether it has differential longitudinal relationships with wellbeing (cross-lagged effects). To address this gap, we apply an extension of the traditional random-intercept cross-lagged panel model that separately estimates autoregressive and cross-lagged effects for deviations above and below adolescents’ (N = 278, aged 14-18) within-person averages, thereby examining asymmetries. Our findings revealed asymmetric temporal dynamics that differed across social media activities and wellbeing outcomes. Autoregressive analyses showed that below-average social media use was temporally “sticky” for both active and passive use, whereas above-average use was “sticky” only for passive use. Cross-lagged analyses showed that lower-than-usual passive use predicted lower subsequent happiness and higher loneliness, while lower-than-usual use across all active and passive use predicted higher subsequent anger. Associations following above-usual use were mostly null, although above-usual video watching predicted lower happiness and higher anger; sadness showed no asymmetry. Overall, these results advance the modelling of person-specific temporal dynamics, highlighting that whether adolescents engage more or less than usual matters alongside the type of social media activity, and that wellbeing dimensions vary in their sensitivity to these temporal asymmetries.</p>