The association between self-esteem and dimensions and classes of cross-platform social media use in a sample of emerging adults - Evidence from regression and latent class analyses

There is growing interest in the role of social media use in young people's mental health, and self-esteem has been hypothesised as a potential link in this association. However, existing studies have tended to use basic indicators of use in isolation and single-platform data, and further, have not controlled for other key variables. To address these limitations, emerging adults completed online questionnaires on social media engagement and self-esteem. In line with the interpersonal-connection-behaviours framework we explored online behaviours that putatively connect and disconnect users, e.g. meeting new people and engaging in social comparisons, respectively. Data were analysed using two methodologies, facilitating examination of the relationship between self-esteem and individual engagement indicators (regression analysis) as well as patterns of use (Latent Class Analysis). Overall levels of use and upward social comparisons independently predicted variance in self-esteem scores, even after controlling for demographic and socioeconomic covariates. Further, membership to meaningful, empirically-derived classes of social media users was predicted by self-esteem. These findings indicate that the association between social media use, social comparisons and self-esteem is robust, and extends to multi-platform data. We argue that such a move away from studies of single-platform data is critical if findings are to be generalised.