The R syntax was used for data prepration, descriptives, correlations, person-specific effects, and assumption checks. The SPSS syntax was used for factor analysis and reliability analyses of the self-esteem contingency data. The M plus syntax was used for the Confirmatory factory analysis and for the DSEM analyses of the effects of time spent social media (SMT) and valence of social media experiences (SMV) on self-esteem, which were estimated in seperate models (see corresponding subfolders). For those who do not have access to Mplus, it would be possible to open the Mplus input and output files with Notepad. For each model, we first estimated a model including correlations with the autoregressive effect of self-esteem. As these models were too complex, we continued the analyses by removing correlations with the autoregressive effect (phi). Each model was tested once with 5,000 iterations and once with 10,000 iterations. We conducted several sensitivity and exploratory analyses: - removing outliers - removing participants whose answers were considered as potentially untrustworthy (for more details, see Pouwels et al [2021]) - estimating the effects of time spent using social media and the valence of social media experiences in 1 model - examining the association of educational level with the effect of time spent using social media on self-esteem and the valence of social media experiences on self-esteem. Note that educational level has not been included in the published dataset.