Maartje Boer

SMU AND MENTAL HEALTH 169 6 behaviors. By including correlations between within-person values in the same year (gray double arrows), the cross-lagged paths were also controlled for associated increases (or decreases) in behaviors within the same year. All RI-CLPMs were estimated using Maximum Likelihood with Robust standard errors (MLR). Modelling Procedure In our first model, we fitted a RI-CLPM with three repeated measures of SMU intensity, SMU problems, and depressive symptoms (M1a). In our second model, we estimated the first model but with life satisfaction instead of depressive symptoms (M2a). We examined depressive symptoms and life satisfaction in separate models due to their collinearity. In subsequent models, we extended the first and second model with the four mediators, with one mediator per model (M1b-e and M2b-e). Model fit was evaluated using the Comparative Fit Index (CFI), Tucker Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA), and the Standardized Root Mean square Residual (SRMR). The standardized results (STDYX) of the models were used for the interpretation of the effect sizes. Analyses were conducted using Mplus 8.4 (L. K. Muthén & Muthén, 2017b). Results Descriptive Analyses Prior to the main analyses, we studied whether adolescents’ scores on the study measures changed over time and associations between the demographic characteristics and the study measures using multilevel analysis on the data in long format (Table 6.2). On average, adolescents’ SMU problems did not change over time. Relative to T1, adolescents’ SMU intensity increased in T2 and T3. In addition, adolescents reported increased depressive symptoms and decreased life satisfaction in T2 and T3 when compared to T1. Also, adolescents reported decreased face-to-face contact in T2 and T3 relative to T1. On average, upward social comparisons, cybervictimization, and subjective school achievements did not change over time. Although on average somemeasures did not change over time, the within-person residual variances of all measures were significant. This suggests that adolescents