Maartje Boer

CHAPTER 8 252 showed a significant decrease in the deviance relative to the baseline fixed effects models (M5a-f: deviance p range = < 0.001 to 0.009). Correspondingly, in all models, the AIC decreased (M5a-f AIC range = -16.9 to -4.6). In contrast, the BIC increased for two models (M5a,b: BIC = 6.8 and 8.7), suggesting that for these models the random slopes deteriorated model fit. However, given that for all models themajority of the indices, if not all, showed significant slope variance, the findings suggest that the within-person associations between all six SMU activities and life satisfaction varied across adolescents. Furthermore, the 95% PIs suggest that in all models, the associations ranged from moderate negative to moderate positive associations (M5a-f: LL- β range = -0.361 to -0.271; UL- β range = 0.231 to 0.314), which is considered substantial. This means that for some adolescents, increased intensity of a SMU activity was associated with decreased life satisfaction, whereas for others, increased intensity of a SMU activity was associated with increased life satisfaction ( RQ3a ). When controlled for SMU problems, the variances in the associations decreased somewhat (M6a-f). However, most of the fit indices indicated that there were still significant variances in the within-person associations between the SMU activities and life satisfaction. Differences by Upward Social Comparison Table 8.6 shows whether the observed variation in the within-person associations could be explained by adolescents’ average level of upward social comparison (see Table A8.4 from the Appendix for all model estimates). All models showed that the higher the average level of upward social comparison, the lower was the average level of life satisfaction. A moderating effect of the average level of upward social comparison on the association between one of the indicators of SMU intensity and life satisfaction was found: Among adolescents who reported lower averages in upward social comparison, increases in SNS liking were associated with increases in life satisfaction, whereas among adolescents with higher averages in upward social comparison, increases in SNS liking were associated with decreases in life satisfaction (M7c: B SNS liking = -0.030*upward social comparison). This moderating effect was also found when controlling for SMU problems (M8c). However, the variance of the slope of SNS liking was not reduced by this moderation (M5/6c relative to M7/8c), which indicates that the explanatory