Maartje Boer

SUMMARY AND DISCUSSION 297 9 our understanding of the association between adolescents’ SMU intensity and wellbeing, it is pivotal that future studies consider SMU problems, by controlling for them, when studying the association between SMU intensity and wellbeing. This would limit the chance of finding a spurious negative association. In addition, our findings illustrate the importance of using multiple analytical methods for longitudinal data to understand the association between SMU behaviors and wellbeing: Although random intercept cross- lagged panel models indicated that changes in wellbeing did not predict subsequent changes in SMU problems on the within-person level, this does not exclude the possibility that lower wellbeing at the between-person level affect adolescents’ susceptibility to SMU problems, as demonstrated by our study using latent class growth modelling (Key finding 4; Chapter 5-7). That is, for example, while adolescents with higher levels of ADHD-symptoms compared to the average adolescent seemed sensitive to SMU problems (Chapter 7), adolescents’ SMU problems did not vary as a function of their temporal fluctuations in ADHD-symptoms (Chapter 5). Other research also suggests that between-person differences in wellbeing impact SMU problems: A longitudinal study among adolescent girls that used latent growth modelling showed that girls with higher levels of depressive symptoms at baseline reported stronger increases in SMUproblems over time than girls with lower levels of depressive symptoms at baseline (Raudsepp & Kais, 2019). Hence, relying solely on within-person fluctuations of behaviors, such as random intercept cross-lagged panel models, to derive the potential causal order of behaviors may not be justified, as this type of analysis does not establish the potential influence of more stable individual characteristics. In line with this suggestion, recently, other researchers stressed that the limitation of random intercept cross-lagged panel models is that it does not provide insight into the effects of between-person differences, while this is often relevant in research on psychology and individual development (Lüdtke & Robitzsch, 2021; Orth et al., 2021). In addition, the finding that at the within-person level, temporal fluctuations in wellbeing were not associated with subsequent temporal fluctuations in SMU problems (Chapters 5, 6), should be interpreted in light of the yearly time intervals of our data. More specifically, the finding does