Maartje Boer

CHAPTER 2 52 representative data on adolescents. As a result, it remained unclear whether prevalence differences reported in previous research (Bányai et al., 2017; Ho et al., 2017; Mérelle et al., 2017) were biased by varying measurement properties across subpopulations. The criterion validity analysis showed that the higher the number of endorsed problematic SMU criteria, the higher the probability of reporting problems related to mental health, school functioning, and sleep, confirming good criterion validity of the test score interpretations. Problematic users typically experience unpleasant feelings such as stress or anxiety when SMU is restricted, which may induce mental health problems. Also, the loss of control over SMU may make it difficult to regulate schoolwork responsibilities, which may increase school problems. In addition, being preoccupied with social media or feeling a constant urge to go online may be associated with sleep difficulties. Or conversely, adolescents with problems related to their mental health, school functioning, or sleep may engage in problematic SMU to cope with their problems (Kuss & Griffiths, 2017). Longitudinal research is warranted to examine the directionality of these associations. In addition, in the criterion validity analysis we also examined the extent to which mental health, school, and sleep problems differed between three subgroups: normative users (endorsement of max. one criterion), risky users (endorsement of two to five criteria), and problematic users (endorsement of six to nine criteria). Although these thresholds for classification were based on observed patterns in the data, research using clinical samples is required to examine whether this classification is justified. Nevertheless, the criterion validity analysis supports the validity of the classification, because the three subgroups differed significantly on mental health, school, and sleep problems, with problematic users being most at risk, followed by risky users and normative users. Furthermore, the finding that risky users were more likely to report problems related to mental health, school, and sleep emphasizes that it is important to study moderate levels of problematic SMU and not only the highest levels, as the presence of a few criteria already seems indicative of problems in several important life domains. In line with former research (Bányai et al., 2017; Ho et al., 2017; Mérelle et al., 2017), our study showed that the number of endorsed problematic SMU criteria was highest among girls, low-educated adolescents, and non-Western adolescents. In addition, the number of endorsed criteria peaked at 15 years,