Maartje Boer

CHAPTER 2 54 adopt addiction-like behavior regarding their SMU, such as loss of control or interference with daily activities (Griffiths, 2013; Kardefelt-Winther et al., 2017). Hence, the distribution of the sum-scores as observed in the present study supports the validity of the test score interpretations. Strengths, Limitations, and Future Directions This study has important strengths related to the nationally representative character of the data and the number and variety of psychometric tests supporting the reliability and validity of the SMD-scale scores and interpretations. Yet, there are limitations that constitute promising directions for future research. First, the present study used a large sample of Dutch adolescents aged 12–16. To establish the generalizability of our findings in other countries and age groups, research using cross-national assessments of the scale among different age categories is required. For instance, a CFA conducted among a sample of 903 Chinese university students aged 18–23 suggested that the scale measured two factors, with the items problem, deception, and conflict representinga separate factor (Fung, 2019), suggesting that the factor structuremay differ across age-groups and/or cultures. Second, the nature of the sample did not allow for clinical validation. Research using clinical samples is required to verify whether the SMD-scale is feasible as a diagnostic tool that accurately identifies problematic users. Third, IRT-analyses showed that the test scores were most reliable for values above the mean of the latent trait, suggesting that the scale provides more precise estimates at higher levels of problematic SMU than at (more common) lower levels of problematic SMU. Hence, the SMD-scale may be most suited to identify moderate to high levels of problematic SMU. This finding is not uncommon for scales that measure exceptional or rare behaviors. For example, validation studies of substance-related disorders and internet gaming disorder scales showed that these scales provide most information at the higher end of the scale’s continuum, that is, for scores that exceed the sample mean (Gomez et al., 2019; Martin et al., 2006; Saha et al., 2006). Fourth, adolescents’ test scores were based on self-reports, which may deviate from their actual behaviors. For example, adolescents may under- or overestimate the extent to which their SMU impairs important life domains. Comparing parent and adolescent scores on the SMD-scale may provide novel insights into the social reliability