332 Chapter 8 In summary, the results from Chapter 6 and 7 support the idea that explicitly promoting video games for mental health present a promising approach to reach individuals with mental health symptoms, even those less inclined to seek professional help. Video games, specifically designed to improve mental health, can be positioned and promoted as mental health games, leveraging their motivational and engaging potential to effectively reach and benefit individuals with mental health symptoms. LIMITATIONS AND FUTURE SUGGESTIONS When interpreting the results across the various chapters of this dissertation, it is crucial to acknowledge that certain conclusions are drawn from null findings. These conclusions pertain to the findings that nonspecific factors do not influence the effectiveness or engagement with MindLight (Chapters 4 and 5, respectively), nor individuals’ choice for and engagement with a mental health game (Chapter 7). In this context, null findings signify instances where statistically non-significant results are interpreted as supporting the null hypothesis, indicating ‘no difference’ or ‘no effect’ (Tabachnick & Fidell, 2014). The justification for such interpretation rests on the premise that the study possesses sufficient statistical power to reject the null hypothesis when it is genuinely false (Leppink et al., 2017). It is essential to recognise that a lack of statistical power can lead to a Type II error, where the null hypothesis is not rejected even when it is not true. With low statistical power it remains unclear whether a non-significant result means that there is actually no effect (i.e., null hypothesis is true) or whether there is not enough power to reject the null hypothesis. In psychology research, the commonly sought statistical power is around 0.80, balancing the need for accurate conclusions with the practical feasibility of the study (as striving for higher statistical power would mean that more participants are needed) (J. Cohen, 1992; Picho & Artino Jr, 2016). The sample sizes in the studies conducted in this dissertation were determined using a-priori power analyses, aiming for 80% power (i.e., 1-β), with a (conventional) significance level of 0.05, and the capacity to detect a medium effect size. This implies that there is an 80% probability of correctly rejecting the null hypothesis if it is false, instilling confidence in the results obtained in this dissertation (Lakens, 2022). Nevertheless, future research is needed to replicate our findings.
RkJQdWJsaXNoZXIy MTk4NDMw