166 Chapter 7 of interest, this can also be explored nomothetically by solely employing personalized EMA data (see Olthof et al., 2024). Analysis Analyze the individual first, then compare It is important that research, nomothetic or idiographic, leads to an understanding of individuals. For example, we found Take it Personal! to be effective on average. Hopefully, this group-level finding will lead to the future implementation of Take it Personal! in practice, which will then help many individuals decrease substance use. But how well do these nomothetic findings apply to the individuals? Average group-level conclusions only equal individual-level conclusions when all individuals are homogeneous and if each of their change processes is stationary (the ergodicity assumption; Fisher et al., 2018). Chapter 5 echoed previous studies that emphasize ergodicity to be an unlikely assumption for humans (cf. Fisher et al., 2018; van Os et al., 2019; Wolfers et al., 2018). The only statement we can make about individuals in our study design one based on chance. The medium effect sizes reported in Chapter 2 indicate that there is a chance of approximately 60% that we pick one person who followed Take it Personal! at random who has decreased their substance use more than another randomly picked person from the treatment as usual condition. Although it is perhaps more nuanced, it is evident that such framing is far harder to interpret than stating that we conclude effectiveness (on average). A notable limitation of our research strategy in Chapters 2 and 3 is thus that it does not allow for inferences about for how many of the 34 people who followed Take it Personal! there was effectiveness. This does not mean that all nomothetic research only focuses on average effects. It should be noted that other randomized trials do attempt to analyze effectiveness at the individual first, for example by analyzing the individuals with the reliable change index (although that index compares the individual to a norm, thus assuming that individual change is best understood when contrasting it to the group; Jacobson & Truax, 1991). Most researchers and laypeople believe that average conclusions apply to most people, but it is possible to simulate data that yield group-level effects based on (non)parametric tests, without any single participant showing that effect (McManus et al., 2023). Although I do not expect this to be the case in our study, the theoretical possibility of it is worrying. There are simply no guarantees that conclusions at the group level generalize to the individual level (Fisher et al., 2018; van Os et al., 2019; Wolfers et al., 2018), but this problem can be mitigated by bridging idiographic and nomothetic approaches.
RkJQdWJsaXNoZXIy MTk4NDMw