160 Chapter 7 Design Design research to accommodate non-stationary change Chapter 5 illustrated the difficulty of capturing non-stationary processes statistically, but the non-stationary nature of change processes also has consequences for the research design. To assess change, at least two measurements per individual are needed. In clinical trials the convention is to use this minimum amount of two measurements to assess effectiveness: immediately before and sometime after an intervention (although sometimes there are additional followup measurements). With this design, one can answer research questions like "Is this intervention effective in reducing substance use?" (Chapter 2) or "Is this intervention more or less effective in reducing substance use for people with higher or lower levels of aggression?" (Chapter 3). The focus is on average change. This can certainly reveal interesting nomothetic change patterns, but designs with retrospective surveys bring issues at the idiographic level that extent beyond the typically evaluated psychometric qualities of a survey (e.g., the SumID-Q (VanDerNagel et al., 2011) and the YSR (Achenbach, 1991) from Chapters 2 and 3). In such retrospective surveys, participants reflect on a process and aggregate this process into one survey response. This requires a good memory, comprehension, and abstract thinking – skills that are challenging for people with a mild intellectual disability (Kooijmans et al., 2022). However, even for people without a mild intellectual disability retrospective assessments can contain bias. For example, Leertouwer et al. (2022) found that approximately half of participants' retrospective assessments of variables that they had previously self-rated in EMA, were either higher or lower than the average of their actual EMA data. But even if we can assume that most people with a mild intellectual disability are able to retrieve the average of a process, it is only sufficient when the process was stationary.
RkJQdWJsaXNoZXIy MTk4NDMw