167 General Discussion 7 Instead of aggregating data from many individuals, and analytically trying to filter out individual differences, each individual should first be analyzed idiographically, and then after that, findings between individuals can be compared. When idiographic methods are repeated across individuals, nomothetic conclusions for the group can be drawn by describing the number of people for which certain conclusions hold. This bridges the gap between idiographic and nomothetic science. In short, the premise should be: "analyze, then compare" (Hamaker et al., 2005). I believe such a bottom-up analytical strategy to be ideally suited for the study of change in a group as heterogeneous as those with a mild intellectual disability (Márquez‑Caraveo et al., 2021; Nouwens et al., 2017; Sajewicz-Radtke et al., 2022). With EMA data it is possible to “analyze, then compare”, as Chapter 5 demonstrated. One of its conclusions was that 12 out of 26 individuals had a significant positive association between restlessness and nervousness, which is a far more nuanced statement than asserting that this effect does (not) exist on average. This approach would bring the person back into the study of change (cf. Molenaar, 2004; McManus et al., 2023). Describe vividly, infer cautiously Idiographic science comes with its own analytic challenges: how does one analytically 'grasp' an individual's change process? In Chapter 5 we described the shortcomings of idiographic network models in this regard. Essentially, the caveat of network modeling is that it attempts to grasp extremely contextualized, nonlinear, and emergent change processes into a relatively simple, static framework of linear associations. Linear models advance quickly, now accommodating some structural changes within persons (e.g., Bringmann., 2024). While this progress is vital, I do think that clinging to general linear models is also driven by the (human) urge to mathematically grasp, predict, and control reality. There is nothing wrong with that ambition – on the contrary. However, our conclusions from Chapter 6 warrant caution here: the daily life of one person with a mild intellectual disability and behaviors may be too complex to "grasp" within a linear model. It may yield some understanding, but at an abstract level. I fear that forcing linear models onto inherently non-linear phenomena (see Table 1) is akin to forcing square pegs into round holes. Instead of rushing to infer a linear model from the raw timeseries, it may be more beneficial to vividly describe the raw dynamics first. The most powerful descriptive tool we have is data visualization. Idiographic network analyses (Chapter 5) inferred a model of partial correlation estimates from the timeseries and then visualized those intuitively, but our analytical strategy
RkJQdWJsaXNoZXIy MTk4NDMw