| Abstract: | In this commentary we discuss and elaborate on a number of topics that have been raised in the articles in this issue. The first topic treated is causality. All scientists are aware of the difficulties of making causal inferences in non‐experimental settings, and in many cases arguments are forwarded that support the causal interpretation of the reported relationships. Questions of bi‐directional causality, nonlinear relationships and the importance of omitted variables are discussed, as is the supporting evidence that in some cases comes from experimental studies. The second topic is the level of analysis employed in the empirical analyses and the difficulty of interpreting findings at a different level of aggregation than the level applied in these analyses. In this context, the importance of applying a person‐orientation is pointed out to ensure that the findings are interpretable at the level of the single individual. Even in individual‐level analyses, this can be problematic if one relies on group statistics produced within a variable‐oriented framework. The third topic concerns how to handle interactions and nonlinear relationships. The fourth topic concerns the importance of improving measurements so that a good correspondence is achieved between them and the theoretical conceptualisations. The fifth and last topic is concerned with the argument that different ways of investigating the generalisability of the results to different populations should be tried. |
