Description
As we write in the introduction, we cannot provide indicators for causal effects for all possibilities. This is because there a combinatorial number of possible effects (from various aspects of Open Science on various types of impacts), and each effect requires a separate approach in and of itself. That is, based on observational data, creating such a causal indicator requires careful thinking about what the relevant mediators, confounders and colliders are, and how they could then be controlled for in order to construct an indicator that reflects causality. Based on experimental approaches, it would of course be easier to draw causal conclusions without necessarily requiring complicated causal indicators.
The idea now is that we could prepare a limited number of causal indicators, preferably at least one for each section (academic impact, societal impact, economic impact and reproducibility). That way, we at least demonstrate what could be possible if a proper causal model is developed for a particular case. Of course, it is possible that for some cases it would simply be impossible to infer causality based on observational data, in which case, no such causal indicator could be developed of course. We should also make sure to clarify this limitation in an introduction.
This issue would involve all of the contributors, at least in order to think of potential causal indicators.