Causation lies undoubtedly at the heart of all explanations. To show a causal relationship is the aim of most endeavours in the social sciences. While experimental designs are still the gold standard for causation (and have become gradually more popular in the social sciences as well) most studies cannot rely on this kind of data. Painting with a broad brush, two other approaches to causation will be made fruitful for this paper. Observational data has to draw on the one hand on Humean regularity approach to causation. It is the “power of the number” that gives us more confidence in claiming causal relationships. The role of the “causality-maker”, however, remains with the theory. On the other hand, non-Humean approaches rely on the deep investigation into a small number of cases. Here, it is claimed that the “causality-maker” lies in the uncovering of causal mechanisms that translate explanans with explanandum.
In this light, the great advantage of QCA lies in showing “conjunctural” or “chemical” causation. In stark contrast to the ceteris-paribus condition which is common to most statistical methods, QCA highlights the configuration of conditions that lead to the outcome. This allows for equifinality, multifinality, and asymmetry. While I do not doubt these great advantages and I do not challenge the conjunctural nature of the QCA-approach, I do, however, raise some scepticism on the causality claim itself. It will be argued that the confidence we gain from QCA-results depends largely on the “power of the number”. Like in quantitative methods, we are in need of a theory that serves as a bridge between correlation and causation. For typical QCA-results, i.e. a configuration of certain conditions, we need an “umbrella-theory” that combines the different conditions into one coherent theoretical frame. As QCA-approaches are often eclectic and aiming at synthesizing different strands in the literature, the quest for this “umbrella-theory” becomes not just challenging, but one of the most pressing issues when engaging in this kind of research.
