Causality and Intuition

The theory of causality described in the previous chapter was derived from three assumptions. In this chapter, the theory is confronted with intuition concerning the behavior of a causal system and of observations arising from it. First, Berkson’s paradox, as an example of the causal effect of two desirable features on selection, is discussed. The paradox is loosely formulated as selection makes two originally independent features dependent. The phenomenon is given a precise description and its relationship with conditioning on a collider in d-separation is described. Then, a very simple type of realizations that apply to every causal structure, called incremental probabilistic causation, is discussed. Using this simple example, many of the theoretical results have direct proofs and intuitive meanings. But one has to be careful as to how far intuition may be relied upon in the context of causal analysis. The last section of the chapter considers a number of claims often made by researchers applying this causal theory, and while these claims may seem intuitively appealing, they are shown not being implications of the theory presented here.