Välkommen till föreläsningen i seminarieserien i matematik.
Given two correlated random variables, the common cause principle recommends looking for the third random variable thatjointly causes them, i.e. explains away their correlation. One obstacle in applying this principle is that details of the third variable (i.e. the cause) are frequently unknown. I will review our recent work showing how the cause can be inferred from the generalized maximum likelihood principle. I will also outline the application of the common cause principle to Simpson's paradox, which is a serious obstacle to establishing a probabilistic association between two events. I will argue that if the principle applies, it offers an efficient resolution of the paradox.
