causal inference

This post was prompted by this tweet (admittedly already rather old by the time of this posting): The phenomenon observed here shouldn’t be particularly shocking — it’s exactly what we should expect in light of something called Berkson’s Paradox. This paradox, much like its more famous cousin named after Edward Simpson, isn’t really a paradox at all, but it does lead to some counterintuitive observations. In this example, we appear to see a negative correlation between product quality and sales team quality....