Fun with Simpson's paradox

This WSJ pieces does a fine job of showing that sub-groups matter, especially, as in the kidney stone treatment example, when they are endogenous.

There is some implicit confusion here about education and unemployment. College graduates pretty much always have the lowest unemployment rate (they also have the highest opportunity cost of being unemployed). Does the fact that their rate is higher now than in the 1982 recession mean that the current recession is worse or that the now more numerous college graduates are a less selected, and so perhaps less able on average, group, and that both education and ability matter for the probability of unemployment?

Hat tip: my colleague Lones Smith