And yet if P does not necessarily lead to Q, it calls into question the value of the original observation, does it not? Where's the benefit of observing that "Every serial killer at some time ate bananas?" As you so correctly point out, eating bananas does not lead to serial killing. So the value of the observation is, at the very least, questionable.
-edit-
That said, the article linked in the OP is insane. Had to check twice to make sure it wasn't dated April 1.
Bookmarks