The Lion and Gun

Nate Silver at FiveThirtyEight.com, in a follow-up to an earlier piece defending RealClearPolitics' decision to exclude the Research 2000 daily tracking poll from its averages, now doubts RCP's explanations for its sometimes dubious choice of polling to present in its widely-touted averages.

Ironically, perhaps the most important lesson imparted by FiveThirtyEight's borderline-fetishistic polling panoply is to treat all numbers with a skeptical gaze. Differences in methodology and the wording of questions, statistical noise and, yes, cherry-picked numbers all render statistics a helpful but untrustworthy friend.