One issue we have in politics is that a lot of terms get bandied about that lack rigorous definitions. “Polarization” strikes me as a prominent one, but the contention that John McCain is or was a “maverick” suffers from a serious problem. To solve it, John Sides offers a new paper (PDF) by Ben Lauderdale, Princeton grad student, that uses a a Bayesian heteroskedastic ideal point estimator (real talk: I’m not sure what that means) to find legislators who frequently cast votes for reasons other than the main axis of ideological conflict. Thus we can see the rise, the fall, and the re-rise, and re-fall of McCain’s maverickiness:
McCain really was something of a maverick after the GOP takeover of congress. Not a moderate, as such, but someone who broke with his party in unpredictable ways. But then as he ran for the GOP presidential nomination, he got more orthodox. Then enraged by his defeat at the hands of George W Bush, he entered super-maverick phase. Then he started running for president again and became totally normal. Now enraged by defeat at the hands of Barack Obama, he’s totally non-maverick.