Who Are the Real Mavericks?

This post will do something extraordinary: it will make you interested in a Bayesian heteroskedastic ideal point estimator.

How do we know whether a legislator is a maverick? Consider this strategy. Several prominent techniques (see here or here) estimate an “ideal point” for legislators using their roll call votes. That is, on what is typically interpreted as a liberal-conservative spectrum, we can tell where that legislator falls, at least relative to other legislators, based on their voting behavior in a particular session.

But no measure of ideology is a perfect predictor. There will always be “classification errors”; legislators can vote in unexpected ways on some votes. It’s easy to identify those votes—they might feature strange bedfellows, for example—but it’s harder to identify the legislators who are consistently more difficult to classify.

These legislators are the “mavericks”—and we begin to identify them thanks to this fancy estimator. The estimator was developed by Ben Lauderdale, a Ph.D. student in the Department of Politics at Princeton. You can read about his estimator in this newly published paper (ungated pdf). Via email, Lauderdale gives this simple definition of a maverick:

These “mavericks” are voting less on the basis of the political dimension that predicts all legislators’ behavior and more on particularistic factors unique to themselves.

Lauderdale shows that this criterion is correlated with being dubbed a “maverick” by the news media, although his data suggest that the news media tend to use the term more to refer to legislators who are “two-dimensional,” e.g., economically conservative and socially liberal. Ben’s estimator goes beyond that to identify people who are truly voting in unique ways—the real mavericks.

So who are they? Well, for starters, they’re not John McCain. At least, not any more. Lauderdale’s estimator generates a score for each legislator in each Congress, where higher values indicate that these “particularistic factors” matter more. Call this the “Maverick Scale.” In each Congress, the average legislator scores 1 on this scale, so values higher than 1 indicate that a legislator is more mavericky relative to others. Here are McCain’s scores for the 100-111th Congresses (with incomplete data for the 111th, obviously).

So while McCain was in some years substantially more mavericky than his peers, this has not been true in the most recent Congresses, where McCain is not much different than average. Lauderdale says via email:

His maverick behavior seems to have peaked in the 107th (2001-2002), just after losing the nomination to Bush. I am not the first to notice this. Further supporting the theory that McCain is simply prone to contrariness is the fact that he didn’t really become much of a maverick until the Republicans got control of Congress (although there is some hint of an upward trend before that).

So if not McCain—or no longer McCain—then who? Who are the real mavericks? Here are the 10 legislators in both the current House and Senate whose roll call votes derive the most from idiosyncratic factors, which are not in the linked paper above but which Lauderdale kindly provided. I plot the same score as in the above graph, where 1 equals the average member of the 111th Congress.

The prominence of Paul and Kucinich as House mavericks seems intuitive. On the Senate side, maverickiness can be found on both the left (Feingold, Sanders) and center (Bayh, Nelson). The presence of Harry Reid might seem surprising, but this largely reflects votes he makes to retain the right to reconsider.

If you go back to the previous Congress, many of the same people emerge as mavericks. Here are the “Top 10” of the 110th House:

1. Paul (R-TX)
2. Kucinich (D-OH)
3. Cooper (D-TN)
4. Flake (R-AZ)
5. Duncan (R-TN)
6. Stark (D-CA)
7. McDermott (D-WA)
8. Blumenauer (D-OR)
9. Taylor (D-MS)
10. Jones (R-NC)

And the 110th Senate:

1. Feingold (D-WI)
2. Voinovich (R-OH)
3. Hagel (R-NE)
4. Reid (D-NV)
5. McCaskill (D-MO)
6. Byrd (D-WV)
7. Gregg (R-NH)
8. Kyl (R-AZ)
9. Coburn (R-OK)
10. Bayh (D-IN)

You can still see that maverickiness can come from the ideological extremes and from the middle.

Lauderdale’s paper also illuminates notable mavericks from history (such as Proxmire), provides insights into the Iraq war vote and the EU Parliament, and gives us this fascinating look at Barney Frank—whose maverickiness declined substantially when he was under an ethics investigation (the shaded Congress on the graph):

Lauderdale writes:

The timing is unmistakable: Frank faced potential expulsion or censure from the House during the 101st Congress because of a scandal. The threat of formal punishment from the chamber appears to have induced Frank to temper his general legislative idiosyncrasy.

For much more, see the paper.

4 comments