This is a guest post from University of California, Berkeley political scientist Eric Schickler.
There are many good things about journalists and other political observers using concepts and measures from political science as they analyze contemporary politics. But an example this weekend from Nate Silver, the excellent analyst of polls and related matters, offers a cautionary lesson. Silver uses NOMINATE scores to argue that Paul Ryan “is the most conservative Republican member of Congress to be picked for the vice presidential slot since at least 1900. He is also more conservative than any Democratic nominee was liberal, meaning that he is the furthest from the center.” Mr. Ryan may well be the most conservative Vice Presidential nominee in decades, but the NOMINATE methodology is not suited to making claims about the relative liberalism or conservatism of politicians over the long time span invoked by Silver.
An obvious problem is that the scores are comparing Ryan to Republican politicians in the 1910s-20s, who were voting on policies in a pre-Social Security, pre-Wagner Act, pre-civil rights, pre-Medicare world. The issues under debate were so different in this period that meaningful comparisons in terms of “liberalism” and “conservatism” are just not feasible. As the originators of the NOMINATE methodology – Keith Poole and Howard Rosenthal – make clear in their work, the meaning of the underlying scale may well change over long spans of time. A negative score on the NOMINATE scale in the 1920s indicates that a member of Congress agreed with the Democratic party line on the main issues of the day – such as the tariff (favoring free trade)—while positive scores designated Republicans who favored their party’s line on the same set of issues. But these members were not voting on policies that bear much of a resemblance to the issues that define liberalism and conservatism today (and in some policy domains, being a liberal in the 1920s would correspond to being a conservative today; see David Karol’s work on free trade for a nice example).
Perhaps the best example of the deep flaws in this approach is provided by the most liberal Vice Presidential nominee of the past 100 years (as revealed by Silver’s chart of NOMINATE scores) … John Nance Garner of Texas! Garner’s NOMINATE score – if interpreted as a measure of liberalism—certainly seem to indicate that he was a liberal: Garner’s score places him to the left of Hubert Humphrey, Harry Truman, and Al Gore.
Who was John Nance Garner? After a long tenure as a prominent House Democrat, Garner was actually a candidate for President in 1932, when Roosevelt selected him as Vice President. In that race, Garner’s support came largely from conservative forces in the party, who – based on his record in Congress – believed he would shun radical solutions to the Depression. The New Republic commented in March 1932 that Garner was the choice for those seeking “a good, safe politician with an innocuous record who knows the game and how to play it. Unconsciously, what they want is a Democratic Coolidge, and they instinctively feel that Garner is their man. They are not wrong.”
These more conservative Democrats were indeed correct about Garner. By the time of Roosevelt’s second term, Garner was leading the conservative opposition to the alleged excesses of Roosevelt’s New Deal. Garner sought to galvanize conservative southern Democrats with northern Republicans to fight the pro-union, high spending Roosevelt policies that they argued were leading the United States to socialism. Garner emerged as the great hope for conservative southerners seeking to block a third term for Roosevelt; instead, Roosevelt’s forces had control of the 1940 convention and ditched Garner for Henry Wallace. Garner had not moved sharply to the right while in the Vice Presidency; instead, the political agenda had been transformed and he refused to be transformed along with it. Garner’s NOMINATE score reflected his position as a good southern Democrat who shared the party’s dominant philosophy and voting record on the issues at play in the 1920s. To call that liberal is, to say the least, anachronistic.
NOMINATE scores are useful measures for a wide variety of purposes. With care and sensitivity to changes in the political agenda, one can make reasonable comparisons over, say, a thirty-year period (as Nolan McCarty, Keith Poole, Howard Rosenthal, and Chris Hare do here). However, when one seeks to compare levels of “liberalism” or “conservatism” across a span of many decades they can be more misleading than illuminating.
There is no doubt that Paul Ryan is a very conservative Republican. Whether he is more conservative than Republican Vice Presidential nominees of the pre-New Deal era – or even more conservative than the segregationist, ardently anti-CIO Roosevelt foe John Nance Garner – is another matter.
 In their post, McCarty and his collaborators rely greatly on Richard Lugar’s long tenure in the Senate, during which Lugar went from being at the center of the GOP to occupying a position on the party’s left. There is good reason to doubt that Lugar moved substantially to the left during this period and our knowledge of recent American politics tells us that the political agenda as a whole has not shifted to the left from 1977-2012. As a result, Lugar’s movement relative to his party is telling. But requires much stronger assumptions to determine whether Lugar and other Republicans today are more or less conservative than Republicans in the late nineteenth or early twentieth century, when the meaning of liberalism and conservatism were far different than today.