Thursday, 29. April 2010 14:48
Anthony Downs in An Economic Theory of Democracy, his famous dissertation on political information consumption by citizens, argued that the median voter is the decisive actor in elections and thus rational political parties would locate their platforms at the median voter ‘ideal’ point in the policy space. Some have argued this bodes well for Charlie Crist’s imminent run as an independent, having abandoned the Republican party as a consequence of his failed bid for the Republican nomination. Rubio has run to the Right of Crist and has solid conservative credintials while Crist has always been a moderate Republican. His ideology is likely closer to the median Florida voter than either Rubio or Meeks.
On the other hand, Maurice Duverger, a French Sociologist, put forward the well-confirmed principle that plurality voting systems favor 2 party systems. A corollary of Duverger’s Law is that plurality systems are biased against independent runs – the logic of the 2 party system leads voters to rationally choose among the major party candidates given the low probability that an independent can win.
So we apparently have two mega-theories of political science at odds with one another on the viability of a Crist run. Which is more important? Crist’s proximity to the median Florida voter or his status as an independent? I will argue here that the simple MV interpretation has alot of problems as a model of a 3-way Florida senate race with Crist as an indepdent, and the DL implications for such a run suggest strongly Crist is a loser. Downs’ MVT model is useful for making predictions about how voters acquire political information, and it unquestionably identified an underlying centripital force in party politics that has a great deal of empirical support…but the Devil is in the Details. The parismonious predictions of Downs’ MVT didn’t hold up empirically. Since then, scholars have created more realistic and empirically-supported models of voting that do not dethrone the median voter, but certainly cast doubt on the efficicacy of using Down’s MVT to predict a specific race.
Here I will cover some of the basic modifications of Down’s model and show why MVT certainly does not predict a Crist victory and, in fact, might suggest he will be a loser under conditions we can reasonably forecast. I’ll also talk a bit about what Duverger has to say about Crist. Preview: it ain’t good for Crist.
1. Downs’ model assumes two parties and a unidimensional policy space. If the political compeititon over the senate seat from Florida is multidimensional, the median voter expectation breaks down (e.g. see McKelvey, Plott, Riker, etc. on the instability of social choice in a multidimensional space). For example, many scholars have identified a valence dimension (likeability) that is orthogonal to the unidimensional policy (or ideology) space that Downs relied upon. While Crist might be closer to the MV on the ideological dimension, his switch could seriously damage his likability. He may seem opportunistic and power-hungry…two characteristics voters rarely approve of in a candidate. Not to mention the hit his integrity takes, having disavowed an independent run just weeks ago on Fox News Sunday. Rubio already has a political ad lambasting Crist over his flip-flop. A low valence score by voters could render his position on the ideological dimension irrelevant.
2. In the work subsequent to Downs on voting behavior on the ideological dimension, the multidimensionality of ideology has been identified (Hinich & Munger, 1994) and the salience of issues has become….er, more salient (see Endersby & Gooch, 2010). Indeed, these factors explain why we don’t usually see parties converge to a single point in their policy platforms in real elections (along with incomplete information, uncertainty, abstention & alienation, etc.). It’s hard to predict what issues will be salient, but non-salient dimensions will play a small or non-existant role in voter decisions while salient dimensions will be weighted strongly in that calculus. Crist may be closer to the MV on some issues, but he may be quite distant on others….particularly those that will count in 2010. For example, Crist supported Obama’s stimuls plan, but that plan was quite unpopular with voters, as only 37% supported it weeks after passage and it has become increasingly unpopular since then. According to Pew, six out of ten (62%) say the economic stimulus package enacted by Congress last year has not helped the job situation. On social issues Crist might be closer to the MV in Florida than the other candidates, but if social issues have low-salience in the coming election, then that proximity he enjoys may have little or no weight in the voter calculus. Likewise, if Crist is viewed as ineffective or incompetent on the most salient issue dimension, economic performance, then Downsian (i.e. spatial) logic makes him as much a loser as Duverger does. So Crist could lose due to valence and salience factors, even if, in the aggregate, his ideological placement is closer to that of the MV. In the 2008 American National Election Survey (ANES), a larger percentage of respondents reported liking Obama (51.7%) than they did McCain (48.1%). Obama rated higher on the valence dimension, and that likely contributed greatly to his electoral victory.
3. We don’t know where the median voter is, but in 2010 she’s probably a Republican. It’s true that the median voter is more likely to be an ideological moderate even in a midterm election. But without knowing the composition of the electorate on November 2010 we can only guess at the MV location. Uncertainty as to the MV location is another reason why parties do not converge. So while it is true that Crist is a “moderate” and the MV is likely to be a “moderate” that doesn’t really tell us whether Crist is actually closer to the real MV than the other candidates. Furthermore, the median voter in FL 2010 will be more right-wing than the median voter in Crist’s two previous campaigns. This may be the most right-tilting electorate since 1994. Consider these Gallup numbers: there is a 20 point gap in partisan identification in favor of the Republicans in the “very enthusiastic” category of potential voters for November 2010 (57% R, 37% D). The “not enthusiastic” category has Democrats edging out Republicans by 4 points. Midterm elections are always a bit more Republican than presidential elections because of the differing turnout patterns of the two party constituencies…but this suggests that the electorate may be ALOT more Republican in 2010. That shifts the MV in the Republican direction…in Marc Rubio’s direction.
4. Downs’ model treats voters as exogenous. In other words, it assumes a distribution of voters and then allows the parties to move with respect to that distribution. But it is important to understand what that unidimensional space actually is or isn’t. What it is not is the ACTUAL policy or ideological positions of the candidates. It is the PERCIEVED policy / ideological positions of the candidates. So while Charlie Crist may be objectively more moderate than Rubio or Meeks, that doesn’t mean that voters will percieve him as such. Objectively speaking, John McCain was closer to the MV in 2008 than Obama was. Obama, for example, was at a -0.465 on the 1st dimension of the DW Nominate scores by Poole & Rosenthal (which estimate the ideological position of members of Congress). McCain on the other hand was at a 0.423. Obama was more extreme than McCain, based on his voting behavior in Congress. But that isn’t how voters percieved the two candidates. It is perceived distance…not actual distance that is the determinant of voter’s candidate preferences. 2008 voters may have percieved Obama as closer to the center than he actually is. The same might happen with 2010 voters and Rubio or any of the FL Senatorial candidates. They may be percieved as more extreme or more moderate than they actually are. If Crist isn’t percieved as the most proximate candidate to the MV, then he isn’t for the purposes of the voting calculus.
5. Finally, turning to Duvergar’s Law, Crist faces an enormous challenge running as an independent. Even if he is more proximate to the MV on the valence, ideological, and issue dimensions…we know voter’s factor in the likelihood that a candidate can win into their voting decisions. There’s a reason that most libertarians vote Republican and not for the Libertarian Party. In a plurality voting system, you get nothing for coming in second. It is a winner-take-all system. What this produces, as Duverger noted so long ago, is a two party system. Third parties cannot get traction, even if they have policies and issues a majority of voters prefer, because their expected vote total is nowhere near the threshold of competitiveness.
Now we have independents in Congress for sure (Sanders, Lieberman)…however successfull independent runs fall into two categories: a) the independent is endorsed by one of the major parties (Sanders has the Dems and Lieberman had the Reps behind him) or b) a narrow victory is won where both of the major party candidates were unpopular. In the first situation the “indpendent” is a de facto member of one of the major parties. The second scenario doesn’t seem to be in play here, as Rubio is a very popular Republican and it is a Republican year. Furthermore, the Dems are unlikely to endorse Crist with Meeks already a viable candidate.
Crist will lose financial support, endorsements, and staff as a consequence of leaving the Republicans. As Rich Lowry argues, channelling Duverger, Republican PID’rs will abandon Crist and as this happens his race will increasingly look like a spolier race. From spolier it is but a short stop to “can’t win” and then you run into strategic abandoment. As Nick Silver points out, Crist doesn’t have to slip far to get to “unviable” candidate. Whereas Rubio and Meeks have much higher floors given their regular party vote supporters. In this scenario, Crist supporters abandon his run to prevent either of the other two candidates (depending on whom they most prefer among the two) from winning a close election. Recall in the 2009 New Jersey governor’s race the third party candidate (Chris Dagget) polled into the upper teens and many speculated he might prove a spolier for Christie’s bid. For example, James Carville’s polling firm “Democracy Corps” had Daggett pulling in 15 points. In the end, he only secured 5.8% of the vote and Christie cruised to a 4 point win over Corzine. If voters don’t believe Crist will win, it won’t matter what his positions are or even his valence characteristics. He will lose a substantial number of supporters in running as an independent…and he simply cannot afford them. The best he has done in current polling is eek out a narrow victory in a 3-way race. But Duverger always wins in the end….which means Crist is likely doomed.
In the end, I think Crist is a loser in this race…and I doubt either Downs or Duverger would disagree. I put his vote total, should he stay in the race all the way to election day, at around 18% give or take a point or two…and even that might be high. Be sure to check back in November and see if I’m right.