Predicting November 21st: What Favors Vitter & What Favors Edwards
We have a runoff... and, for the moment, it appears to be the real deal, a two-party competitive fight between Democrat John Bel Edwards and Republican David Vitter. Several polls released over the past month show either an Edwards lead or a tossup between these two candidates. Something weird is happening in Louisiana.
I write "weird" because this is not what the structural 'fundamentals' (the things that tend to predict elections quite well most of the time) say usually happens. Instead, as of today, it looks like Edwards has a real shot at winning the governorship of a red state. Will it still look that way on November 21st?
When making predictions – or, when deciding how much credence to give someone else’s prediction – I find it useful to think in terms of balancing what the fundamentals tell me about what usually happens with what the immediate evidence tells me about what is happening at the moment. For example, this past Saturday, early in the second quarter, Western Kentucky was tied with LSU 7-7. The immediate information showed me a close game, but I also knew that things usually fall in line with the fundamentals. In this case, LSU was a deeper and more talented team playing at home. So, there was no reason to expect it to remain close despite Western's highly touted quarterback, the rainy conditions, or the tied score in the second quarter.
But, every so often, things do not go the way of the fundamentals. To keep the sports analogy going, you may recall LSU's loss to Kentucky in 2007. These are rare cases, but they do happen.
We have an interesting case in the 2015 Louisiana gubernatorial race – a clash between the fundamentals and the current polling. Which are we to believe? For the time being I would caution you from dismissing either.
Why a Republican Should Be Favored in a Runoff in Louisiana…
1. Partisanship Matters, a Lot. I’ve written on this before. In American gubernatorial elections, 85% of partisans end up voting for their party’s candidate. This is probably the single most important feature of most elections. We like to tell stories about the twists and turns of the campaign, but it almost always comes down to people doing what they tend to do election after election.
To be clear, I am not talking about party registration, which is not a great indicator in Louisiana. Instead, I mean the way people think about their political orientations, the side they think of themselves as fitting with. Louisiana tilts Republican in its voting habits in statewide races because it has a lot of Republicans and a lot of conservative, Republican-leaning independents.
Vitter's favorability rating has declined in recent weeks, including among Republicans. Even so, he has large pockets of support among Republican identifiers, independents who lean Republican, conservatives, Catholics, and born again Christians. These overlapping groups make up a substantial share of the electorate in Louisiana and tend to vote for Republican candidates. That favors Vitter.
2. Obama is Not Popular in Louisiana. Of course, Vitter needs more than just Republicans to vote for him. He’ll need the conservative and moderate independents who tend to vote Republican too. He has not nailed them down yet. In the primary, his net favorability among all independents was negative.
But the fundamentals would lead them to Vitter. Based on Vitter’s election night speech and the recent slate of ads sponsored by the Republican Governor’s Association, John Bel Edwards’s association with President Obama will be a prominent theme of the campaign. This can help Vitter get these non-Republicans who usually vote for Republican candidates. The president is not popular among white voters in the state, giving Vitter an easy way to corral these almost-Republicans.
The days of “all politics is local” are largely over. While there may be limits to how far Vitter can go with this, voters’ partisanship increasingly colors their view of state races in a way much like it does for national races. Consider this: Heading into Saturday’s elections, only 13% of seats in the Louisiana Legislature were held either by a Democrat in a district Romney carried or by a Republican in a district Obama carried. If voters think of the 2015 governor’s race in the same way as they think about national politics, then John Bel Edwards suffers.
… And Why David Vitter Might Not Be Favored
1. Mood about Direction of State. Partisanship may be the chief fundamental in American elections, but it is not the only one. Also important (though not often as important) is the electorate’s mood about how well things are going in the state (or country, or city, or wherever the election is happening). This matters more when an incumbent is seeking reelection, but it matters at other times too.
Right now, the Louisiana electorate appears primed for a change of direction. A majority of voters (56%) think the state is heading in the wrong direction.
In the primary phase of the governor’s race, this did not amount to much. No candidate consolidated the support of these voters. There are a few possible reasons for this. One is that different people are upset for different reasons. Large shares of both Republicans and Democrats say the state is heading in the wrong direction. If these groups are upset for different reasons – perhaps one is concerned about spending cuts while the other is concerned about rolling back tax breaks – then they would not gravitate toward the same candidate.
But the disgruntled mood of voters may have had little effect in the primary because the campaign largely focused on the the three Republicans. Edwards has a record of disagreeing with the Governor Bobby Jindal. Now that the race pits him against a Republican, we may see Edwards try to capitalize on this mood by touting the stands he’s taken against the governor. (Vitter is also trying to tap this mood by running against “Baton Rouge politicians.”)
2. Candidate Quality. After partisanship and overall mood, candidate quality matters too. That is a catch all phrase designed to include things like experience, gaffe proneness, and scandal. Vitter has certain baggage in this regard. In a future post I'll summarize what the political science literature says about the effect of scandal on elections. For now, I’ll just note that the effect varies quite a bit by the kind of scandal, voters' prior attitudes toward a candidate, and media coverage. The best evidence indicates that, on average, scandal costs an incumbent seeking reelection about five points. There is no direct test (that I could find) examining what happens when a candidate is seeking election to a different office.
3. Divisive Primary. I suspect this is the point most of us are focusing on in the immediate aftermath of Saturday’s election. Vitter was caught in a highly negative contest among three Republicans. Will the Angelle and Dardenne voters support Vitter?
There is reason to suspect they may not. The last Reilly Center for Media and Public Affairs poll from the primary (taken late September and early October) did not include the horse race question about vote choice. To work around this, I looked at Vitter favorability and Edwards favorability among the set of voters who liked Dardenne more than any other candidate and among the set of voters who liked Angelle more than anyone else.
People who were more favorable to Dardenne than to any other candidate in the primary are pretty unsure about how they feel about Edwards (86% can’t say), but they really, really, really do not like Vitter (70% unfavorable).
The pattern is similar among those who were more favorable to Angelle than to any other candidate. Edwards has a positive net favorability rating among these voters (35% favorable and 11% unfavorable), while Vitter has a negative net favorability rating (36% favorable and 58% unfavorable).
This might help explain what could be Vitter's biggest stumbling block ahead. He ran just about a neutral net favorability rating during the primary: 38% unfavorable and 41% unfavorable. It is often said that a Democrat needs about a third of the white vote in Louisiana to win statewide. This has been tough for Democrats in recent years. Last year, former U.S. Senator Mary Landrieu secured only 18% of the white vote in the primary. White voters are considered a problem for southern Democrats. Right now, the polling suggests Vitter is having his own problems with them as well.
That being said, the typical pattern after a primary is that people who support a Republican in a primary also vote for a Republican in the next round. Intra-party sour grapes or bad feelings tend to get set aside when facing off against the opposing party. The numbers we saw during the primary today showing a close runoff between Edwards and Vitter are likely due to Angelle and Dardennne voters saying they will not vote for Vitter in a runoff. There is a good chance that this is a temporary reflection of grievances that will subside.
They might return to a Republican candidate (as is usual) if they have other reasons to vote for him. Majorities of both the pro-Dardenne and pro-Angelle voters identify themselves as conservative. So long as Vitter is perceived as the more conservative candidate, these voters have a reason to set aside any distaste they may have for him. Even that may not be enough. If all the Dardenne and Angelle voters return for the runoff, he needs about 80% of them. Vitter needs most of the people who voted for a Republican in the primary to vote for the Republican in the runoff. If the usual pattern holds, he would be fine. But if this time is different...
The hard part of election predictions is knowing how much to weigh the fundamentals versus the polls and knowing how to adjust that balance over the campaign. Right now, when the polls look close but it is still early in the four week runoff phase, we need to keep the fundamentals in mind. There is a general tendency for the polls to move toward the fundamentals - in this case toward a Republican advantage as voters are drawn by the force of partisan and ideological habit (and a lot of anti-Obama ads).
But we also need to remember that today’s polls are not meaningless. They show us that, as of now, other factors are in play. If they remain in play, the contest remains close. Every day that the polls remain tight, the more weight we should shift to them and away from the fundamentals. Elections may follow the fundamentals most of the time, but every now and then the ball bounces the other way.
[NOTE: In an initial draft of this post I wrote, "What’s more, the online betting markets – where people can place money on which party they think will hold the governor’s office – have tightened up quite a bit. As I write this Sunday evening, PredictIt transactions indicate nearly even chances for both the Democrat and the Republican. Just a few weeks ago, a Republican victory had 2 to 1 odds. For further contrast, take a look at the KY, WV, and MS markets. None of those are remotely close." PredictIt seems to have taken down the market for the Louisiana governor's race as of Monday, October 26. So, that is no longer available to cite as further evidence of a close race.]