Vitter’s Republican Problem: Will a Divided Party Hurt Him in the Runoff?
One of the major determining factors in the Louisiana governor’s election will be whether or not Republican voters coalesce behind Senator David Vitter. This is the latest twist in what increasingly appears to be an unusual election. So often, discussions of high profile elections in Louisiana seem to focus on the Democrats’ base: Will black Democrats turn out? This year turns that on its head. It looks more like a question of whether Vitter can persuade his base – Republican voters – to support him.
Although Louisiana does not have party primaries, the first round of voting on October 24th effectively served as a party primary for the three Republican candidates. As the lone serious Democrat in the race, Representative John Bel Edwards was all but assured a spot in the runoff, leaving the Republicans to vie for the remaining spot. After a primary campaign featuring internecine attacks among Republican candidates, the party's voters were left divided. The three Republican candidates combined for a majority of all votes, but no single candidate won a majority of Republican votes. Vitter garnered 40% of all votes cast for Republican candidates, allowing him to narrowly edge out Angelle for a spot in the runoff by just four points. Reportedly, one of Vitter’s former opponents from his own party will now endorse Edwards.
Vitter will need a substantial portion of Angelle and Dardenne voters to cast a ballot for the sole remaining Republican candidate. Imagine for a moment that exactly the same people who voted in October show up to vote in November. Imagine also that no one who backed Edwards in the primary switches to Vitter or vice versa. In this scenario, Vitter would need about 79% of Angelle and Dardenne voters to support him to win a majority among all voters.
[The actual target will likely be higher. Blacks made up only about 28% of the primary electorate. If Democrats launch an effective GOTV effort to boost black turnout, Vitter will likely need more than 79% of Angelle and Dardenne voters to back him unless he finds new voters of his own.]
So far, Angelle and Dardenne voters are not flocking to Vitter. The first of three new polls released in the past week, conducted by Democratic-leaning Anzalone Liszt Grove Research (538 grade of C-) and sponsored by the anti-Vitter Gumbo PAC, pegged Vitter at 46% among Angelle and Dardenne voters. Another poll, conducted by JMC Analytics (538 grade of D+) and sponsored by WVLA, put Vitter at just 32% among all the “also ran” candidates. Finally, Market Research Insight (538 grade of C+) had Vitter at 48% among Angelle voters and 43% among Dardenne voters.
Why This Might Not Turn Out to Be As Bad As It Seems for Vitter
Divided primary contests do not typically tank candidates in a runoff or general election. Louisiana has had too few gubernatorial runoffs to provide enough data to thoroughly examine this question specifically in the context of this state, but it is possible to look across gubernatorial elections from all 50 states. I took all gubernatorial runoff or general elections in the U.S. since 1980 in which a candidate had to first win a primary election. [I am excluding any Louisiana elections that did not include a runoff and any candidates in other states who did not face a primary challenge.] I compared the vote share in the general/runoff election to their primary margin of victory over the candidate from their own party with the second highest vote total.
Generally, candidates with smaller margins of victory in the primary get fewer votes in the runoff/general election. On average, every point in the primary margin of victory is associated with eight-tenths of a point in runoff/general election vote share after accounting for state and election year. This does not necessarily mean that a closely divided primary electorate causes them to do worse in the next round; it could simply be that candidates who struggle in the first round also struggle in the next round. Still, whatever the cause, candidates with narrow primary wins tend to suffer a penalty in the runoff or general election. No surprise there.
What’s probably more important is that the difference does not seem to matter all that much in determining who ultimately wins the runoff/general election. Winning a close primary contest (defined as a victory margin of less than five point) has no statistically discernible association with the chances a candidate actually wins the runoff/general election. In other words, the electoral cost of divisive primaries are generally not important enough to tilt the outcome.
Why is the relationship between primary performance and runoff/general election performance so limited? Regardless of how a voter who backed a losing candidate in the primary feels about the candidate from his or her party who won the primary (and polls currently show Dardenne and Angelle voters are not particularly favorable toward Vitter), that voter still has other reasons to support his or her party's candidate. For example, in this year's election in Louisiana, these voters have already revealed a predisposition to vote for a Republican (because many of them are themselves Republicans or independents who lean closer to the Republican Party). Similarly, majorities of those who liked either Dardenne or Angelle the most among the primary candidates identified themselves as conservative.
As I’ve noted before, in recent American gubernatorial elections, 85% of partisans on average end up voting for their party’s candidate. In current polls, Vitter falls below this mark... way below... way, way, way below. He is currently polling at anywhere from 60% to 72% among Republicans. That would be enough to doom him if those numbers held through Election Day. But there is reason to doubt they will.
It is almost unheard of for a candidate to perform among his own party. Sure, it happens, but only rarely. Since 2006, gubernatorial candidates across the US [or at least in states that featured exit polls] carried 72% or fewer of their own partisans in about one out of every seven elections. They won 60% or less in just one out of every 25 elections.
In other words,it would be very unusual for Vitter to do as poorly among Republicans on Election Day as current polling suggests. If, instead, the typical pattern holds then we’ll see Republican-inclined voters gravitate to Vitter and, as a result, his chances of winning would rise.
Why This Might Turn Out to Be Every Bit As Bad As It Seems for Vitter
There are at least two reasons to not dismiss the polls showing Vitter's struggles with Republicans. First, this may be an even more divided party than what’s captured in a generic competitive race. My analysis of divisive primaries above is based on competitiveness alone, and incomplete measure that ignores the tone or tenor of the campaign. After all, one of the Republican candidates who lost in the primary is now endorsing the Democrat. That degree of intra-party division is probably rare. So, maybe this is the one out of seven or one out of 25 elections when a candidate does very poorly among his own party. [Update: The cross-party endorsement is so rare I could find no solid social science evidence about what the effects of this kind of endorsement typically are. It does appear the endorsements by the party matter, so it seems logical that a leading party figure who endorses the other party's candidate could matter too - perhaps by giving his former voters cover to resist the gravitational pull back to party.]
Second, and probably more important, the clock is starting to run down on Vitter. It’s easy (and usually smart) to downplay early polls from six weeks or two months before an election if they clash with predictions based on the 'fundamentals' like the usual partisan voting patterns. It makes less and less sense to do so as we get closer to Election Day.
Polls are better predictors of election results the closer they are to Election Day. My analysis of one firm’s polls (Public Policy Polling) from across a number of governor and senate races last year shows that early polls miss the Election Day vote margin between the top candidates by more than eight points on average. The average “error” drops by half that amount for polls taken the week before an election.
We’re now barely more than two weeks away from the runoff. The polls today are not necessarily great predictors of what will come on Election Day (as a cautionary tale take a look at Kentucky), but we are in the phase of the campaign when polls tend to become better predictors of actual vote margin.
If things are going to shift back toward the fundamentals - and in Louisiana the fundamentals favor a Republican win - we should start to see that happen soon. If, instead, over the next week to ten days we continue to see similar numbers as we have seen in prior weeks, then that would truly be a sign that this race is atypical. [Unless, as in Kentucky, things shift toward the fundamentals but the polls miss it. Before tossing the polls out the window as useless, recall that they did pretty well in Louisiana's primary.]