As John Sides has explained here and here, our Election Lab forecasts are based on a model of Senate election outcomes from 1980 to 2012 combined with polling results from the current cycle. Two factors from the model combine to weigh heavily in the Republicans’ favor in Georgia. First, with a Democratic president the midterm penalty (about 3 percentage points, on average) favors the Republicans and Perdue. Second, in recent presidential elections Georgia has been about 6 percentage points more Republican than the country overall, which our model suggests will translate into about 2 percentage points of the Senate vote for Perdue (compared to a state whose presidential vote mirrored the national vote). Together, then, these two factors give Perdue a significant head start. Obama’s low approval ratings and sluggish economic growth also help Perdue.
Another factor to consider is campaign fundraising. Our model currently is based on relative party fundraising rather than candidate fundraising. That is, we sum up the money raised by all Democratic and Republican candidates running for a House or Senate seat, and then examine the balance between the candidates in each party. The logic is that early in the campaign, party fundraising is better at capturing how eager the supporters are on each side to open their wallets. Moreover, this helps us produce a measure of fundraising even before we know who the general election candidates will be. In our analysis of previous election cycles, we have found that early on this measure predicts outcomes better than candidate fundraising.
In Georgia this year, the Republican Senate candidates have out-raised Nunn by about 3 to 1. Based on our model this gives Perdue another boost of almost 4 percentage points compared to if the fundraising between the parties was even.
Of course, there was a hotly contested Republican primary and primary runoff on the Republican side while Nunn faced no other serious Democratic challengers. If the amount raised by all the Republican candidates combined is not a good indicator of overall Republican strength, then our model is overestimating Perdue’s chances.
Indeed, if we focus on only the funds raised by Perdue and Nunn, they are nearly even. In this case, Perdue’s advantage from our model would be limited to the 5 percentage points based on the midterm penalty and Republican tilt in Georgia. So based on the model, Perdue has an estimated advantage over Nunn of either 5 percentage points or 11 percentage points.
Our forecast also takes into account polling. In Georgia, as in many states, there have not been many polls and none since Tuesday’s Republican primary run-off. In fact, there have only been two public polls since May. In early June, Survey USA gave Perdue 53 percent of the two-candidate vote (that is, 53 percent among those with a preference for Perdue or Nunn). More recently, Public Policy Polling, a Democratic polling firm, had Nunn at 54 percent of the two-candidate vote. Overall, our polling average currently gives the advantage to Nunn.
At this point in the election cycle our forecast gives the model more weight than the polls. That’s why our forecast is more pessimistic for Nunn than the other ones tracked by the Upshot. Currently we give the Nunn just a 2 percent chance of winning. If we were to exclude party fundraising and focus only on her fundraising relative to Perdue’s, her chances would rise to about 30 percent. Either way the advantage is to Perdue. The only question is how much.
Of course, this raises the question whether it would make more sense to rely on candidate fundraising. We continue to feel, based on our analysis of past elections, that right now party fundraising gives a better sense of the race, and that it will continue to do so until after Labor Day. It’s important to give these candidates a chance to “recover” from the primary, especially after a competitive one. Will Perdue bring Kingston’s supporters on board? If he does, we can expect his fundraising to spike. Time will tell, and our forecast will adjust accordingly.
Going forward, as Election Day approaches our forecast will give growing weight to the polls, assuming that more are reported. If Nunn continues to poll well, then our forecast for her chances will, of course, improve. But given the hurdles she faces, a Nunn victory would qualify as a notable upset.