Despite all the turmoil in the country this year, our forecast for the presidential election has been remarkably stable. Since June 1, the first date we made the forecast, at no point had only two states switched between Donald Trump and Joe Biden: North Carolina and Ohio. On Wednesday, however, a third state was added, Georgia, where Joe Biden is the favorite for the first time of the year – the smallest favorite! – In our forecast.
Of course, it’s a bit silly to talk about states switching sides when FiveThirtyEight’s predictions are probabilistic. With the addition of a new poll from Quinnipiac University Biden’s chances in Georgia rose from 46 percent to 51 percent as he was 7 percentage points ahead there – more on this poll. (They have since dropped to just 50/50.) So a) this isn’t a really big change, and b) the race is a throw; Trump could easily become the nominal favorite again (Georgia turns red again on the snake chart) if you read this.
In another sense, however, the shift in Georgia should come as no surprise. Consider:
- In 2016, Trump won Georgia by just 5 points in an election in which he lost 2 points in the national referendum. Georgia was only 7 points more republican than the whole country.
- In 2018, Brian Kemp won the Georgia gubernatorial election by 1 point over Stacey Abrams. In this election, the Democrats won the referendum for the US House with 9 points. Georgia was 10 points redder than the country as a whole.
And if Georgia is 7-10 points more republican than the whole country, then you’d expect it to turn blue when Biden has a 10-point lead in national polls, as he is currently doing.
Our model, however, was a little more skeptical of Biden’s odds in Georgia for several important reasons. First is the state relatively inelasticThis means that it doesn’t vibrate as much as the whole country. That’s because Georgia has traditionally not had many swing voters. Republicans have a base of conservative, religious, white voters; Democrats have a base of black voters as well as younger white, Asian, and Hispanic voters in Atlanta and in university cities. The Republican base was a bit larger historically, though, so while Democrats could hit 45 percent, or 48 percent, or 49 percent nationwide … 50 percent was tough.
Second, Georgia has strict electoral lawsand after the Voting cost indexOur laws show that passing stricter electoral laws also harms Democrats, which is a measure of how easy it is to vote in any state. This is taken into account in our estimates of the partisan base in each state.
Even if the suburbs of Atlanta have shifted blue – leading to a 2018 victory for Democrat Lucy McBath in the 6th Congressional District in the upscale northern suburbs of Atlanta – these are still some pretty big hurdles Democrats have to overcome to achieve national victories. That’s why Biden even though he was already ahead of our Georgia polls forecast – which also takes these other factors into account and slightly constrains Biden’s current national leadership – Trump was still slightly favored in Georgia.
Then came the Quinnipiac poll. What should you think of a poll where one candidate has a 7 point lead when everyone else tied, or practical? Well, you should … toss it into the survey average. In general, it is good for quality pollsters like Quinnipiac to be willing to deviate from consensus. it means they don’t focus on what everyone else is saying.
At the same time, Quinnipiac might have produced the best numbers of any major pollster for Biden. Another recent release of Quinnipiac polls Biden was 11 points ahead in Florida, 13 points in Pennsylvania, and 5 points in Iowa, far better than the FiveThirtyEight averages for any state.
I’m not the one to play the doctor of choice, delving into the crosstabs and explaining whether a particular company’s approach is right or wrong. I think the right approach is to trust the process of averaging the polls unless there is something outrageous. At the same time, a 7-point lead from Quinnipiac doesn’t quite mean the same thing as from a pollster like Monmouth University, who tends to show results for Biden that are close to or slightly worse than the consensus. (Monmouth’s recent Georgia poll – conducted prior to the debate – had shown Trump a slight head start.)
Our model tries to account for all of these tendencies with our adjustment of the home effects. it explained the fact that Quinnipiac had before that, Biden showed a 3 point lead in GeorgiaFor example, at a time when Trump was in the lead there in our survey average. Even so, a 7 point lead was enough for Biden to impress our model, even if it came from Quinnipiac.
Next question: While Georgia could obviously go either way, it could indeed be crucial when choosing? In other words, it could be Turning point condition, the one who gets Biden or Trump to get his 270th vote?
That’s pretty unlikely: the model says there is only a 2.4 percent chance Georgia is the tipping point. That’s because if Biden won Georgia, he likely also won at least one of his neighbors, North Carolina or Florida, where he has slightly clearer leads in polls. And if Biden wins North Carolina or Florida, Trump is likely a toast with or without Georgia.
In theory, Georgia could play a role if Biden collapses completely in the Midwest: if he loses Michigan and Minnesota and Pennsylvania and Wiscosnin then Biden could still win the electoral college by turning around from Georgia and North Carolina and Florida, while the other states hold, which Hillary Clinton won. However, we are again talking about some long-term scenarios.
But where Georgia could be of great concern is in the US Senate, where there are actually two races: one standard Class II Senate elections between Republican incumbent David Perdue and Democrat Jon Ossoff, plus a special election in which appointed Republican incumbent Kelly Loeffler defends her seat against challengers from both parties.
Georgia needs runoff elections if no candidate gets 50 percent of the vote. Therefore, the Perdue-Ossoff race could potentially require a runoff election if the libertarian candidate gets enough votes there. And the special election will very likely require a runoff because it has more than a dozen candidates and no candidate there is particularly close to 50 percent of the vote.
Recent developments have also had a positive impact on the Democrats in the special elections. One Democrat – Raphael Warnock, the pastor of the Ebenezer Baptist Church – has consolidated the vast majority of the Democratic vote after advocates from Barack Obama and other prominent Democrats. Based on the latest polls, it is very likely that he will advance to the runoff election. Meanwhile, the second runoff slot is a bitter feud between Loeffler and another Republican, Rep. Doug Collins, who tried to outdo each other brag about how conservative they are and how much they support Trump. This is potentially toxic news in an increasingly purple state that Trump is not as popular in. In fact, Loffler and Collins have fallen behind Warnock in some recent polls that tested individual combat.
Our model tries to account for all of these complexities – as well as the likelihood that the environment for Democrats won’t be as favorable for January as it was in November – and now has Warnock with a 50 percent chance of winning, Loeffler with 29 percent chance, and Collins with 21 percent Chance. The other candidates like the Democrat Matt Lieberman have practically no chance.
In the Georgia Senate regular race, Perdue has a 72 percent chance and Ossoff has a 28 percent chance, according to our “deluxe” forecast. There is a one in four chance that the race will also require a runoff. This could mean there could be two uncalled Senate members by the January 5 runoff.
So there are certainly no safe things for Democrats in Georgia. But the fact that a formerly red state has become perhaps the most competitive battlefield in the country is a bad sign for Republicans.