Remember in November how everyone got super mad at the pollsters for overestimating Democrats’ standing for the second straight election? Well, there are two pretty big runoff elections in Georgia on Tuesday that will decide which party controls the U.S. Senate, and guess what? The polls once again give Democrats a slight edge.
As things stand now, most polls have the runoffs between Jon Ossof and Sen. David Perdue and Rev. Raphael Warnock and Sen. Kelly Loeffler as virtual tossups, with Democrats Ossof and Warnock holding slim leads. As of January 5th, FiveThirtyEight’s polling average has Ossof leading the Republican incumbent Perdue 49.1 percent to 47.4 percent, and Warnock leading Republican incumbent Loeffler 49.4 to 47.2.
Both races have been more or less dead even since the initial November 3rd election. They were even closer than they are now until about a week ago, when Ossof and Warnock crept ahead by 1-2 points, likely a result of Perdue and Loeffler’s entanglement in President Trump’s very public whirlwind of delusional regarding the the state’s voting system, as well as a fuller picture emerging of their own incompetence.
So, the polls think it’s going to be close, and maybe even that the Democrats are favored to win? Great. They also thought Jamie Harrison was going to give Lindsey Graham a run for his money in neighboring South Carolina (he lost by over 10 points), that Sara Gideon was going to unseat Susan Collins in Maine (she did not come close to doing so), and that Cal Cunningham would knock off Thom Tillis in North Carolina (again, no).
It wasn’t just the Senate. The polls called Biden’s win, but drastically underestimated Trump’s support, particularly in the Upper Midwest, the same place they drastically underestimated his support four years earlier. Again, not good.
Why, then, would anyone pay an iota of attention to the polls coming out of Georgia ahead of Tuesday’s runoff elections?
One reason is that despite notable swings and misses in key swing states, the polls were pretty accurate in predicting how Georgia was going to vote in November. Biden and Trump were just about tied in the polls, and they ended very close to that after the votes had been counted (Biden won by less than 12,000 votes out of the nearly 5 million cast). “We did a great job in Georgia,” Don Levy, director of the Siena College Research Institute, told us in November in explaining why the polls missed so badly. “We had it tied and it is tied, for all intents and purposes. Others did pretty good in Georgia, as well.”
As for the Senate, the polls had Ossof and Perdue about even. Perdue won by close to two percentage points. They had Warnock winning by a greater margin than he did, but not dramatically, and the fact that there were three formidable candidates instead of two likely made things more difficult to predict. In the end, the polls were off in Georgia, but not by very much.
It’s unclear how much this relative success in the state in November will translate to success this week — if it will at all. There are a few new variables in play, to say the least.
For one, Trump is not on the ballot this time. This could be a good thing for polling accuracy. “There might be something about Donald Trump as a phenomenon that is just so different that it makes it difficult to get an accurate picture of whether people want to hide their feelings about him, or whether they simply don’t want to talk about it at all, or whether he has a greater level of support among people who basically reject participation in the normal institutions of political behavior,” says Monmouth University Polling Director Patrick Murray. “It might be the first time we’ve had a populace like this [to the extent that] it might have an impact on polling, and we might not see it again without Donald Trump in the mix.”
Murray points to 2018 as an indication that “Donald Trump” is indeed the problem. “Without Trump on the ballot two years ago in these congressional races, we were able to poll them fairly accurately,” he says. “The factor seems to be that when the name ‘Donald Trump’ is part of the ballot test, that’s what seems to impact it.”
Trump may not be on the ballot this week, but he’s certainly played a large role in the runoff campaigns — from pressuring state officials to overturn the election results, to hosting rallies Perdue and Loeffler, to incessantly promoting conspiracy theories about the state’s election system. Gauging support for Trump in the months leading up to the general election could be child’s play compared to gauging how his weeks-long effort to sow doubt in the state’s ability to carry out the democratic process could affect conservative turnout. All of this is taking place within a compressed campaign timeliness against the backdrop of actual issues like congressional gridlock over pandemic relief and a rising Covid-19 death toll. It’s an unprecedented calculus of issues to consider, and it’s entirely likely that fence-sitters across the conservative spectrum have been changing their minds weekly if not daily about whether to vote with the party, whether to vote Democrat as a rebuke of Trumpism, or whether to even vote at all.
There are some early signs that appear to favor the Democrats. Three million early votes have already been cast, more than the record 2.1 million recorded in the state’s last Senate runoff election, in 2008. Early voters tend to lean Democratic, and as Nate Cohn of the New York Times noted last week, this has indeed been the case prior to the runoffs in Georgia. The question on Tuesday is whether enough of the MAGA sect will turn out to vote for Perdue and Loeffler to overcome whatever early gains the Democrats may have made. The polls say it’s going to be close. We feel more comfortable saying we have absolutely no idea.