The Georgia runoff is upon us! Today, the last U.S. Senate election of 2022 will finally be decided. And according to the polls, Democratic Sen. Raphael Warnock is slightly ahead of Republican Herschel Walker — but the election is still very close.

During the Nov. 8 general election, Warnock received 49.4 percent of the vote and Walker 48.5 percent. Although other Georgia Republicans won their races comfortably, Warnock’s advertising and fundraising prowess and Walker’s many scandals and controversies helped send this race to a runoff. (Georgia is one of the few states that requires runoff elections when neither candidate gets a majority of the vote in the general election.)

We aren’t issuing a forecast for this runoff as we did for the general election, but we have calculated a polling average. And as of Monday at 4 p.m. Eastern, Warnock was averaging 50.2 percent support, and Walker was averaging 48.2 percent. Again, this isn’t a prediction of the final results, just a summary of the polls conducted of the runoff.

The polls in Georgia ahead of November’s election were pretty accurate. In our final polling average of the state’s election for governor, Republican incumbent Brian Kemp led by 7.9 percentage points. He then won the election by 7.5. And in our final polling average of the U.S. Senate general election, Walker led by 1.0 point, which was 2.0 points off the actual election result. Although the polls had the wrong candidate in the lead, they were mathematically quite close to the actual margin. Polls are meant only to approximate the final result, not predict winners.

So, of course, Warnock’s 1.9-point polling lead is not ironclad. Although most of the nine runoff polls give Warnock the lead, every single one has been within the margin of error. (Because margin of error applies to each candidate’s vote share, a candidate’s lead has to be almost twice as large¹ as the margin of error to be statistically significant.)

What’s more, a few of these polls have been from pollsters without established track records. That doesn’t mean they will be inaccurate, but it is a reason to take them with a grain of salt. For example, Phillips Academy just started polling this year, so they don’t have a FiveThirtyEight pollster rating, our assessment of how accurate and methodologically sound pollsters are.² The same is true for Patriot Polling, which only surveys people via landline telephone (not cell phones), a methodological red flag.

In addition, Frederick Polls is a Democratic pollster that rarely releases public surveys. From 1998 to 2020, it released only seven polls that qualified for pollster-rating consideration, earning them only a provisional grade of B/C. Finally, we’ve analyzed 20 polls from Impact Research (under its former name, ALG Research), but only four from Fabrizio, Lee & Associates. Moreover, the two organizations’ partnership is new this cycle, so we haven’t had a chance to judge how accurate their polling is in combination.

However, five surveys from more established pollsters (Emerson College, InsiderAdvantage, SSRS, SurveyUSA and the University of Massachusetts Lowell) all consistently put Warnock a few points ahead of Walker. That said, it wouldn’t be shocking to see an unexpectedly strong performance from either of the candidates. Other factors agree with the polling that this race is tight, and Warnock’s polling lead is still smaller than the average polling error in U.S. Senate races since 1998.³ While Warnock may be better positioned than Walker, either candidate could still win.