Stop me if you’ve heard this before: A massive early vote lead. A higher minority share of the electorate. Large turnout in the suburbs. Favorable polling for the Democrats. Almost every possible factor pointing to a Democratic win.

We all know exactly where we saw this last time. It was November 3rd, and we were all settling in to watch the results from the US Presidential Election, and what we all thought would range anywhere from a decisive Biden win to a historic blowout. It turns out that almost everything we thought about that election was wrong, with the only correct prediction being the ultimate winner. Trump would make large gains with those minority voters that were turning out in droves. With Democrats setting an unbelievably high bar for what Trump needed to win the election day vote by to hold many states, Trump met it and beat it in the key states of North Carolina, Florida, Texas, Ohio and Iowa to keep them in his column. The large suburban gains that Democrats expected were either much more muted than expected, or almost completely negated by losses with minority voters in the cities those suburbs surrounded.

However, as the calendar has changed, many people believe that the Democrats have a renewed chance at taking the Senate, with the two Runoffs in Georgia of Ossoff vs Perdue, and Warnock vs Loeffler.

As we have seen in past runoffs in Georgia, and in low(er) non-main cycle elections, the electoral turnout becomes much more educated than in the general election as mostly only higher information voters know about the election happening, and find their way to a voting booth. Even though Biden’s gains with white college voters was lower than many expected, those suburban gains still happened, specifically in Georgia. In fact, in the sun belt states, where the polling largely missed Trump’s levels of support, one of the states where the polling was actually pretty accurate was Georgia. There can be many possible reasons for that, but the most likely is probably the notable absence of a large Hispanic population, which was present in Florida, Texas, and Arizona/Nevada (even though Biden did win Arizona/Nevada, he did by a smaller margin than predicted).

Georgia was one of the lone exceptions on election night, where Biden didn’t massively underperform his polling numbers, and the expected gains with white college voters did materialize. Additionally, in an environment where those same white college voters are expected to make up a larger share of the electorate than in November, where Black turnout from the early voting numbers appears to be higher than the general election, and where turnout in the rural areas appears to be crashing through the floor, it creates an irresistible storm where most analysts would take the Democrats. Of which I am one, to be perfectly clear. If someone were to hold a gun to my head, and demand I make an opinion on the races, I would say that Ossoff and Warnock both win, by small margins. However, all of this just feels too similar to November. An accurate model is only as good as the polling data, and we just aren’t getting good data.

Here at LeanTossup, we don’t put much stock in polling misses repeating. In the UK election, many people were waiting for a Labour resurgence in the final weeks of the campaign, waiting for an inevitable Conservative Minority government. Our model clearly showed the Tories in a solid majority. Even in 2020, we (incorrectly) didn’t believe that there would be a polling error in the same direction as 2016 (in favour of Trump). We know that polling misses are random, and that pollsters overcorrecting for one polling error can very easily create a new error in the other direction, so I’m not one to believe that the same polling error will occur again and again. Additionally, in many of those cases, those misses were sometimes years apart. Non-zero numbers of new voters join the electorate, and voters actually have months (years) to change their mind as new information about the candidates/parties becomes available. Many voters chose Jeremy Corbyn in 2017 because they disliked Theresa May and her approach to Brexit. Fast forward to 2019, and many of those same voters feared that Brexit would not happen if they didn’t support Boris Johnson and the Conservatives, which is a clear example of many voters decidedly changing their minds due to current events. We have to remember that a large number of voters are not partisan. They vote based on emotion.

However, in the US, that last polling error was literally 2 months ago. While yes, over time, pollsters might attempt to correct their 2020 errors, I highly doubt that they have had an opportunity to fix them in such a limited time since November. Since the Presidential election, very few new voters have entered the electorate, and nothing of substance has changed since the election, with the possible exception of Trump challenging the results, of which we don’t know what effects that might produce (if any). Additionally, in terms of that Presidential election, the Democratic Senate candidates did worse than Biden in the suburbs. Those same suburban white college voters that now make up a larger share of the electorate could be a double edge sword. Are they more likely to be Biden straight-ticket voters? Or are those split-ticket voters that voted against Trump but for GOP control of the Senate going to be more motivated to continue to vote for that same preference they expressed in November? Has Trump contesting the results of the election flipped any of those voters towards the Democrats? We don’t know. With almost no polling in the race we have no idea about those pivotal questions.

While everything I know about politics and electoral composition points to Democratic wins in the Georgia runoffs, it just feels like we are all falling into the same traps as November: No actual error in analysis, as all of the data leads us to the conclusion of a pair of Democratic wins that many analysts are making, but somehow, against all odds, the GOP still finding a way to hold onto the seats. That all of the GOP voters decide to vote on election day after Trump has spent two months bashing early and mail-in ballots. That those suburban numbers giving hope to the Democrats are a mirage, as they are GOP leaning voters that wanted a President Biden, but are coming out now to stop a Senate Majority Leader Chuck Schumer. That the minority voters that have already voted, somehow, against all odds and laws of math, decided they want a divided government instead of a unified one.