The Case for Cautious Early Vote Analysis

Why the Early Vote Analysis You're Reading is Probably Wrong

I’m a big believer in the value of early vote analysis. With tens of millions of votes being cast before Election Day, we have a very rich dataset from which to draw conclusions. Yet, early vote analysis is complicated. When we are seeking to gauge relative intensity of partisans, it’s not enough to look at which party has turned out more of its supporters in the early vote. A good analysis benchmarks current early vote data against previous cycles. Where one party or voting group is outperforming the historic baselines, it tends to be a sign of higher intensity and engagement for that group overall. And that is where the challenge lies in this unique cycle. 

As I noted ahead of the rollout of TargetEarly over 3 weeks ago, the fact is that when it comes to understanding the early vote, looking back at past cycles can only take us so far. A lot has changed. In the last few years alone, voting habits, state laws, party messaging, and even public opinion around early voting have shifted drastically.

Let’s take a moment to dig into those points. What has actually changed since 2020?

Well, for one, we’re not in the peak of a deadly pandemic. The 2020 election saw the biggest liberalization of access to early voting as states adapted to the realities of the pandemic. And it was a great success, with over 100 million Americans safely casting their vote before Election Day. Of course, there was an asymmetry here. Democrats were more covid-conscious, and therefore more likely to cast an early vote (take Pennsylvania, where registered Democrats outnumbered Republicans in the early vote by an almost 3 to 1 margin). And at the same time, Republicans largely abandoned voting by mail, due to Donald Trump claiming that mail voting was fraudulent.

Republicans acknowledge that their failure to drive their voters out early in 2020 put them at a strategic disadvantage, and have since committed significant resources to turning that around in this election. Just yesterday the New York Times reported “Republicans have spent months and millions of dollars on an effort to push former President Donald J. Trump’s most loyal supporters to change their minds about voting early.”

It is also important to keep in mind that many states have changed access to early voting. Michigan added early in person voting, while North Carolina put stricter voter ID laws in place. Georgia put additional regulations into place on mail voting. 

All of these changes will make any comparisons in the current early vote to 2020 numbers extremely difficult, at best.

With Democrats shifting from early voting to Election Day, and Republicans doing the opposite, it is safe to say the expectation is that the early voting will skew much more Republican than it did in 2020. So at what point can we draw conclusions as to which side has an intensity advantage? We may not be able to, though when the results in a state defy expectations (as they currently appear to in Michigan and Wisconsin, where partisan models suggest the early vote is actually more Democratic than it was at this point in 2020), that is noteworthy. 

Let’s look at the early vote data in one state, just to better understand how complicated this analysis can be. We’ll use North Carolina as an example.

North Carolina

Taken at a high level, the early vote data in NC would appear to be very good for the GOP. At this point in 2020, the early vote was +16D, and right now it is +1.5D. But, as I’ve said, it's a lot more complicated than that. 

It does appear that the GOP is performing well in early voting so far in North Carolina, right? Well, we can’t be so sure. In fact, as evidenced in the graphic below, turnout is down relative to this point in 2020 among all parties, but the GOP is much closer to their 2020 turnout than Democrats, which is entirely consistent with what one would expect.

Here’s some important context: according to our data, in 2020 in North Carolina, a whopping 82% of votes were cast before Election Day. By 2022 that number dropped to 58%. It is quite possible that the early vote share will drop again in this election. So you can see how comparing current early voting data to a cycle where more than 4 out of 5 voters cast an early vote is problematic.

In 2020, the Election Day vote in North Carolina was +19 Republican. By 2022 that dropped to +15 GOP. We expect that trend to continue, and to see Democrats perform much better on Election Day, due to all of the factors I’ve detailed above. 

This same analysis could be applied to almost any other battleground state. In Pennsylvania a striking 42% of registered GOP early voters this cycle voted on Election Day in 2020, as compared to only 12% of registered Dem early voters. Nevada is another good example, as much has been made about the gap closing from +16D to +2D at this point compared to 2020. Yet, at this point in 2022 the early vote in Nevada was just +3D, and the result was a close and competitive election that saw Senator Cortez-Masto win statewide. And let’s remember, the final Democratic advantage among the early vote in Nevada in 2020 was +3D and in 2022 it was +1.5D. So, Democrats started off with a massive advantage in 2020 – but things settled back down and once again, all things point to a close race. It’s also worth noting that the addition of automatic voter registration in Nevada, with a default to register new voters as unaffiliated, has resulted in a larger pool of unaffiliated voters who lean Democratic, relative to 2020. 

While banking votes early is obviously important, it’s also important who is voting. The early evidence is that Republicans are cannibalizing likely Election Day voters, not turning out lower propensity voters early, which is always the priority. As a matter of fact, across the battleground states, 92% of voters who have cast their ballot so far also voted in the 2020 election. 

So, what does this mean? 

In short, it means that something that was already difficult before the massive vote-mode shifts we're seeing happening this year will be exponentially more difficult. And in turn, we are seeing exponentially more flawed analyses, so buyer beware. As I noted earlier, my general approach this cycle is to stay in this context of expected vote-mode shifting, so when the early vote looks close to 2020 (or better) for Democrats, that is a very good sign, and when it looks worse for Democrats, like in North Carolina, we are left with the question of how much worse is problematic. 

As we head down this home stretch, we'll continue to share our ongoing analysis, and you can always follow along at @tbonier and @TargetSmart.