By Tom Bonier

I’m excited to announce the public release of the 2020 iteration of our TargetEarly site. TargetEarly will provide a real-time look at the American electorate as voters cast what will certainly be a record number of votes before Election Day. We believe this site will provide important context to an election rife with uncertainty and volatility.

While TargetEarly will provide an unprecedented level of access to voter turnout data, it is important to understand what we can and cannot glean from the early vote data.

In 2016, many of us fell into a trap of overreacting to a clear Democratic bent in the early vote data, without considering the context of prior turnout among early voters. In the end, the Democratic early vote advantage was built largely upon high propensity voters who would have voted on Election Day had they not voted early, thereby not representing any significant surge in intensity relative to Republicans.

For the 2018 election, we applied those lessons to our TargetEarly analysis, allowing users to assess early turnout data in the context of previous voting behavior. It was this level of analysis which showed that Democrats were building an advantage among “surge” voters – those who might not otherwise be expected to vote in a midterm election. These advantages extended to younger voters and people of color, portending surges in turnout among those groups, something that held true through Election Day.

The 2020 TargetEarly site allows users to view early vote data in the context of various groupings, including age, gender, race/ethnicity, partisanship, past vote history, and urbanicity. You may also compare current early vote turnout data to the same time period in the 2016 and 2018 election cycles. These comparisons, while interesting, must be considered in the context of an electoral system which has changed dramatically in response to the pandemic. It would not be shocking if the number of votes cast before Election Day more than doubled, relative to 2016.

So what should we be watching for as we seek a better understanding of which party is performing better in the early vote? I’ll be looking closely at the partisan gaps among voters who have never voted before or have ​infrequent history of voting. Due to the massive partisan differentials in likelihood of voting early, I’m not as interested in the share of voters from each party who are voting early, but rather the partisan gaps among these lower propensity voters.

Building an advantage among low turnout propensity voters may not only suggest an intensity gap between the parties, it also represents a tactical advantage. The earlier a turnout target voter casts their ballot, the more money the campaign can reallocate to targets who have not yet voted.

What can’t be known from the data? It’s important to keep in mind the limitations. First and foremost, it’s important to stress we only know if someone voted, not for whom. Any partisan data we apply is based on either partisan voter registration data, or modeled partisanship. These models only predict the probability that an individual would self-identify as a Democrat or Republican, not their specific vote choice. Especially in this unique environment, it is reasonable to assume that the candidates will perform divergently to base partisan identification among certain groups of voters (such as Vice-President Biden’s relative strength emerging among senior citizens).

Perhaps more importantly, while we will naturally be curious regarding the comparisons between current early vote data and the same point of time from previous cycles, these comparisons must be viewed in context. And they must be taken with a very large grain of salt.

For example, as I write this, our TargetEarly dashboard shows voters under the age of 30 accounting for a diminished share of the early vote relative to this same point in the 2016 and 2018 cycles. Does this mean that younger voters are demonstrating a lower turnout intensity this cycle? Not likely. In raw numbers, they are turning out at a rate of almost 7 times their turnout in 2016. But in the context of this pandemic, it is reasonable to assume that older voters are pivoting more to the safer early methods of voting than younger voters, thus resulting in a diminished youth early vote electorate share.

It’s also important to keep in mind that these numbers will evolve. For example, in previous election cycles, older, white voters have tended to vote earlier in many states, though it is impossible to say if that will be the trend this year. In the end, while the information on TargetEarly will be incredibly helpful in assessing the state of the electorate, in the end we will have to wait until all of the votes are counted to draw conclusions!