Stop Trying to Derive Meaning From Early Voting

The system is so varied and each state so specific that it is nearly impossible to assess.

Mail-in ballot.
A mail-in ballot for the midterm elections in California’s Orange County.
Robyn Beck/AFP/Getty Images

Election ballots are already in the hands of millions of voters, who either requested an absentee ballot or received one by virtue of living in one of three states that send them automatically (Colorado, Oregon, and Washington). Millions more have cast a ballot at an early voting location. As of this writing, 22 million votes have already been cast, but there is no reliable way to know for whom.

Political journalists are desperate to fill space right now, and early votes are at least actual votes. As a political scientist who studies early voting, I get the same questions every election: Which party benefits from early voting? What can we tell from early voting so far? And—the most off-base—who will win the early vote? (Pro tip: No one “wins” the early vote—candidates win the final vote.) As Yogi Berra famously said, “it ain’t over until it’s over,” and I’d urge you to look at early voting returns the same way.

As attractive as it may be to try to use early voting results as an indicator of the final results, there are vast areas of uncertainty that make them, at best, an indicator that needs detailed analysis and context, the kind that some campaigns have but to which most observers aren’t privy. The reasons why it’s so hard to get good analysis is because the processes for early voting are very complicated and dissimilar. There are three methods or “modes” of early voting:

1) Early in-person voting: A voter appears in person to cast a ballot, usually using the same voting technology as on Election Day. Thirty-four states and D.C. provide for this.

2) Absentee and no-excuse absentee voting: A voter submits a request to receive an absentee ballot through the mail and returns the ballot via the mail or drops it off at a designated location. Twenty states require an excuse to request an absentee ballot, while 27 states and D.C. don’t require an excuse.

3) Vote at home/Vote by mail: All registered voters receive a ballot in the mail, and ballots are returned by mail or dropped off at a designated location. This is what happens in Colorado, Oregon, and Washington.

These three modes give the general gist, but even still, this masks enormous variability across the country in the methods and times for early voting, consequently influencing which and how many voters cast early ballots. There is a ton of random variability otherwise—for example, Massachusetts allows early voting only in even-year November elections. Colorado uniquely allows you to surrender your vote-at-home ballot and cast a vote in-person at “voter service and polling centers.” Eight states and D.C. allow you to put yourself on a permanent no-excuse absentee list, while some states make you request this status each election or each year. Eighteen states hold some elections fully by mail (usually some municipal or special elections).

If this all seems really complicated, it is! There has been some federal legislation that has standardized some election procedures, but like most voting rights issues, it’s often motivated by partisan competition. Even the original introduction of absentee voting fits this mold: The practice started during the Civil War when Republicans suspected that soldiers would cast a ballot for Lincoln over McClellan and introduced mail-in ballots.

All of the complications around how early voting works should help to explain why it’s so hard to answer the main questions people always want answered, about whether it increases turnout and whether we can use early voting numbers to predict anything meaningful.

The scholarly experts are split on whether or not early voting laws increase or decrease turnout, in part because there’s so much variation. Early voting is just one of many ways that states can make voting easier—or harder. Registration requirements vary substantially across states, and registration has a consistently strong impact on turnout. But most importantly, scholars have known for decades that the primary barrier to higher turnout in the United States is voter disengagement and disaffection, so whatever impact early voting has, it’s probably not that significant.

To the second question, about whether we can infer anything useful from early voting returns, the answer is again familiar: “Early voting” is really complicated, and it’s really complicated to infer anything from early voting returns. I’m not alone in taking this viewpoint. Nate Silver describes why the FiveThirtyEight model doesn’t consider early votes, and Nathaniel Rakich from FiveThirtyEight provides a detailed list of the ways that early vote returns can be misleading. Harry Enten and Eric Bradner, elections analysts for CNN, give us more reasons to be cautious about over-interpreting the early vote.

These caveats coalesce around the idea that something can be learned from early voting returns, but most commentators don’t have all the information needed to put the returns fully in context, e.g.,  how voters have behaved in the past, how likely they are to vote in 2018, and what campaigns are doing to stimulate early voting.

The highly competitive governor’s race in Georgia is a great example where we may be reading too much into the early voting returns. Both Brian Kemp and Stacey Abrams have put a heavy focus on early voting, but Abrams is a very distinctive candidate. She is trying to make history as the first black female gubernatorial candidate in United States history. It’s no surprise that early voting is way up, especially among African Americans and Democrats. This may be an indicator of higher voter enthusiasm in that state, or it may be cannibalizing voters who would have waited until Election Day. There’s compelling evidence for both arguments.

Early voting is extremely important because it’s becoming a major part of the way people vote in the United States. And yet there are still vast areas where we have only a partial sense of the best way to administer early voting, how it changes campaigns, or how it impacts voters. There’s a lot of work ahead of us.

There are some scholars making a good run at this kind of cross-election, comparative analysis of individual voting histories. Michael Bitzer at Old North State Politics and Daniel Smith aka ElectionSmith are doing excellent work on North Carolina and Florida respectively, but this is only because these two states make individual voter registration and voter history files readily accessible. Most states do not.

It’s hard to sort the geniuses from the fools in this hyperpoliticized environment. My advice is pretty simple: Don’t obsess over the early vote.