What to Expect from 2022 Early Voting
Early Voting is upon us with North Carolina leading the way by sending domestic and overseas voters their mail ballots on Friday, September 9. The National Conference of State Legislatures has full calendars for mail balloting and in-person early voting periods. Of course, always check with your local election officials about the rules and procedures as they apply to you if you wish to vote prior to Election Day on November 8.
As in prior elections, I will track early voting statistics for the country. Much has changed since the last midterm election in 2018. Understanding these changes is important when interpreting early voting statistics to forecast who will win, as in some cases these changes greatly diminish the value of early voting statistics for many states. These statistics can still be informative, if you know how to consume them.
So what can early voting tell us?
Turnout
States with high volumes of early voting tell us something about overall levels of voter participation. If a state nears or exceeds their 2018 turnout in their 2022 early voting alone, that sends a strong signal of high voter engagement, as 2018 had the highest turnout rate for a midterm election since 1914.
I recommend caution when comparing the 2022 level of early voting on the same number of days prior to Election Day in 2018 as a signal of overall turnout. As I document in my new book, From Pandemic to Insurrection: Voting in the 2020 US Presidential Election, voters changed their behavior in 2020 in a number of ways, including by voting prior to the election at much higher rates and returning their mail ballots earlier. If that behavior continues in 2020 we may need to wait until closer to Election Day to get a clear picture of overall turnout.
Why is predicting turnout important to predicting the election outcome? When pollsters release surveys of “likely voters” they are modeling what they think the overall turnout will be. Some of the more transparent pollsters release estimates based on varying levels of turnout — low, medium, and high. Early voting participation can provide clues as to which of these turnout scenarios is more correct.
Horserace
Predicting who will win from early voting can be challenging, and will be more so in 2022.
We do not know voters cast their ballots until election officials report election results. What we know in states with party registration is the relative frequencies of registered Democrats, Republicans, and no party affiliates who have voted early. In some states without party registration there may be other clues as to who is voting, such a race (in the few states that track such information on their voter registration rolls) and region, such as urban/rural divides.
A difficulty with the partisan breakdown of early voting is that states vary in how they encourage early voting.
Some states’ early voting provide strong horserace signals — particularly the vote-by-mail states that send every active registered voter a mail ballot. By Election Day, in California, Colorado, Hawaii, Nevada, Oregon, Utah, Vermont, and Washington we will likely have a fairly good picture by Election Day of the relative frequency of registered Democrats and Republicans who will vote in these states. We can compare to 2018 to see the relative frequencies in that election. Putting these two statistics side-by-side (in what is known as a “difference-in-difference” approach), we can gauge relative enthusiasm of registered Democrats and Republicans compared to the prior election. This method is not full-proof to predict election outcomes, but has an excellent track record to predict winners for high volume early voting states. For states that with medium to low volumes of early voting, it is important to wait for Election Day, when larger proportions of Republicans tend to vote.
A key for the difference-in-difference approach to work well is that there must be a comparable election to use as a yardstick to measure how the electorate changed across elections. The last midterm election of 2018 is an obvious best comparison. However, the 2020 presidential election and the pandemic has thrown a gigantic monkey wrench into early voting behavior. Overall levels of early voting increased in 2020 to record proportions — about 70 percent of votes were cast prior to Election Day in 2020. There is evidence from the 2022 primary elections that interest in early voting has carried forward. Furthermore, Republicans who used to dominate mail balloting are now tending to vote more frequently in-person while Democrats who used to flock to in-person early polling locations are more often casting mail ballots.
Voters’ changing behaviors will likely upend past patterns of early voting, degrading the value of the 2018 election as a comparison benchmark. Since mail balloting tends to occur before in-person early voting, I will not be surprised or give much weight to registered Democrats establishing early leads in returned mail ballots. Even in-person early voting patterns appear to have changed. In the past as Election Day neared in-person early voters as a whole became younger and more likely to register as Democrats or have no party affiliation. That pattern was disrupted in 2020, with more Republicans voting in-person early as Election Day approached.
Despite these limitations, I believe there is value to early voting to divine the horserace. I’ll be paying particular attention to the large volume vote-by-mail states for overall turnout forecasts and relative voting interest among partisans within these states. These states in turn can provide confirmation of what is seen elsewhere. Still, midterm elections are challenging because — unlike the presidential election — there is no national election to drive interest across the country. We will likely observe state and district variation due to campaign and candidate dynamics.