Voter turnout is likely to decide the winner. In prior elections, a candidate for President could win by capturing the lion’s share of “swing voters,” “undecided voters,” or “independent voters.” This year, that’s no longer the case.
There’s just not very many of them. Research by political scientist Corwin D. Smidt indicates that in Presidential elections between 1952 and 1980, an average of 12% of voters switched parties from one election to the next. Since 1984, that average has fallen to 6%.
If a candidate has little hope of converting his opponent’s partisans, his best path to victory involves motivating his own supporters to turn out to vote.
A variety of indicators suggest this year’s turnout will set records. One thing that both sides agree upon is that this election is hugely important, as demonstrated by the following chart.
These attitudes affected the 2018 Midterm Elections, which had the highest turnout in recent history, even though President Trump wasn’t on the ballot.
The conditions that drove the dramatic increase in 2018 appear to be even more potent now. To better understand how they might affect voter turnout in 2020, it’s helpful to analyze which demographic groups had the largest changes in voter turnout in 2018.
Age. The change in the age makeup of voters was incredible. People 18–29 were 75% more likely to vote in 2018 than they had been in 2014. By contrast, those 65+ were only 11% more likely to vote. This reduced the skew towards older voters (though old voters were still almost 2x more likely to vote than young voters).
Gender. Not much difference between genders. Men and women were each about 27% more likely to vote.
Race. Although not as dramatic as Age, the changes in the Race makeup of the electorate was significant. Hispanics and Asians were 50% more likely to vote. Whites and Blacks, who both traditionally vote at much higher rates than Hispanics and Asians, increased their voting rates by about 25%.
Educational Attainment. The changes in this demographic ran counter to the changes in the others, in that, for both Age and Race, the groups with the lowest propensity to vote increased the most. For Educational Attainment, each group increased roughly in proportion, ranging from 19% for those with Advanced Degrees to 31% for those with Some College or Associate’s Degree.
Metropolitan Area. There was a somewhat significant difference in the changes based upon Metropolitan Area. Voters in cities increased their participation rate by 34%, while those in non-metropolitan areas only grew by 17%.
Will these rates of growth be duplicated in the 2020 Election?
It’s hard to be certain that they will be. Midterm elections generate much lower turnout than Presidential elections. Even with those large increases, the 2018 Midterms didn’t match the 2016 vote count. It could be that people who generally vote in Presidential elections but not in Midterms chose to vote in 2018, but that people who never vote at all continued to sit out the election. If that’s the case, the voting rates in 2020 may not be any different from 2016.
Still, there are other indicators that suggest it will be quite different. One of those indicators is opinion polls. Pollsters ask voters questions to tease out how likely they are to vote. This helps eliminate from the data those voters who are not likely to go to the polls, and thus makes the forecasts more accurate. For example, the Public Religion Research Institute, which conducts a very thorough poll of approximately 2,500 people, asks respondents whether they are registered, how they plan to vote, and how likely they think they are to vote. In Appendix 1: Survey Methodology, they say… (This is not the place one normally finds dramatic news, but read on!)
Due to the overwhelming majority of respondents saying they are both registered and absolutely certain to vote or had already voted (76%), we determined that a “cutoff” model based on those factors alone resulted in an unrealistic estimate of voter turnout. Using the assumption that if people are unsure of how they will actually cast their ballot, they are a bit less likely to vote, we used the plan to vote question to decrease the likelihood of voting among those who report not knowing how they would vote. Then a probability model was applied in order to avoid the sharp exclusions of a cutoff model. This process resulted in a model in which 68% of adults are deemed likely to vote and constitutes the high turnout estimate reported. Because 68% would be historically high turnout, we estimated a second likely voter model that forced the data to replicate the roughly 55% voting age population turnout of 2016.
In other words, if you take people at their word, we would have 76% voter turnout. The statisticians deemed that unrealistic, so they artificially brought the number down to 68%. But even a turnout rate of 68% would be a 24% increase over 2016’s turnout rate— roughly in line with the growth in the 2018 Midterms.
There is no historical precedent for voter turnout close to 68%. Since the 26th Amendment expanded voting rights to 18 year-olds in time for the 1972 election, the highest rate of turnout was in 2008, and it was only 57%.
But if early voting data is any indication, we are on a path to getting there. With two weeks to go before the election, 37 million votes have already been cast. That’s more than 25% of the total ballots cast in 2016. Millions more are being added every day. Believe it or not, in Texas the early vote total is already more than 50% of the total vote count in 2016.
Of course, there’s no way to know for certain whether these early vote totals merely indicate a shift towards early voting due to the pandemic or if they point to higher than normal total turnout.
A nice, round number to keep your eye on is 150 million votes. A turnout rate that matched 2008 (57%) will generate about 150 million votes. If we actually do hit a 68% turnout rate, that would imply roughly 180 million votes — 40 million more than four years ago.
If there is a massive increase in turnout, it bodes well for Democratic candidates. The disproportionate increases in young voters, Hispanic and Asian voters, and urban voters all play to Democratic strengths. Because most polls tend to assume the electorate will look similar to the previous election, unusually high turnout will probably lead to Democrats outperforming expectations.