This Chart Shows How Much Turnout Could Matter in November
By Anthony Rentsch and Brian Schaffner
One of the greatest challenges pollsters face is determining who is likely to vote. Americans notoriously overstate their likelihood of voting when they talk to pollsters. For example, in 2016, nearly four-in-five respondents to the Cooperative Congressional Election Study survey said that they would definitely vote in the presidential election. However, when we matched the survey to voter files, we discovered that one-third of these “definite” voters did not actually have a record of voting.
Pollsters try to deal with this uncertainty in a variety of ways. Along with our co-author Justin Gross, we developed a technique that uses patterns from previous elections to predict how likely each survey respondent is to actually vote in November. This is based partly on how likely respondents tell us they are to vote, but of course we can’t fully trust those responses, so we also include a variety of other demographic and political predictors to help us fine tune a better estimate for each respondent’s actual probability of voting.
We recently applied our approach to a YouGov Blue survey of 1,059 registered voters. Using our likely voter methodology, we are able to generate a predicted probability that each respondent will vote in November, based on how likely they said they were to turn out as well as the other predictors in our model. Weighting respondents’ stated candidate preferences by these turnout probabilities, Joe Biden currently leads Donald Trump by a margin of 51% to 45%.
Of course, there is as much uncertainty as ever about who will vote this year, and so it is even more useful to think about how different scenarios may affect the ultimate outcome. Indeed, it is not always clear which party benefits from increased turnout. This is what the chart below does. Specifically, we use our turnout probabilities to plot a range of scenarios where anywhere between 60% and 92% of registered voters turn out in November. We also use data from the Current Population Survey’s Voter Supplement to benchmark various turnout rates over the past 30 years. For example, according to the CPS, over the past 7 presidential elections, the average turnout among registered voters was 87%. The lowest figure was in 1992, when just 82% of registered voters cast a ballot while the highest rate was 90% in 2008. And, of course, turnout is much lower in midterm elections, where an average of 71% of registered voters typically turn out. Given precautions over COVID-19, some fear that turnout could drop well below what is typical for a presidential election year.
Notably, the chart shows that as turnout increases, so too does Biden’s margin over Trump. If turnout was especially low, say around 70% of registered voters, then Biden would be in a tossup race with Trump. This would be abnormally low turnout for a presidential election, but consistent with an average midterm election.
When turnout is 80% of registered voters, still lower than it has been in any election since the CPS has kept track of these statistics, Biden’s lead grows to about 5 percentage points. But this 5 point lead comes with a fairly large margin of error which would make an electoral college victory far from certain. Once turnout reaches the 1992 benchmark of 82%, Biden’s lead grows to a somewhat safer 7 points. And if turnout in November hits the heights achieved in 2008, then Biden’s margin would be a fairly hefty 9 points over Trump.
Overall, the chart helps to demonstrate how important turnout is to a Biden victory in November. As turnout increases, so too does Biden’s margin. And, of course, this also helps to show why Trump seems so determined to make it more difficult for Americans to vote in November. As the wave of Americans casting their ballots in November grows larger, it is increasingly likely that it will carry Trump right out of the White House.
Anthony Rentsch (@Anthony_Rentsch) is a data analyst at Blue Labs.
Brian Schaffner (@b_schaffner) is the Newhouse Professor of Civic Studies in the Department of Political Science and Tisch College at Tufts University.