The Washington PostDemocracy Dies in Darkness

Opinion We could have record turnout in the 2020 election. We’re not ready for it.

"I Voted" stickers at an Alexandria polling place on Nov. 7, 2017. (Alex Brandon/AP)

Election officials are already bracing for a record voter turnout in next year’s presidential election. A closer look at historical data, however, shows they are still likely underestimating potential turnout by as many as 42 million people.

The passions President Trump inspired have caused a skyrocketing interest in voting. According to University of Florida associate professor Michael McDonald’s United States Election Project, the most comprehensive data source on historical voter turnout patterns, more than 118 million people voted in last year’s midterm elections. This was the largest number ever and the highest percentage of the voter-eligible population since 1914. This has led McDonald and others to hypothesize that between 156 million and 161 million people will cast ballots next year.

That alone would be a significant increase from the 139 million that voted in 2016. Already thinly stretched election systems would strain to efficiently handle such a surge. But the data suggest that even this estimate is low.

The American Enterprise Institute recently reported that a CNN poll shows a record 71 percent of Americans say they are extremely or very “enthusiastic about voting for president” next year. The report also cited data from polls for the past four years preceding the presidential election, which showed similar levels of interest ranging between 51 and 55 percent.

Follow Henry Olsen's opinionsFollow

There is a strong and consistent relationship between these numbers and the actual share of voter-eligible persons who cast ballots the following year. Data from McDonald’s website show that share is always between 5.1 and 8.6 percent higher than the numbers reported by AEI. For example, almost 61 percent of eligible Americans cast ballots in 2004 while the 2003 CNN survey data showed 53 percent of Americans expressed serious interest in voting, nearly an 8 percent increase.

McDonald estimates there were 235.7 million eligible voters in 2018, which has increased by an average of about 4.4 million every two years. If the historical relationship between pre-election year voter interest and actual voter turnout continues, then between 182.5 million and 192.4 million people will vote next year.

That’s a shocking number, but other historical voter turnout data confirm it. Data from McDonald’s website also show there is a consistent relationship between the number of ballots cast in the midterm preceding a presidential election and the number of votes cast the following election. McDonald’s data for the past four presidential election cycles going back to 2004 and the 2002 midterm show that between 43 and 67 percent more ballots are cast in a presidential election than in the immediately preceding midterm.

There were an estimated 118.6 million ballots cast last year. If past trends hold, that means between 170 million and 198 million votes will be cast next year. Taking the average increase from the past four cycles, 55 percent, again yields an estimate of 183.5 million cast votes.

Our state election systems are almost certainly not prepared for this. We already face complaints that there are too few polling stations, especially in inner-city areas, to accommodate the people who wanted to vote in past years. Imagine if those two-hour waits double to four-hour waits. Affected populations would surely cry foul, leading to even more charges of intentional voter suppression and election manipulation.

This is something that should concern everyone, left and right alike. Partisan hatred is already increasing at an alarming rate. Imagine what would happen if after an incredibly bitter campaign, millions of people faced insuperable burdens that lead to them either not voting or extending polling hours into the wee hours of the night to accommodate voter demand. Both parties would likely end up crying fraud, with the loser possibly even claiming the election was stolen. We cannot allow that to happen.

We have enough time to plan so it doesn’t happen. State and local election authorities should dramatically increase the number of polling machines and available staff so that waiting times do not significantly increase over 2016. Both parties should urge as many of their supporters as possible to vote early so their ballots are cast and counted on time. Digital technology should be used to accurately record the names and registrations of people subjected to long waits so they can leave the lines and get a call back or a text telling them when they can safely come back. People who say they cannot wait should be given a fast track to vote so election incompetence does not cause them to become discouraged and leave.

The voter tsunami is coming. It is up to us to prepare for it. Failure to reach bipartisan agreement to prevent this catastrophe would do more to discredit U.S. democracy than anything anyone wants to imagine.

Read more:

Henry Olsen: Republican senators are right to resist impeachment

The Post’s View: The U.S. government needs to protect against 2020 election meddling

Michael McFaul: A high-level Senate report confirms it: Our elections still aren’t safe

Greg Sargent: Want to beat Trump in 2020? Register people to vote.

E.J. Dionne Jr.: North Carolina’s election results provide a thought experiment for 2020