Midterm Election Turnout Was Up. How Much? We Don’t Yet Know.

Voters turned out Tuesday at rates not seen in a midterm election in half a century, driven by strong opinions of President Donald Trump and a string of competitive races in states where robust contests have been rare.

It could be months before we know the demographic breakdown of who voted in this election, but there are early indications that turnout boomed especially among women, Latinos and young people.

We crunched the numbers and spoke with experts. Here’s what the data can tell us (and what it can’t yet).

Based on preliminary — but incomplete — data made available by the states and analyzed by Michael McDonald, a professor of political science at the University of Florida, it looks as if more than 113 million people voted, which would be at least 48 percent of eligible Americans. That’s up from the 83 million votes cast in 2014, when Republicans made sweeping gains in the House and the Senate. In fact, it’s closer to the turnout in the last presidential election — about 138.8 million — than to the last midterms.

By percent of people eligible to vote, it was the highest turnout of any midterm election since at least 1970 and the first time midterm turnout topped 100 million, said Tom Bonier, chief executive of TargetSmart, a data analytics firm that studies voter data. He based his analysis on the county- and precinct-level data reported so far.

In some counties, something almost unheard-of happened: More people voted in the midterms than in the last presidential election. One example is fast-growing Travis County, Texas, which contains the left-leaning city of Austin. Preliminary numbers show that 775,950 people voted there Tuesday, compared with 725,035 in 2016.

“That’s totally crazy,” Bonier said. “You don’t generally see midterm turnout that approaches presidential year turnout, and that is just something that is true in this country. I have never had an example to point to before.”

The increase was most dramatic in states with competitive races, but it also happened in states without them. In at least 41 states, turnout was up by double-digit percentages from 2014.

Texas, which had the nation’s lowest percentage turnout in 2014, saw the biggest increase this year: Sixty-three percent more people voted Tuesday than voted in the last midterm elections. Four other states, all with big Senate races, also increased by 50 percent or more: Nevada (up 60 percent over 2014), Missouri (58 percent), Indiana (51 percent) and Tennessee (50 percent). Georgia, where the governor’s race could be headed to a runoff, was up 42 percent; Arizona, where the Senate race is still too close to call, was up 40 percent. Turnout in Florida — which had competitive races for both the Senate and the governorship — increased less than in other places, though still significantly: 24 percent. That may be partly because its turnout in 2014 already put it in the top half of states.

Only in three states: Maine (down 2 percent), Louisiana (4 percent) and Kentucky (7 percent). But these numbers might still improve as states count absentee and provisional ballots. (Maine’s turnout was the highest in the country in 2014, when voters had a competitive governor’s race and also re-elected Sen. Susan Collins. So despite the decrease this year, the state’s turnout, at 57.8 percent, was still the seventh-highest Tuesday night.)

For one thing, there were exciting and competitive races in some states that rarely have them. In Texas, which had the biggest turnout increase in the country, Beto O’Rourke launched a headline-grabbing challenge to Sen. Ted Cruz and lost by less than 3 percentage points. (For comparison, Cruz was elected by 16 percentage points in 2012.)

“People will vote when they believe their vote matters,” said McDonald. “What induces people to believe their vote will matter? Well, having a competitive election where they may actually be able to cast a ballot and affect an outcome.”

But turnout also increased in states without electrifying races. Everyone expected Delaware would be blue, but its turnout went up 46 percent, the seventh-largest increase in the country. Everyone expected Alabama would be red, but its turnout increased 42 percent over 2014.

The reason for that can be boiled down to one person: Trump.

“Midterms are always a referendum on the president,” said Mia Costa, an assistant professor of government at Dartmouth. But presidents don’t usually say that out loud, as Trump did regularly at rallies.

“With Trump at the center of the election, I think voters were just fired up on both sides,” Costa said.

It could be months before we have a detailed demographic breakdown of the electorate. Hundreds of thousands of ballots haven’t been counted yet, and right now we have only aggregate data — that is, the total number of votes cast in each state, county and precinct. We don’t know, for example, how many voters were women or African-Americans or in their 20s.

Be wary of reports that already cite exact percentages for specific groups. Those numbers are mostly based on exit polls, which can give us a broad sense of what the electorate looked like but are imprecise and sometimes inaccurate.

We can extrapolate some trends, though, from voter data at the county and precinct level.

For instance, precincts in Virginia, Texas, Florida and California that are composed mostly or entirely of college campuses reported a significant increase in turnout, suggesting a broader surge of young voters, Bonier said. Similar trends are visible in counties with large nonwhite populations. “It is highly likely that this election was younger and more diverse than any midterm election this country has seen in some time, based on early voting data and matching it up with county-level data,” Bonier said. “It is imperfect, but the data does point in that direction.”

Early voting numbers also point to an increase in young, female and Latino voters, Costa said, but not to a significant increase in African-American voters. That could change as more detailed information comes in.

Looking at socioeconomic divisions, data suggest that voters in areas where most people have a college degree, or where unemployment rates are low or falling, turned out at higher rates than voters in areas with lower education levels or more unemployment. The average turnout was 62 percent in counties where a majority have college degrees, compared with 43 percent in counties where less than 10 percent are degree holders, said Christina Coloroso, director of analytics at Catalist, a data firm.

In counties where unemployment was lower than the national average, as many as 57 percent of voters turned out, she said. Where unemployment was above average, turnout plunged into the mid- to low 40s. Counties where the unemployment rate has risen since 2016 also saw turnout in the low 40s, compared with 53 percent in counties where unemployment has dropped.

It could be a while. Even the fastest states won’t officially certify their results until next week at the earliest; others could take as much as a month. Complete data sets for all 50 states may not be available until January or February.