The Washington PostDemocracy Dies in Darkness

It took more than twice as long to call House races in 2020 than 2018

It’s not clear how much of this is pandemic-specific

The Hunter’s Moon rises behind the dome of the U.S. Capitol on Oct. 9 in Washington. (Alex Wong/Getty Images)

There is no inherent reason that it matters how long it takes to count votes cast in an election. A vote is a vote, and whether it is tallied at the instant polls close, or by noon the following day, or a week later, it is still the same vote it always was.

In recent years, though, the duration of counting has become a source of increasing agitation for Americans. One reason is simply impatience: We had come to expect over the past few decades that most races will be called on election night and that those not called wouldn’t affect the balance of power in Washington.

Then there’s the reason that came to the forefront in 2020. President Donald Trump, picking up a line of argument from Florida Gov. Rick Scott (R) in 2018, suggested that counting mail-in ballots over the hours and days that followed the election allowed for fraud to occur — something neither Scott nor Trump was ever able to show had occurred. But this idea has become something of a mantra on the right, in large part because those votes have heavily come from Democratic voters and, therefore, offer a ripe target for depiction as nefarious and illegitimate.

There’s a third reason for the agitation, of course. In 2020 in particular, counting ballots really was quite a bit slower than it had been two years before.

Sign up for How To Read This Chart, a weekly data newsletter from Philip Bump

This is admittedly somewhat subjective, since there are different mechanisms and triggers that prompt an official race call. Many news outlets, including The Washington Post, rely on the Associated Press’s calls in federal and state races to determine official winners. So, to evaluate how rapidly or slowly states were counting, we pulled the AP’s formal race call in each of the 435 House races for which it made a call in 2018 and 2020 and assessed how long it took to be made.

Different states close their polls at different times, of course, but our interest was in determining when results were confirmed generally. So instead of comparing results to poll closing times, we compared them to a universal metric: noon of Election Day on the East Coast. A race called as soon as polls closed in California, then, would be recorded as happening 11 hours after noon Eastern.

Here’s 2018. Because some races took far longer to call, the graph below is shown on a logarithmic scale, from about 6 p.m. Eastern on Nov. 6 until more than 300 hours later.

You’ll notice — and may perhaps remember — that several of the races that took longest to call were in California. A combination of heavy mail-in balloting and lots of House races consistently means that California races come in more slowly than other states. The same holds for Washington state: Voters vote by mail; that it’s on the West Coast has little to do with it.

One of the last races to be called, though, was the race in Utah’s 4th Congressional District — a race that took a long time to call because it was very close. That’s the other factor here, of course. Races with narrower margins (shown with smaller circles) take longer to count and then to adjudicate. So calls come later.

With 2018 as a baseline, let’s move to 2020. During that year, you will certainly remember, states relied more on mail-in ballots in an effort to reduce the threat of coronavirus transmission. Notice not only that the number of races that took a while to settle is larger below, but so is the time-frame the graph depicts.

Again California stands out — but so does New York. Changes to voting in that state slowed election results dramatically.

Overall, the average House race in which the AP made a call in 2018 had a winner identified about 15 hours after noon Eastern on Election Day. In 2020, that extended to nearly 35 hours.

A lot of that shift is a function of New York, where the average increased a lot across its large number of House seats. It’s also shaped by the increase in Alaska, where it took far longer to determine the winner in the state’s sole House race than it had two years before.

It’s interesting to compare Virginia and Washington on that graph. The arrow indicating the change in Washington starts further from the left-hand axis, in part thanks to the time difference. But while Virginia’s results were an average of 12 hours slower in 2020 than in 2018, Washington’s were more than 18 hours slower than they had been two years prior. This despite the state’s experience with mail-in ballots.

The biggest increase was New York, where the results came, on average, 200 hours later than they did in 2018.

Again, there’s nothing about this that is particularly fraught. For one thing, there’s no evidence that counting more slowly injects illegal ballots — or even makes such an injection more feasible in the first place. It’s also not clear whether 2020 was an aberration caused by the pandemic or a mark of how long it will take to know House results moving forward.

The moral, as always, is to be patient. In 2018, the state that saw its last House race called by the Associated Press the earliest was Kentucky, where the last race was called before 9 p.m. Eastern. But the last race was called in each state at an average of 11 p.m. Eastern on the day after Election Day. And in 2020, it took more than three days for all of the results to come in for an average state. That’s a bit after noon Friday.

That year, of course, the biggest race wasn’t called until Saturday: Joe Biden’s presidential win and Donald Trump’s reelection loss.