Patrick Healy: Happy Election Day, folks. On Tuesday evening The Times will publish what we call our “results pages” for today’s elections — charts, maps and analysis showing up-to-the-minute data on voting in New York City, Virginia, New Jersey, California and other states with big races.
Both Republicans and Democrats hang on those results: When I was reporting on election night in 2016, and called Donald Trump on his cell around 9 p.m., the first thing he mentioned was The Times’s latest results in Michigan and Pennsylvania. In 2017, Roy Moore, the Republican Senate candidate in Alabama, showed The Times’ up-to-the-minute results charts at his election night party. Joe Biden’s campaign aides kept asking us in 2020 about results in the Arizona and Nevada nail-biters.
So when readers ask if The Times plays a role in free and fair elections in America, one of the things I point to is our work as a reliable source of accurate results.
Much of those efforts are led by Wilson Andrews, who oversees The Times’s presentation of live election results, and Will Davis, who leads the Election Analytics department that is responsible for our election night statistical model, which is often visualized as The Needle. I asked Wilson and Will about some of the work that goes into election night results at The Times.
Will, how does the election analytics team prepare for the vote count to make sure you guys go into it well informed?
WILLIAM P. DAVIS: American elections are run by counties and states, so we talked to officials in nearly every county in some states and, in other places, to a mix of state and county officials. We spend a lot of time thinking through potential scenarios for election night. We have rehearsals of the whole team in a sort of a mock election night and make sure all our code and graphics work as we expect. And then, on election night, we get the most granular results possible. That’s in addition to working with our colleagues around the newsroom to make sure that our report reflects information we know on the ground.
Wilson, your team takes all the work of Will’s team and makes those results accessible and understandable to our audience. In your years here, what’s been the most effective way to do that?
WILSON ANDREWS: We think a lot about how to be responsible with results data and our live model, because sometimes there’s a large amount of uncertainty in an election. For instance, voter turnout can be a big question mark: We saw that in the June mayoral primary in New York, where Zohran Mamdani’s victory was driven by this surge in turnout among young people that was not really expected by many experts, certainly not by the polling.
The Needle has proved to be a very popular way to visualize the results — it’s often the first thing that tells readers which direction an election is heading. But there are other features we build to explain what’s happening — we have maps that show where a lot of votes are still to be counted, and on Tuesday, we’ll have a comments thread where reporters are answering readers’ questions.
You’ve decided not to use the full array of The Needle visualizations on this election night. Take us inside how you reached that decision.
DAVIS: We will be running our model, and you will be able to see those insights on our results pages. But we have typically not published Needle visualizations in off-year elections, especially when one candidate is plainly favored. When The Needle starts out indicating a very likely outcome, it can give the impression that the race is over before any votes have been reported. So we want to be careful with how we deploy that graphic live in those cases.
On election nights, people want to know one thing: the winner. I’ve heard from readers over the years who asked why we didn’t declare winners as fast as some other news media in some races.
ANDREWS: In my time here, and I’ve covered elections here since 2014, being first to call a race has never been our No. 1 priority. It is for some news organizations. We care a lot more about accuracy than being first. We want people to know that when they are getting election results from The Times, they’re always getting the most reliable information that can be had at that moment. Now, that said, we do prioritize speed — we want to publish the results as fast as we can as we know them and can verify them.
When I was The Times’s politics editor in 2018, I remember some readers and some of my own reporters asking how long we’d wait to declare winners in some House midterm races in California swing districts. It took a while for California to count those votes. We didn’t jump the gun. In 2020, Fox News and The Associated Press declared pretty early that Biden had won Arizona. We had additional data that indicated the result might end up closer than it was at the time, though, and we held off. But have we ever jumped the gun? Or had to struggle with whether to make a race call when there was a real difference of opinion?
ANDREWS: I can’t remember a time where we have felt like we needed to make a call before we were comfortable. We don’t force it. In 2020, the information we knew on election night was that Arizona very well could go either way. And that’s why we held off on making that call.
Handling results and making race calls in a live setting is very different from reporting data for a typical article, with time for fact-checking and going through layers of editors. We’re publishing results in real time, every second, for hours, sometimes days. The type of staff you have on the team to make that possible looks a lot different. In my first presidential election at The Times in 2016, I think there were about a dozen people who worked on our results pages. Last November, our data team alone numbered close to 100.
DAVIS: We usually rely on The Associated Press for race calls, and there’s a good A.P. explainer on how they do it. They say, “If our race callers cannot definitively say a candidate has won, we do not engage in speculation.” We do make our own calls from time to time. But when we make those calls, it’s because we have our own additional data on the race and the evidence is indisputable. There can be circumstances where there’s a downside to waiting to call a race that is clearly over, like candidates spreading misinformation or the result being questioned.
Wilson, you said something to me once that stayed with me — that our election night live results work is one of the biggest news products built in the Times newsroom. What does that look like?
ANDREWS: When I started in journalism, it was very typical to subscribe to The A.P.’s data feed, produce some results tables and maps, make sure it all worked, and then go home at the end of the night and go to bed. That no longer cuts it — both in terms of needing to take in election data from around the country, verify and explain it, and in terms of stability to display the election results no matter what happens technically.
Everyone probably remembers a week or two ago, Amazon web servers went down and a lot of the internet sort of clunked out. We have a contingency for that type of technical problem to still be able to report election results if certain technical providers go down.
In addition, our reporting team researches and collects granular data in key states so that we can dig deeper into the results than the county-level data that most news organizations publish. And we have engineers, designers, reporters, statisticians and graphics editors all working together to publish the results.
When candidates show our results pages at their election night parties, does The Times have anything to do with that?
ANDREWS: No. That’s their decision. We don’t work with candidates. That Roy Moore example is a great one from 2017 because we had never seen the kind of audience demand for special election results as we saw with the special elections during the first Trump administration. And that helped us realize that there was a huge opportunity to report more deeply on this type of data.
You’ve been working on elections results for years, and as you’ve laid out, this work isn’t simple. I’m curious if there’s a journalistic issue you had to wrestle with.
ANDREWS: In the lead-up to the 2024 general election, the results team had a lot of discussions about how to help readers understand The Needle throughout the night. A key aspect of the Needle presentation is how readers first see it when they come to our homepage. It’s a dense information environment, with a lot of different elements, including live results, reporter analysis and news articles. The Needle itself is a great summary of our live model, but we wanted to tell the story behind its numbers more fully.
The ultimate solution was simple, but a first for us: a block of text that we’d update in real time throughout the night to tell readers what was causing The Needle’s movements at that time. We walked readers through different stages of the night, highlighting states where we were seeing trends emerge, while also advising where we were waiting on votes. It became a key tool for explaining how the election would unfold, and seems like an essential explanatory feature that we’ll add to our Needle playbook.
Explaining what’s happening in a moment reminds me of one of the hardest things I worked through — the Iowa Democratic presidential caucus night in 2020, when there was a real breakdown in the Democratic Party’s voting system. In those first hours of the night, we were focused on our traditional playbook of determining who’s winning, and we had to quickly pivot and start doing a better job of explaining what was going wrong with the Iowa vote and what it meant. We were hearing from readers who wanted to know who was ahead, but we really had to move from results coverage to explanatory coverage, and it underscored how much of our job is in the explanatory business.
DAVIS: I’d just add there are always races that in hindsight we could have been more focused on. There’s races this week in Texas, Maine and Pennsylvania that we’re not investing my team’s time on as much as other races. Just no matter what, there’s always some race that captures attention that you didn’t expect.
A couple of last quick things. Wilson, early on in election nights, we often get a big batch of results that show a wide vote margin between candidates, and then it closes. Do the results always tighten? What do you expect will happen tonight?
ANDREWS: It tends to be all over the map. Sometimes you get the first votes from a single county. Some release the early vote count in a single bulk report. So the first wave of results can be quite different from where the final result is going to end up. One of the things we’ve done as a standard feature on our results pages in the last few elections is what we call our “expectations text,” which tells readers: Here’s when we expect to get votes and how fast we expect to get them.
That sounds like the Slack messages I used to send you on election nights when I worked on our Politics team.
ANDREWS: Yeah. Setting expectations is a big part of the explanatory work we do on results these days.
There’s a real difference between Virginia and California. Virginia counts its votes quite a bit faster than California, which can take weeks. I do think we can be clear that we’re unlikely to know the outcome in some races on election night. Or that we know there’s a certain amount of absentee ballots that can come in after Election Day in certain states — that can sometimes make the difference in the outcome.
Last question — do you think Tuesday will be a long night?
DAVIS: I mean, define long.
ANDREWS: 2020 shifted people’s expectations that votes take a long time to count now. And it’s true: It does take longer than it used to. But these states have also learned from pandemic-era elections, and some have improved their vote counting and vote-reporting operations. New York City, in the last mayoral election, took forever to count votes. Then, in the mayoral primary in June, most of the vote was counted on election night.
DAVIS: I would expect there’d be enough information to make race calls in New Jersey, Virginia and New York by midnight. But the vote count may be close in some places, and there will be other races that will make this a long night. And that’s fun.
Wilson Andrews is an editor in The Time’s Graphics department who leads the election results team.
William P. Davis is the director of election data analytics for The Times.
The post How Two Times Editors Work on Getting Election Results Right appeared first on New York Times.
