How we know when someone wins an election

It's election day, and starting later in the day, you're going to hear a lot of this. That's according to a race call from the A.P. That's according to a race call made earlier this morning by the Associated Press. But what does a race call mean? And why the Associated Press? Well, let's find out. It's the NPR Politics Podcast. I'm Tamara Keith. I cover politics. And I'm Domenico Montanero, Senior Political Editor and Correspondent, and I want to know who then what this Associated Press is. We're going to find out because David Scott is here from the Associated Press to help us with all of these questions. Hi, David. Hey, how's it going? Good. We are so glad to have you here. Let's start with the basics. What's your job at the A.P.? So I am the Vice President of A.P. Elections. So everything that has to do with the vote count, with declaring winners, with our public opinion research, if it has to do with Election Day at the A.P., it's my responsibility. And that's a very big responsibility. And many times tonight, someone on our air is going to say, so and so has won a race based on a call from the Associated Press. How do you make those calls? What is a call? So it goes back to how the American democracy functions, right? And if you look at the Constitution, it says, basically, hey, states, go have some elections. And it stops. And there's no further instruction than that. And so every state does it differently within states, counties do it differently. And two years after the A.P. was founded in 1848, we said, that's a story, right? We, who got the most votes is a story. And we don't want to wait for all the mechanics of this unspecified process to play out. We're going to go out and we're going to count the vote. And we're going to find out who got the most votes and we're going to break that news. And so because it was a story, and that's what we do, we cover breaking news at the A.P., we started counting the vote on election night. And we started declaring winners and saying, who got the most votes? Are we certain that the trailing candidates can't catch up? And when we are certain, that's when we declare winner. And I want to make a distinction for our audience on what the A.P. does in making calls versus what we might hear on other broadcast networks about projections. Maybe you can sort of like, you know, distinguish between those two things and why you guys say that you're making calls versus what projections are. Fundamentally, what we're doing and what the broadcast networks are doing with their decision teams is the same thing. It's a choice that the A.P. has made deliberately not to use the word projection. It's a reminder to us about the headline that we create, the news that we tell when we say someone has won a race. So it is a little bit of a rhetorical device that we say, this person has won a race because the certainty required for us is that word. It's certainty. We are certain based on all the data, on all the evidence that we are looking at that the trailing candidates can't catch up. And so we don't give ourselves an out by saying projections or an apparent winner. I don't think anyone uses apparent winner anymore. But it is really a reminder to us about the seriousness with which we take our race calls and the role that we play in the democracy. We're going to call 7,000 elections today, well, maybe not all today, but be nice if we gave them all done today. I mean, we could dream. We've got a few weeks, I think. Yeah. But in this general election, 7,000 winners and more, you know, because some will go to run off and it's a, again, it's a reminder to us that we have to be certain. So how do you gather all of the information that you need to be certain to make those calls? It starts with the vote count. It's the most important thing. And one of these things I really sort of like saying is that every number you see in a graphic on election night, every number you see on a screen or you hear on the air, every number is a vote. Every number is the voice of someone who said, this is who I want to represent me. This is who I want to lead in this country. And so we start with that. We start with the vote count, which means we are going out to about roughly 5,000 reporting units, counties, towns in New England, parishes in Louisiana to find out who got the most votes. And you're literally sending people. We are. It's not the only way. We're like any good journalist, journalists, we like multiple sources, right? So we send a reporter to a county election office to physically, they're not actually touching the votes. We don't do that at AP. That's the election officials. But they may get a piece of paper that says, here's our tally to this point or they may be posted or they may get told directly in an interview. So we're doing that. And then counties or towns are also posting their results online. And sometimes it's a really sophisticated website that they have. And it's great. And sometimes it's a picture of a whiteboard that they post to Facebook. Is it early? Yeah, absolutely. Yes. A hundred percent. Or they might email us a photo of a handwritten note. Here's our tally. And sometimes it's something really sophisticated like a feed from a secretary of state's office. So multiple sources, right? And do, are they all the same? Are you getting the same story, depending on those sources? And then we look at those votes and we would do what we call quality control. So like, again, do the sources align? Or do we see anything in these results that make us think that maybe election officials made an early mistake? It doesn't happen often, but it does, you know? And then we publish. And we send out those votes counts to NPR and all of our media customers and our commercial customers and really the world and they're getting sort of this real time look at what the state of the vote count is and the state of the election is from our vote count, which, if it wasn't for us, you know, you wouldn't have that. You'd be waiting because every state does it differently and the timeline is really long. Well, it's a big country. I mean, how are you able to put that many people out there? How many people is it again? So on election night, we'll have about 5,000 people. 5,000 people. Yeah. We like to brag and call it the single biggest act of journalism there is for a single story. It's a huge effort and it's a full time effort. And so what you know, we do general elections in this country every November and then obviously in even number years, they're a lot larger, but AP elections is a full time job for me. It's a full time job for my team. It's a full time job for our staff, not not 4,000 people. Those are folks that we hire for the night. But it's full time work to make this make this go. Now there are some races that get called almost immediately, like the polls close and AP says we can call this race that race with 0% in like no votes. How are you making those calls? Yeah. How are you doing that? So if you look at a very small number of races for statewide office where all of the evidence is that the traditional, very well established dominance of one party will carry through into this election, you can call a race at poll close. And so the registration is overwhelmingly in favor of one party. The early voting statistics that we have are overwhelmingly in favor of one party. The history of the state is just dominant by one party. The candidate quality in which you see on the campaign and how much that dominant party spent on advertising compared to their challengers. And then we do a voter survey and we talk to a sample of American voters. We do that in partnership with the US Broadcast Networks. It's called the voter poll. And that's the last piece of confirming data. And so all of that taken together tells you that a Republican is going to win the statewide office in Wyoming, which is exactly what you would expect. The poll close calls get a lot of attention. It's a tiny number of the calls that we'll make on election night. So can tell us a little bit more about the voter poll, is it an exit poll? Is it a poll beforehand because of all these people that are voting early? Is it a mix of both? You guys used to do something called vote cast. So how should people think about what they're hearing when they hear that there is this information as part of the voter poll? So what the voter poll is, it's an evolution of what AP did with AP vote cast, which was our survey of the electorate and what the networks did with the exit poll, traditional exit polling. I think the goal of both us and the networks has always been to understand the electorate in full and in totality. And so what we were doing with vote cast was really great at understanding people who voted early. And what the exit poll did was really good at understanding people who voted on election day. And so we're working together. And we brought those things together. We hired a great company called SSRS to execute that survey for us. And so it is designed to, however you vote in America, we want to know what you're thinking and who you voted for and what you cared about and what motivated you to cast your ballot and what motivated you to make the choice that you did. And so we use multiple methodologies, including traditional exit polling to conduct that survey. And Tam, when people hear me talk about, you know, this group voted X way or changed from the last election by this many points, or you know, Trump won, you know, the highest percentage of Latinos in history for a Republican presidential candidate. That's all coming from these kind of exit polls, slash pre-election poll, you know, numbers that are coming from the networks and the AP because that stuff just isn't all written on the ballots. Right, exactly. No one's going to the polls. And no one knows, like, information about you. There isn't, like, some metadata within the ballot. None of that information about you is in some sort of ballot. It's all part of what APN the networks do. And then others will go later, like the Pew Research Foundation and do what's called validated voters survey where they try to find exactly those people who did vote and then verify that they actually did. All right, we're going to take a quick break and we will have more in a moment. And we're back. And David, I want to ask you maybe a slightly uncomfortable question, which is, has AP ever gotten a call wrong? And what did you learn from that? Yeah, we have. And we do. Let me start by saying elections, and this is not making an excuse because our standard is 100% accuracy. And our standard is certainty. Elections in the United States are largely exercises conducted on pieces of paper with pen and pencil counted by humans. And it's an amazing thing when you think about it that more than 100 million ballots will be cast today, somewhere cast in advance, but ultimately 100 million. And most of them are still pieces of paper counted by people using tabulators and using machines. But still, we're still a paper ballot democracy. And humans do make mistakes. We try really, really hard not to make them when we're calling races. Again, 100% is our standard. Anything less than that, we've fallen short. Certainty is the standard. For us, we are certain that the trailing candidates can't catch up. This year, heading into today, AP has declared winners about 12,000 winners throughout the primaries and special elections, and four of those race calls were incorrect in just this cycle, just this year, just this year. So since the start of the year heading into today, we made about 12,000 calls, and we had four incorrect. One was incorrect because we got bad information from election officials. They reported the votes wrong. One was incorrect because a candidate who was ineligible for the ballot was reported as getting votes by election officials. And as it turns out, that that person wasn't eligible to get votes, and we took them out, a race went to a runoff rather than a clear winner. One was a technical mistake. We had data come in, and our team just made an error in how we were getting the data into the system. And then one was a mistake of analysis. We underestimated the amount of votes that were left to be counted. Every single one of those gets picked apart. Every single one of those gets studied so that we can do the work tonight, starting at six o'clock, not to make a mistake when we declare a winner. And it's the thing that keeps me up at night, and it's also the thing I'm most proud about from our team. Well, what was the one with the underestimated vote? It was a state legislative race, and it was-- And those are hard. You know, what you have in those smaller races is you don't have the benefit of seeing patterns develop across an entire state. The one that you probably still have some nightmares about that you got right, ultimately, but game pretty close was Arizona in 2020. Talk a little bit about that and what you guys have learned from what happened there, because you initially, early on, essentially, said that Joe Biden had won the state, and then the margin got very, very, very close to potentially it being overturned. Yeah. Look, there are things about that race call that in hindsight, we would do a lot differently. But at no point were we worried that Joe Biden wasn't going to win Arizona. Was it a comfortable few days while they can't out the rest of those votes? No, it wasn't. But if we had been certain that Joe Biden wasn't going to win, we wouldn't have called it. And if any point our certainty hadn't held, we would have taken the call back. And we didn't do that because we were certain that he would win. Again, we learned a lot from that part of what we learned is how we analyze advance vote. Arizona is a state that has multiple ways to think about how they cast their ballots. You may have heard the phrase the late earlys. And that was something that we have adjusted how we think about advance vote and how we think about using advance vote into clearing winners and trying to understand with certainty someone has caught up or has the ability to catch up. But again, it's like I said earlier, every single race call where we make a mistake, we pick it apart and we try to learn from it. One election that was sort of the foundational earthquake of all of this, which by the way, now is so long ago that there's a lot of voters in this election who weren't even alive for it. But the 2000 election where we had this issue in Florida, where some networks called the race for Al Gore. They had to take that back. It wound up going to George W Bush. It was essentially some 500 plus votes in one county with these weird ballots. Now you guys didn't get that wrong. You never called it for Al Gore. Talk about that and about kind of how all the race calling sort of changed for you guys. And obviously for others, but how it sort of bolstered the model that you're talking about. I think the difference between today and 2000 is just the level of sophistication that both AP and everyone who's running decision teams puts into declaring winners, right? First off, we have so much more information than we did than our polling is better. The voter poll is substantially better than the exit poll that we had in 2000. The amount of information we have about types of ballots and when they were cast and when they were counted is election officials just provide us with so much more information than they did in the past. State of Florida, boy, completely redid how they count ballots after 2000. And if we wanted to have a democracy where a complete, fast, accurate count was the standard and the value, every state would do it like Florida. Now some states value different things. So California values maximum participation. They've made a different choice. Both of them are completely valid and not for me to say which one is better. But we just, we have so much more information and the level of sophistication of what we can do with that information is just a lot different from 30 years ago. - So like other than our need for instant gratification, why is it important to do race calls? Why not just wait for elections officials to put things out slowly? - 'Cause it's the news, it's a story. And I think in the same way that why is it necessary for the Congress team at NPR to cover votes and committee hearings? Why don't you just wait until the president signs laws and then put out a list of everything that he signed into law, right? We cover the process. Also, it's respect for the voters. I mean, we're talking, it's NPR and it's AP and let's what's AP's role in this. We are merely a reflection. We are a mirror of what the voters are doing on election day. And I talked to my team a lot about that. That every number is a vote. Every number is a voice of an American who has said, this is who I want to represent me. This is who I picked to lead. And they deserve to know what that outcome is. And so we, like I said, this is news. This is a story and we tell stories. And we break news and so that's why we're doing it. And also, arguably, the absence of information leaves room for bad information. One of the things that our team talks a lot about is we don't ever want to go fast, right? If the question is, how do you balance speed versus accuracy? It's always accuracy at 100 and speed at zero. But at the same time, if it's very clear that someone has won a race and we haven't called it, it does create a question. Why hasn't AP done this? Why hasn't this race been called? And so because we're in this business, we take on the responsibility of calling those races and not allowing misinformation to enter into the ecosystem. But with so much misinformation that's out there, is there some degree to which there is actually oversight from journalists like the AP going out to each of these precincts or counties, to so many of them that really didn't exist in the 1800s or the early 20th century, that it gave more room actually back then for people to cheat. The idea of now there are independent eyes. All five, that was enough of them all over the country. And then all of your listeners, all of the audience of the AP, everyone who's looking at our data through the many media customers that we serve, there's this conard, right? Every election is the most important election in history. - Yes. - Okay. But every election moving forward will have more attention on it and more scrutiny on it, just because of that's the way. And it's no different than any other type of story, any other type of trade on the stock market or whatever the case may be. We live in an era where people are paying attention more closely to everything that happens. And that's fine because we're out there to do the job. - Domenico, why does NPR rely on AP? - Well, I think you've heard David very thoughtful answers to a lot of what we're talking about. And I think our audience is smart, right? And I think-- think that what we value at NPR is getting it right a little bit more than we value, you know, being absolutely the first. And I will tell you as somebody who came from commercial television or worked a significant portion of my career in that, getting it first was way more the thing, that speed was way more the thing that they really wanted to tout, especially internally to be able to say we got it first, we did it first. And that's important, right? You don't want to be like days and weeks behind the story, right? But getting it right, having a degree of patience and being able to know what's under the hood for why the decisions were made the way they were. And I think that makes us most comfortable with how AP does what they do. Yeah, an AP has vast resources dedicated to this in a way that we're relying on them. Yeah, we don't have 5,000 reporters. No. And we're happy to have them. We only do an election day. Fair. He's like, wait a minute. We're both non-profit organizations. So both of you, what should people be watching for on election night itself in terms of following results, understanding how the night is coming together? Well, I mean, there's the actual analysis of what's going to happen on election night and what things we need to watch. But I think that as the night progresses, people need to be patient and understand that every state's got different poll closing times, right? And some states are in split time zones. So you're going to see data come in from Florida, for example. They close tonight at 7 o'clock Eastern time in most places, but because of the Panhandle, the entire state doesn't close until 8 p.m. So you're going to see lots of results on our website coming in from the AP, but you're not going to see race calls for statewide races until the polls close in the Panhandle, which is in that central time zone. It's a big country. And one of the things that someone asked me, "Why do you have this job?" and I was like, ask Alexander Hamilton, right? Like ask the founders, right? They're the ones who decided to create this system that basically delegated all of this out to the states. It's a big country. Every state has chosen to value different things and how they organize their elections. By the time Florida gets done basically counting its votes, polls will just be closing in California. And we work really hard to tell that story as quickly as we can, but accuracy is the most important and the states are different. You think about a huge urban county like Harris County, Texas and all that goes into how they count their ballots and how they set up early voting and the electronic tabulators and the army of people who will count those ballots. Think about Los Angeles County, which is like the fifth largest state in the country, right? And then you have these townships in Maine that may have 15 votes, 20 votes. And they're essentially the same in our system, right? They're a reporting unit. And so that's what election night is like in America. It is not like what you see in the Brazilian election from a couple of weeks ago where it's a very straightforward national system where there's a lot of homogeneity between how it's just they do it one way across the whole of the country. That's not how America is. Every place gets to do it different. They get to pick their own way of doing it. They get to pick their own rules. And that story takes a little bit of time to tell. It's less rigable, too, by the way, for the people who think that it's that there's problems going on across the country. It's a lot harder to rig an election where it's decentralized. And our system is very much decentralized. Yeah. All right. Well, David Scott from the Associated Press. Thank you so much for spending some time with us today. Yeah. Happy to do it. We will be back in your feed with insight and analysis from the election on Wednesday morning. I'm Tamara Keith. I cover politics. And I'm Dominica Montenarro, Senior Political Editor and Correspondent. And thank you for listening to the NPR Politics podcast.