The evolving art of predicting elections

“In the early 2000s there became many more players in the kind of polling marketplace.”

By Sean SaldanaOctober 22, 2024 2:56 pm,

Trying to figure out who will win elections before they happen is a ritual as old as American democracy itself, but over the years, the game certainly has changed.

The latest story from David Freedlander, a contributor to New York magazine, is titled “When election modelers attack.” He joined the Texas Standard to explain how election modeling has changed in recent years.

This transcript has been lightly edited for clarity.

Texas Standard: I’m wary of any political prognostication. I mean, the polls don’t even claim to predict outcomes. So set this up for us: What were you trying to do with this article? 

David Freedlander: I guess I’m trying to sort of, you know, address folks like you who find themselves sort of annoyed or frustrated –

We’re very cautious. 

I think people who are not political junkies don’t quite understand why political junkies, you know, seem to obsess over this topic so much when of course, like we’re going to find out who won, you know, in due course and any any amount of, you know, trying to sort of read the tea leaves to find out what’s going to happen is probably not going to meaningfully change the outcome.

And so there’s just a lot of sort of energy and anger and confusion around what these polls and the modelers are doing in their efforts to sort of predict the winner.

Well, I think you put your finger on it there. And I’ll just admit I’m one of those who follows especially the battleground state polls. Because we’re dealing with the Electoral College and the overall vote count.

But you see these polls go up, you know, two points on one side and then down four points on the other. And still there’s a two-point margin, and the margin of error is three or more sometimes.

What does it all add up to? I mean, it’s just kind of hard to glean any real meaning or significance from it. 

I mean, I think the pollsters would say that what they are doing is not actually meant to tell us the winner of the election, especially an election this close. They just don’t have that kind of predictive power. They’re not sort of calibrated to the level of certainty that’s required.

What the polls are telling us right now is that the election is close. And they’re saying that over and over again. And I don’t think they’re saying who’s going to win, I mean, because I think that’s sort of unknowable.

And, of course, you know, they’re also just a sort of moment in time. They could maybe tell you sort of like who is up right now, but that doesn’t mean who’s going to be up in the two weeks that we have left until Election Day.

» MORE: A familiar refrain: Some polls say Texas could turn blue, but can they be trusted?

Well, in the world of election predictions, I guess there’s no bigger name than Nate Silver. Can you say a little bit more about who he is, for those who don’t follow him and election prognostication in general, and what has he done to change the way we view these forecasts?

I think if you sort of back up a little bit: You know, in a previous era, a major news organization like CBS News or whomever would come out with a poll, and that was just sort of it. It was sort of taken as gospel.

And what ended up happening in the early 2000s was there just became many, many more players in the kind of polling marketplace. And they didn’t always have the best reputations or sometimes didn’t even really quite knew who they were, which is not to say their polls were necessarily inaccurate.

But what Nate Silver sort of figured out was that the best way to determine the accuracy of all of these polls was just put them into a model and sort of weigh them essentially kind of based on their like, previous predictive power. So if a pollster had gotten the answer, they got the winner, right the previous election cycle or the midterms or something, their poll would be given more weight than someone who got it wrong.

And, you know, tossing a couple of other kind of complicated factors into a mathematical equation such as the state of the economy and the power of incumbency and, you know, demographic factors. And they’re trying to figure out this sort of meta analysis of polls. It was supposed to be even more predictive of polls to tell us who was going to win.

But at the same time – and you point this out in your article – you have a number of people who have gone out there wanting to be the next Nate Silver. It’s almost become a kind of a career option for some folks. 

Nate Silver sort of became famous, at least famous as a political analyst, in 2012 when he correctly predicted the winner of every state in the Barack Obama/Mitt Romney election, even as a lot of people were sort of doubting him. And once that happened, and he achieved notoriety while other people sort of wanted a piece of that.

And so there’s now like many competitors in that space, all of whom have kind of slightly different formulas that they plug into their mathematical models to determine the winner.

» MORE: How Texas pollsters get their job done when everyone is screening their calls

But I wonder if this isn’t a kind of a scientific patina over the whole poll process. I mean, it’s one thing to go out, talk to respondents, get specific data. But once you start weighing factors, you know, you start looking over your shoulder, it seems to me a bit.

You look back at 2016, and everyone remembers that little prediction meter in The New York Times swinging from a Clinton victory to what actually happened.

I just wonder if part of this is that a desire to predict the political future is so powerful that no matter the warnings we hear about the polls, that we are putting a lot more faith in election prognostication than we ever have before as the body politic. And in a way, you have the Nate Silvers out there who are now feeding that hunger.

I think that’s absolutely right. I mean, the stakes of this election are really high. Feelings are raw. Anxiety is spiking, and people are looking for comforting blanket of certainty.

And, you know, they’re just not going to find it now, like doing what Nate Silver and the election modelers are doing is going to sort of provide you with slightly more feeling of certainty than any one given poll because they’re kind of compiling all the different numbers.

But even Nate Silver and in all the models at this point basically have the election of the coin flip. It’s essentially a sort of 50-50 tie. Now, in some days, based on the polls that come in, it might be 55-45 in favor of Harris or, you know, 53-47 in favor of Trump on the sort of probability of who wins.

But what Nate Silver and others would say is those actually aren’t meaningful distinctions. If it’s 55% Harris is likely to win there one day and 53% Trump the next day, it’s just basically tied. And you should just think of it as tied.

» GET MORE NEWS FROM AROUND THE STATE: Sign up for Texas Standard’s weekly newsletters

You know, if the polls say anything, as you were pointing out, this is a close race. We don’t know if we’ll know who won the race by the next day. It could be much longer than that, if the polls are as accurate as they claim to be.

One of the things I’m hearing more and more from some who have been noting what you’re describing here is that half the country needs to be prepared to lose. And we’re not really talking about that very much.

Yeah, that’s a great point. I think injecting that into the kind of public political conversation would be time well spent. There hasn’t been quite enough of that, I think, in the last couple of years.

But democracy is kind of dependent upon recognizing the rightful winner. And so, you know, regardless of the outcome, it would be useful for both sides to kind of acknowledge that fact ahead of time.

If you found the reporting above valuable, please consider making a donation to support it here. Your gift helps pay for everything you find on texasstandard.org and KUT.org. Thanks for donating today.