What do you think? Leave a respectful comment.

How FiveThirtyEight calculates the data of a divided nation

Nate Silver's FiveThirtyEight blog burst onto the political scene in 2008, when he forecasted the popular vote for president within one percentage point. Becoming a key point of reference during elections, he started a podcast in 2016, and says his methods continue to be more reliable than popular narratives. So what do they say for the 2018 midterms? NewsHour Weekend’s Christopher Booker reports.

Read the Full Transcript

  • Jody Avirgan:

    Hello and welcome to the 538 politics podcast, my name is Jodi Avirgan, we are 8 days away from election day.

  • Christopher Booker:

    As he has done many times times before, host Jody Avirgan has assembled the "538 politics" podcast team to talk about the upcoming election.

  • Clare Malone :

    I think everyone knows this happens. Which is people call up reporters and they say this is why we are going to win and this is why the state is looking rosy."

  • Christopher Booker:

    In a crowded field of over a half million podcasts, the show has emerged as a must listen ranking among to the top 150 podcasts available on itunes.

  • Nate Silver:

    The midterms are really important. It's the country's first major chance to say how it feels about everything that is happening.

  • Christopher Booker:

    538's forecast model burst onto the political scene during the 2008 Presidential election, when founder Nate Silver calculated the results of the popular vote within one percentage point.

    In 2012, he got it right again this time forecasting the electoral college votes correctly 332 votes for Obama, 206 for Romney.

    But 2016, was a bit different. 538's forecast model sat well outside much of the popular narrative. Reuters had given Hillary Clinton a 90% chance of winning, The New York Times model gave her an 85% chance of winning. While 538, only gave her just over a 71% chance.

  • Christopher Booker:

    There was criticism before the election that you had the probability of a Trump victory around 30 percent. It was amongst the highest of the forecasts and then there was criticism afterward that you didn't have the probability correct. Did 2016 change the way you model probability?

  • Nate Silver:

    No, because we think our forecast was right. that very few people thought it was a competitive election. If you looked at the data carefully and built a good model, you would have known that Trump winning was not that big a surprise.

  • Christopher Booker:

    While they haven't changed their model, they have changed the way they describe their forecasts….emphasizing 538 trades in probability, not prediction.

  • Clare Malone:

    We have thought about people's misinterpretations and I think everyone, all news consumers are getting smarter. Now we do it in, "You know this person has a 1 in 6 chance of winning" so it's more the vernacular of when we talk about everyday life and we talk about gambling or chance we usually do one in whatever the number is.

  • Christopher Booker:

    Silver says, this clarifies what the forecast is indicating for the audience.

  • Nate Silver:

    Well first of all people don't confuse it as much Because they think 20 percent, they think it means "Oh, Trumps only getting 20 percent of the vote right" instead of having a 20 percent or 30 percent in our model chance, It forces them to think about what it actually means right? But if we have a model that gives Democrats for example one in six now chance of holding of winning the Senate. That's not just meant to like hedge our bets, we're saying that given the uncertainty in polling that if you have this election six times under similar circumstances one of six times they would, they would wind up winning.

  • Jody Avirgan:

    I wonder if it that is worth parsing out?

  • Christopher Booker:

    According to Silver, the podcast gives the team a wider avenue to explain the complicated methodologies that drive 538's forecast models.

  • Nate Silver:

    I think It's a way to make content that's like pretty technical and pretty dense, more accessible to people. It's a way to let people kind of see our thought process.

  • Micah Cohen:

    This is what is so tricky about covering politics, is, getting the facts right is hard enough, but what are you putting emphasis on, what are you spending the majority of the time talking about.

  • Micah Cohen:

    I think it's good for journalism to level with the reader or the listener or the viewer and say, "hey here's what we know, here's what we don't know, here's how we're thinking about it." I think that helps and I think that builds trust.

  • Clare Malone:

    I thought it could be helpful for readers to kind of see right next to each other Republican spin and Democratic spin on the Senate map, which we are currently giving the Republicans a better chance to control the Senate.

  • Micah Cohen:

    It's I mean it's huge advantage of having the model, as the model keeps us in check. It forces you to take a step back and take take a long view. Calm down. You know it's very easy in the news cycle to get wrapped up in things.

  • Nate Silver:

    "I remember like, being pretty shocked by that, and that crossed a line and thinking that the way the media covered that was inappropriate."

  • Nate Silver:

    You know a big part of what 538 does, a big core part of like what we're known for is debunking media narratives that are overblown or overdramatic or overwrought or whether there is groupthink around something and typically like in an election campaign that gets worse and worse and worse, things that make a 1 pt difference in our model, people will talk about them like it is a 5 pt dramatic swing And so it's always like a little bit more more even keeled I think

  • Nate Silver:

    "There are so many things that go down the memory hole. I mean like, who is talking about Brett Kavanaugh this week?"

  • Christopher Booker:

    Is there one specific thing you can cite that you would change for the way media covers politics?

  • Micah Cohen:

    Ah Just one? I would have the media be smarter about selecting what tool in its journalistic toolkit it's using to answer what question. So in other words the question of who's going to win the House is really a question that's best answered with data that's just empirically true. Looking at the polls, fund Raising doing all that in a smart systematic way is better at answering that question than going to Allentown, Pennsylvania and talking to six people. Now that data is not a good tool to answer other questions you know what will Trump do on health care if he wins a second term. You know you can't use data to answer that question.

  • Christopher Booker:

    The narrative of 2018 is that we are a divided nation. Is this reflected in the data?

  • Micah Cohen, Nate Silver, Clare Malone:

    Yes, yes (all of them).

  • Nate Silver:

    For sure. Although less in some ways than in other elections we have in our forecast a 110 or something congressional districts where we think either party could win. There are 10 or 12 Senate races that either party could win out of 35. The times I'll really start to buy into critics of what's happening to American democracy will be when people give up and aren't turning out to vote. So this is a very 538 thing, right ? America is as divided as ever is like is like mostly true. But I think we kind of instantly are looking for the ways in which it might be slightly not true or might be a little bit exaggerated as a narrative.

  • Micah Cohen:

    It's pretty true though.

Listen to this Segment

The Latest