Nate Silver: ‘Bubble mentality’ is affecting the way we understand data

The 2016 presidential election led many to second guess the soundness of political polling. But Nate Silver, the editor-in-chief of FiveThirtyEight, says the polls are just fine. It’s the media and the public who could use a Stats 101 lesson.

In the media world specifically, Silver said, “a bubble mentality” can influence the way people understand data.

“When [people in media] think that they’re looking objectively at evidence, their assumptions play very heavily into how they interpret that evidence,” Silver said. In an election where the polls showed Clinton ahead but narrowly, “people took the ‘Clinton ahead part’ and stopped paying attention to the ‘narrowly’ part.”

Silver spoke with Brian Stelter in this week’s Reliable Sources podcast. A data wiz himself, Silver thinks that mediocre math literacy prevents people from understanding probability.

The media overestimated the certainty of a Clinton win, when the Democratic candidate only really had a three point lead.

The same happened in the case of the Brexit vote: The media was “very confident that Britain would not vote to leave the EU,” even though the polls were showing that the referendum results would be “as close as it gets to being a toss-up,” Silver said.

In both the 2016 election’s and in Brexit’s case, journalists “looked at who’s ahead” without paying attention to “how much that lead is.”

Lessons in data literacy

Silver recognized that FiveThirtyEight has been “a little bit prickly” towards the media, but his vision on how to educate the public to be more data literate has evolved over time.

“I used to think there are two separate sides, the conventional wisdom side and the data-driven side,” he said, but only when those two sides work in synergy are there opportunities to help readers understand complex information.

There are a few practical lessons Silver thinks journalists should take away from the 2016 election experience.

First of all, reporters need to feel comfortable explaining uncertainty in the data and disagreements between polls, the results of which differ based on the methodology they employ.

Not only is uncertainly acceptable, but “that’s where the fun is,” he said.

Keeping an eye on the margin of error is another rule to abide by. “Pollsters report it for a reason, it shouldn’t be the fifth footnote that you mention in passing,” Silver said.

All the stats that’s fit to read

When there are no presidential elections going on, the readers of FiveThirtyEight are interested in two things: sports, and other elections.

On FiveThirtyEight.com, which started out as a blog in 2008 and partnered with The New York Times until ESPN acquired it in 2014, the Montana and Georgia special elections were “generating traffic like that of a Midterm election.”

“People are in permanent politics mode,” Silver told Stelter.

Among the most popular pages on FiveThirtyEight is Trump’s approval rating tracker, which averages the results of opinion polls on the subject and weighted based on the sources’ reputation.

Currently Trump scores a 39.3% approval, down from about 45% in the early days of his presidency. Two events “really moved those numbers,” Silver told Stelter: The failure to pass a health care reform bill in March, and the firing of former FBI director James Comey in May.

Silver thinks those two events “are the two biggest developments” in the presidency’s life cycle.

The public, Silver told Stelter “is fairly rational, waiting for major stories before they reformulate their opinion about the President,” and relatively unfazed by presidential tweetstorms and other character manifestations by Trump, including his attacks on the media.

Bashing the press “serves two strategic purposes for Trump,” Silver told Stelter. “It can be a distraction from other stories,” like the unsuccessful health care negotiations, the Russia probe, or the North Korean crisis. It also has the added benefit of unifying his fragmented constituency.

Trump is “a pretty effective troll,” said Silver, but the media is guilty of taking the bait and losing sight of other important stories: “I don’t think it serves the media well when they drop those stories to cover the president tweeting about Joe and Mika, for example. In some ways, that brings out the worst in both sides.”

Listen to the full podcast with Silver and Stelter here, and subscribe on iTunes.

Exit mobile version