The Slatest

Data Journalism Has Become a Confetti Cannon of Contradictory Takes

Nevada Desert Mountain Landscape Aerial View.
The hotly contested Nevadan landscape.

KatieDobies/Thinkstock

Here’s FiveThirtyEight’s Harry Enten on Twitter.

Safe for Clinton! Here’s what you get on FiveThirtyEight right now when you click through to its “2016 Election Forecast”:

Screenshot/FiveThirtyEight

That’s red, for Trump! Bwuh?

There’s a simple explanation for why this is happening, but it speaks to a larger political/media problem. The explanation is that Enten is looking at early voting data to see how actual voting is going, which apparently favors Clinton. But FiveThirtyEight’s projection system—the one it touts on its site and via its social media accounts—does not account for early voting. The model that comes up first when you click through to the forecast is a polls-only model, and right now the polls are tilting very slightly to Trump, giving him the tiniest of projected Nevada wins (46.6 percent to 46.5 percent).

Nonetheless, we still have a situation in which a FiveThirtyEight senior political writer, someone with the word “Forecaster” in his Twitter handle, is sending out a forecast that’s different than the page labeled “forecast” on FiveThirtyEight’s site. You can’t blame people for being confused.

It’s kind of our fault as readers, though: Our appetite for polling projection, aggregation, and analysis is basically insatiable, currently supporting FiveThirtyEight, the Times’ Upshot, HuffPo’s Pollster, RealClearPolitics, and several other sites. And everyone who works for those sites has their own Twitter account with which they can opine—sometimes with data support and sometimes without—about what each new poll might mean.

None of this is necessarily irresponsible: Polls are fallible, and reasonable people may differ about how they should be interpreted. (FiveThirtyEight acknowledges as much by printing roundtable chats in which the site’s various empiricists argue with each other.) Still, a style of journalism that became popular in 2008 and 2012 precisely because it cut through the noise of punditry has now become a confetti cannon of oft-opposing takes—and there’s a 70 to 80 percent chance that, for the lay reader, it can all sometimes be exhausting.

Read more Slate coverage of the 2016 campaign.