Blog Post

When it comes to elections, data journalism without context is noise

It’s election day, and my hopes are high. Not because I expect my team will win, but because it should mean the months-long slog through midterm election forecasts will finally be over. What was once novel — and, for many Americans, relatively isolated to Nate Silver at the New York Times over the past few elections — is now ubiquitous. And it’s becoming a case of information overload.

At the Times, Silver and a small group of contributors published a blog post daily, give or take, leading up to the election. At his new site, FiveThiryEight, he and his team publish lots of posts breaking down the numbers about who’ll win everything from control of the Senate to individual gubernatorial races (the actual forecast page is here.). Unless you’re a politico, your brain probably doesn’t have space, or your day enough time, to absorb the shifting chances of Republican versus Democratic control of Congress, who’ll have to win if someone else loses, and the myriad polls constantly coming out and affecting the forecasters’ models.

That’s right, forecasters. FiveThirtyEight isn’t the only game in town anymore. Silver’s former stomping ground, the Times, has copied the FiveThirtyEight formula with its The Upshot blog, right down to election forecasting. Silver’s stomping ground once removed, the Daily Kos, has an election forecast, too. Ditto the Huffington Post and the Washington Post

If you really want to geek out, meander on over to the Princeton Election Consortium, PredictWise, the seemingly much-less-scientific Rothenberg Political Report or any number of other sites where statisticians apply their craft to predicting political races. Or don’t. The Upshot handily shares the latest predictions from several forecasts. They’re largely the same.

Source: The Upshot
Source: New York Times / The Upshot

There are plenty of situations when the prediction is the thing that matters, but politics shouldn’t be one of them. Predicting who’ll win is just the starting point, and we probably don’t need dozens of people, publications and websites doing that on a national level. We know now that Silver and his peers can predict winners with high accuracy, so for the sake of everyone’s sanity and the nature of political discourse, let’s stop obsessing over the numbers and start obsessing over what they mean.

The nature of the midterm elections might lead to part of the problem with forecast overload. Without a presidential race to make the focal point, there’s no equivalent to the home screen, the top Google News headline or a count or perhaps an activity score on your fitness tracker — the types of things many people want to see first before diving into the weeds, if they choose to dive deeper at all. When the only races of national import are congressional and gubernatorial, there’s only weeds.

This is all the data most people really need to see. Source: FiveThirtyEight
This is all the data most people really need to see. Source: FiveThirtyEight

But the bigger issue than just being bombarded by lots of numbers that aren’t that interesting is a lack of context with our election data. We get it: Republicans will likely control the House and the Senate. A shift of a couple confidence points in either direction probably doesn’t necessitate its own detailed analysis.

Tell me why the results of these elections matter. Who are these men and women we’re voting into office and what do they stand for? Is there any chance they’ll actually be able to pass any legislation, or is a 53-47 Republican majority in the Senate not that big a deal with a Democratic president and a 60-vote threshold to pass many important bills?

Here’s a chart I’d like to see: a sliding scale of the chances we’ll actually see meaningful movement on issues like climate change or immigration based on the makeup of Congress after this election.

I like data journalism as much as the next guy, but when it comes to elections, I don’t need the data behind the data. I need to learn about what comes next. If data journalists can’t deliver that, it’s a quick glance at one chart on one site, and then off to find information somewhere else.

Newspapers still exist, right?