The past year has seen the rise of data-driven journalism websites that were intended to replace the type of horse race coverage that typically dominates political news. But in an ironic twist, election forecast models have become the horse race coverage of 2014.
Nate Silver was a web pioneer of the concept of employing rigorous modeling based around aggregating polling data to predict the probabilities of election outcomes. In 2012, as many political reporters speculated about how various gaffes and news events would affect the outcome of the race — or exaggerated the importance of fluctuations in national polls — Silver, then at the New York Times, calmly and repeatedly reminded everybody that President Obama was the odds on favorite because of state-level polling data.
Recommended Stories
Since 2012, there has been an explosion in outlets using this approach to election analysis. Silver himself founded FiveThirtyEight, while the Times and the Washington Post started their own election models.
But what’s happened this year is that political coverage has moved from feverishly covering horse race polling, to hyping up daily fluctuations in predictive models of which party will control the Senate after 2014.
For instance, on Tuesday, the political news world was shaken up when the Post model gave the Democrats a slight 51 percent probability of maintaining control of the Senate. But then a number of polls came in that were favorable to Republicans. So by Thursday night, the model was back to giving the GOP a 62 percent chance of gaining control.
This isn’t a knock on Silver or the Silver-inspired approach to polling analysis. Part of what’s happening is that compared to a presidential race, there are fewer Senate polls, many of them have been contradictory, and several key races are close enough that it’s easier for models to fluctuate based on a smaller number of polls.
But there’s also something else at work. Political news abhors a vacuum, and when trying to appeal to a broader audience, it’s inevitable that journalists will boil everything down to the question of “who is going to win?” Data journalism isn’t changing that. All that’s changing is that people are freaking out over fluctuations in statistical models instead of just daily polls.
