Trump’s Success Exposes the Limits of Nate Silver’s Predictions
Failed analysis shows statistics alone can't explain elections
As of this week, Nate Silver — FiveThirtyEight founder, political prognosticator, and wunderkind — is giving Donald Trump just a 25 percent chance of defeating Hillary Clinton in November .
I’m considering putting a few benjamins down on Trump.
Those who remember Silver’s prediction last year that The Donald had just a five percent chance of winning the GOP nomination will understand why.
Silver, you might recall, was feted for his 2008 and 2012 election predictions. Meanwhile, skeptics of his 2012 analysis were deemed “poll truthers. ” So it seems only fair that we not immediately let him off the hook for blowing what might turn out to be the primary election of the century.
Keep in mind, it’s not just the Trump nomination that he botched. In recent years, he has gotten Super Bowls wrong, the Oscars wrong, the 2014 Scottish independence referendum wrong and the 2015 U.K. elections wrong.
We all make mistakes, of course. The difference is that Silver’s entire raison d’etre — and he has been wildly celebrated for this — rests on his ability to make picks. But really, his track record for the last, I don’t know, four years, is spotty on some pretty big things. Now, if Silver were known for his flowery prose or for cultivating terrific sources on the House Budget Committee, we might forgive him for botching a big prediction. But this is his bread and butter. This is what he does. To paraphrase “Thank You For Smoking,” Michael Jordan plays ball, Charles Manson kills people and Nate Silver gets predictions right. (Except, of course, Manson didn’t technically murder anyone — and Silver doesn’t always make good picks. Perception is reality).
[Will ‘Electability’ Sink Trump?]
I know what you’re thinking. This is when you school me on understanding the concept of a probability, right? Look, I get that sometimes the unlikely actually happens. Sometimes it rains when there’s only a five percent chance of showers. This excuse also works as a cop-out — a sort of get-out-of-jail-free card that allows you to get credit when things go right, but avoid blame when they don’t. This is like when Jon Stewart or Rush Limbaugh messes up and then get to claim they are mere “entertainers.” If your local meteorologist says there’s just a five percent chance of snow and then the biggest blizzard of the century destroys your city (an apt Trumpian analogy), it’s a pretty good bet that weatherman will lose his credibility, if not his job.
If you’re sensing some schadenfreude coming from columnists such as yours truly, you’re probably right. Silver has demonstrated more than his fair share of arrogance over the years, suggesting that political columnists, who rely on things like wisdom, experience, reporting, and even intuition, are mostly worthless. “He took this tone that journalism that was not data journalism … was inherently stupid and wrong,” explained Washington Post reporter David Weigel on a recent episode of Slate’s “Trumpcast” podcast.
But Silver’s bad Trump analysis was potentially more harmful than just causing those who believed him to have egg on their face. Ideas have consequences, and it’s plausible that Silver might bear some responsibility for some of what has transpired after his call.
Is it possible that a #NeverTrump movement might have begun sooner had Silver taken Trump more seriously? Is it possible that Ted Cruz might not have buddied up to him for months? There are too many variables, and, unlike Silver, I won’t assign a statistic to something that is unknowable. But it’s at least possible that Silver might have talked some of us into a false sense of security. After all, if Nate Silver — Nate Silver! — says you have nothing to fear, then only a fool would be freaked out.
By the way, it’s important to remember that Silver’s post was literally headlined with the click-baity title: “Dear Media, Stop Freaking Out About Donald Trump’s Polls.” This wasn’t some cold or sterile scientific analysis of Trump’s long odds, it was a suggestion from a guy who had a lot of credibility at the time that only kooks should be afraid Trump might actually win.
He doesn’t get enough credit for it, but conservative radio host John Ziegler penned a remarkably prophetic piece for Mediaite in November rebutting Silver’s prediction. He diagnosed Silver’s many mistakes — for a guy who prides himself on being analytical, Silver’s five percent prediction was mostly bad punditry and faulty assumptions. Ziegler ultimately concluded, “the time to ‘freak out’ is clearly right now. In fact, it might already be too late.”
Sadly, nobody freaked out.
The good news is that balance has been restored to the universe. We can now return to a time where journalists will hopefully utilize data as a tool, without fetishizing it as some sort of panacea. For a while there, it seemed as if we had it all figured out. Statistics, some thought, held the key to our understanding of elections. But just as it’s a fatal conceit to assume that elites and experts can plan an economy, the belief that we had finally figured out political predictions was naive. The world is too complex for such an assumption.
Roll Call columnist Matt K. Lewis is a Senior Contributor to the Daily Caller and author of the book “Too Dumb to Fail.” Follow him on Twitter at @MattKLewis.
Get breaking news alerts and more from Roll Call on your iPhone or your Android.