For weeks now, some in the traditional punditocracy – the folks who look at a poll, call a campaign official, then consult their gut feelings over lunch – have hammered Silver for his statistics-heavy approach. He gave a false appearance of certainty, they said. He was way too bullish on President Obama’s chances, they said. His numbers were skewed, they said.
The classic exposition of these views was “Nate Silver: One-term celebrity?” a piece by Politico’s Dylan Byers published on October 29.
Then came the actual voting. Silver got all 50 states right, down to his last-minute prediction that Florida would be a virtual tie.
CNBC pundit/host Jim Cramer? He thought Obama would roll up 440 electoral votes. We’ve gone over the states several times and we don’t see how that’s even mathematically possible.
Given this disparity, has the venerable art form of political punditry been discredited beyond redemption?
We’ve got some thoughts on that, surprise, surprise. The first is that it’s easy to make pundits look like witch doctors. All you have to do is cherry-pick the worst predictions, which we’ve done above, and suddenly a whole class of cable news analysts appears foolish.
Some pundits were right, or at least more right than Mr. Morris. Ron Brownstein of the National Journal had Obama to win, but a low predicted total of 288 electoral votes, for example. Donna Brazile of the Democratic National Committee said Obama would get 313 electoral votes, which was pretty close to what happened.
Slate has a fun dart-board graphic of pundit hits and misses, which you can peruse here.
After all, Nate Silver isn’t that special. That’s our second point. Many analysts produce prediction models based on lots of polls, plus the addition of economic indicators and other data. If you know your way around a regression analysis, it isn’t that hard.
Political scientist Josh Putnam of Davidson College did a math-based forecast at his Frontloading HQ blog, and he was dead-on, just like Silver. Sam Wang and the Princeton Election Consortium thought Romney would win Florida, but got everything else right.
Heading into the next election cycle, more and more media outlets will want their own Nate Silvers. After all, in the run-up to Election Day, 20 percent of all New York Times web visits included a stop at Silver’s FiveThirtyEight blog. In that sense Silver has dented the old way of doing things, which may never be quite the same. The future of political journalism includes more numbers. We, um, veteran types will have to get used to that fact.
But the Dick Morrises of the world aren’t going away either. In today’s polarized media landscape, one purpose is to inform, but another is to make the news consumer feel comfortable. Fox News will give lots of air time to pundits who just happen to lean Republican. MSNBC will do the same for liberals. Viewers who want to break out of partisan closed-feedback loops will need to try to discern which “experts” know what they’re talking about and which are just repeating what they think the partisan skew of the audience demands.
Because – and here’s our last point – it’s really about time that journalism stepped up its performance in this area. The politicians themselves adopted a quantification-heavy approach to their business long ago. In terms of voter analysis, microtargeting, and other techniques they’re far beyond what the media discusses. In the book “Moneyball,” Michael Lewis wrote how statistics revolutionized baseball. Well, the “Voterball” revolution is upon us, if it hasn’t already occurred.
Why do you think Mitt Romney went to Pennsylvania on the campaign’s last weekend? Given the scope of his loss, it’s clear he wasn’t trying to run up his score or force Obama to play defense in the state. No, he knew he was quite likely to lose Ohio, and Pennsylvania provided the slim chance of an alternate path to 270 electoral votes. That’s what his quants told him. Dick Morris? He was out of the loop.