Sam Wang's projected electoral college map as of 11/3/12
As Election Day approaches, there's been an interesting dust-up surrounding some of the predictions being made by various analysts, especially the ones who specialize in poll aggregtion and statistical modeling. Nate Silver and Sam Wang are two of the best known, but there are many others.
Silver has particularly been the target of commentators and pundits, mostly on the conservative side (though not entirely) who don't seem too happy that the NY Times blogger is forecasting an 80% chance of an Obama win on November 6. National polls indeed show a race that is tied, but, as any 8th grade civics class knows, we don't elect our presidents through a national popular vote. State polls show Obama has a small but consistent lead in electoral votes, a lead which appears to be increasing in the last few days, not receding.
It's interesting to look at what these commentators are saying about Silver and other analysts who forecast a Democratic win. They say the stranges things. Ta-Nehisi Coates of The Atlantic discussed this in a recent post that was headlined,"It just 'feels' like your numbers are wrong."
Yes, some commentators actually get as silly as to say things like, "it just feels like it's tied in Ohio."
It's important to put this in proper perspective. Nate Silver and his cohorts have been working for years to develop statistical models that forecast how people will vote based on the way they respond to surveys. Silver has also built into his model various economic indicators to supplement the polling data. These models have been getting better all the time. In 2008, Silver called the right outcome in 49 of 50 states, and he predicted the final tally in the popular vote to within 0.1%.
Silver is an Obama supporter, though one could read his posts and never know that. He comes from an unusual background--including work as a baseball statistician, a practioneer of sabremetrics. That experience gives him considerable credibility, in my view. Sabremetrics experts introduced something into baseball that it lacked: clear thinking.
It's all about debunking misguided perceptions
Here's an example: it's conventional wisdom in baseball circles that walking a hitter leading off an inning is worse than allowing a hit. Supposedly, runners given free passes score more often than those who reach base safely. This belief is ingrained in baseball thinking--from the dugout to the press box to the loud mouth fan in the seat.
But it's wrong, and analysts like Silver have proven it wrong. Studying thousands of occurrences over the past 30+ years, they showed categorically that 38% of the time a hitter who is walked leading off an inning comes around to score--the exact same percentage for those hitters who got on base with a hit.
Silver, Wang and others are not flatly predicting an Obama victory. In fact, they go to great pains to point out that they are simply stating probability based on well formulated models. As I write this, Silver's model shows Mitt Romney with a 1 in 5 chance of winning. That's not a great chance, but it's certainly not a longshot.
As a planner who must always assess market situations and future trends, I cannot ignore the lessons of all this. While statistical modeling is by no means perfect, it is an extremely powerful tool, a tool that has consistently gotten better over the years in part because of the technologies at our disposal.
And while in marketing we don't always have the budget (monetarily or timewise) to construct tight analytical models, it certainly makes sense to do our very best to look for hard data instead of simply jumping to conclusions because things "feel" a certain way. This is especially true when it comes to looking a social and cultural trends.
I'll address this in another post in the next few days, but the gist is this: Social and cultural analysis can't simply be based on offhand, impressionistic assertions. If we're going to predict that society is changing in a big, crucial way (or even in smaller ways), then we had better have something convincing behind us to provie it.
Comments