By Fareed Zakaria
When I was getting my PhD in political science, the course I probably found least interesting at the time was the mandatory one on statistics. And yet it has probably been the course that has, at a practical level, been the most useful in helping me understand politics, because it gave me a framework and understanding of how to analyze data. So to watch the divisions during the presidential campaign between political operators who think that politics is all art and no science on the one hand, and the statisticians on the other, has been fascinating.
I do think that the social sciences – even economics – are quite different from areas like physics and mathematics, because subjects like international relations require analysis to be textured and historical. However, it is also clear to me that one area where statistical methods have worked very well has been in analysis of voting, because this is an area where you have lots of very clear, tangible data. And you also have much repetition of the experiment, allowing you to reduce the chances of misleading anomalies.
What has this meant in practice this year? Statistician Nate Silver and others have turned these tools on American elections. And in Silver’s case, the results have been extraordinary – he looks to have correctly called all 50 states in yesterday’s election.
So why so much handwringing among political operators? Because Silver’s predictions suggest that a lot of what they are paid handsomely to do is actually worthless. Silver’s model has shown Obama in roughly the same place for the last four months – through the conventions, the debates and up until polling day. This suggests that the manipulations and tactics employed by the campaigns over this period have had little impact despite the fact that political operators get paid millions of dollars to effect week-to-week and month-to-month movements in support.
This isn’t just an intellectual divide. You can’t help but feel that when people like Karl Rove were saying that the momentum was with Romney, and were confidently predicting that he was going to win, that it was also in their own interests to be saying such things. After all, this is how political operators make their money, and they are keen to report to candidates that momentum is on their side, suggesting that as long as the candidate keeps writing checks, then he or she has a better chance of winning.
But I think the bigger and more fascinating takeaway from this debate is that the combination of the sophisticated statistical analysis (which can only provide us with this kind of accuracy when it is a big presidential election, during which there are many, many state polls) coupled with micro-targeting of voters, is the future of politics. Why? Because it allows campaigns to spend money incredibly efficiently, for example by placing ads in TV shows that certain audiences are much more likely to watch.
But all this raises a troubling question – if politics gets overrun by all this highly statistical analysis, will it start to look like drone warfare, where what you see outside is merely a shadow, while what is really happening is going on deep inside some control room where a bunch of geeks are analyzing and manipulating data with the latest technology?
So statistical-based targeting may very well be the future not just of American politics, but campaigns elsewhere – one thing we know about U.S. politics is that it is often a trendsetter for the rest of the world. Don’t be surprised if at Britain’s next general election you start seeing American-style, statistical drone warfare.