The world of sports is in the latter stages of a revolution of data-driven analysis. The ultimate question in sports analysis is, what determines whether a team wins or loses? Traditionally, most people believed that the determining factors were those that sports journalists liked to write about and that coaches felt like they could control.
Back in the day, everyone talked about leadership, and chemistry, and clutch hitting, and all sorts of other human factors that were unquantifiable. The numbers show that the truth is more boring than that — once you learn how to properly measure player performance, statistics show that teams with better players usually win. When teams underperform or outperform their statistical predictions, more often than not it’s due to luck.
A classic example of received wisdom from the institutions of sport is that great teams win most close games. What data analysis teaches us is that performance in close games is essentially random, and that great teams don’t play in as many close games — they tend to beat inferior teams badly. That’s what makes them great.
The new insights that quantitative analysis has brought to sports are applicable to many fields, including politics. That is perhaps best illustrated by the fact that Nate Silver, once known to baseball fans as creator of the PECOTA system for forecasting the performance of baseball players, has become famous as a political analyst, specializing in breaking down polling data and using it to make predictions.
The adjustment to a more data-driven approach to analyzing sports has been tough on sports journalists, because increasingly they are being forced to acknowledge that their impressionistic analysis of sports may be entertaining, but isn’t very informative. Political journalists are just starting to face the same challenge to their positions of authority. As political scientists have improved their data-driven analysis, it has become increasingly clear that most of the things that political writers (and indeed, politicians) think are are important, really aren’t.
This was all a very long segue into a link to an article about the tension between political scientists and political journalists in the Columbia Journalism Review, Embrace the Wonk. Here’s the crux:
That perspective differs from the standard journalistic point of view in emphasizing structural, rather than personality-based, explanations for political outcomes. The rise of partisan polarization in Congress is often explained, in the press, as a consequence of a decline in civility. But there are reasons for it—such as the increasing ideological coherence of the two parties, and procedural changes that create new incentives to band together—that have nothing to do with manners. Or consider the president. In press accounts, he comes across as alternately a tragic or a heroic figure, his stock fluctuating almost daily depending on his ability to “connect” with voters. But political-science research, while not questioning that a president’s effectiveness matters, suggests that the occupant of the Oval Office is, in many ways, a prisoner of circumstance. His approval ratings—and re-election prospects—rise and fall with the economy. His agenda lives or dies on Capitol Hill. And his ability to move Congress, or the public, with a good speech or a savvy messaging strategy is, while not nonexistent, sharply constrained.
The important take away for people who are consumers of political news is that press coverage of politics on a day to day basis is at best useless and at worst pernicious. What really matters is that there is a blowout spewing millions of gallons of oil into the Gulf of Mexico and that it probably won’t be fixed until August. The degree to which President Obama seems to be upset about it doesn’t matter. What really matters is that the unemployment rate is right around 10%. What doesn’t matter is that Congress is holding votes on Don’t Ask Don’t Tell rather than doing things that create the appearance that they’re working on the economy.
I try to pay attention to what our actual problems are and the degree to which we’re making progress in solving them. It turns out that not only is this what’s important in our day to day lives, but it’s what’s important politically as well.
June 3, 2010 at 9:24 am
Your overall point is well taken, but political science has been a mature field of study for far longer than sabermetrics. What makes Silver’s work so important is that it is accessible to laypeople: most journalists seem to believe that poll results are brought down from Mount Sinai, but he takes the reader behind the numbers and explains why they may or may not mean something. I think there is a case to be made for FiveThirtyEight as the most important contribution to journalism made by the blogging revolution. He is not merely breaking stories– he is creating (or has created) a methodology.
June 3, 2010 at 9:39 am
Political science has been a mature field for a long time, but I don’t think it was as quantitive 20 years ago as it was today. I seem to remember most political science classes almost being literature classes. You read the political philosophies of the greats and talked about them. These days, a lot of it is about breaking down data and finding patterns and relationships.
I do agree with you on 538, though. Both Sabermetrics and 538 were about revealing to laymen who are not statisticians what data can teach us about trends and performance. What 538 really blew up was the idea that each news outlet could sponsor a poll and then just report on the results of the poll they commissioned, or that you could compare the results of last week’s Gallup Poll to this week’s Rasmussen poll and infer something from what you see.