Math Nerds Rule (Editorial)

by Arkansas Business Editors  on Monday, Nov. 12, 2012 12:00 am  

Nate Silver, the nebbishy statistical genius who conquered poker, baseball and then politics, was the man of the hour on Tuesday. While the biggest names in conservative punditry — George Will, Peggy Noonan, Rush Limbaugh — were predicting a Romney victory to the end and the most familiar names in polling were still concerning themselves with national surveys, Silver’s proprietary formula for distilling meaning from thousands of data points predicted the winner in 49 states and the District of Columbia. And if President Obama maintains his paper-thin lead in Florida, Silver will be 51 for 51. (He even predicted the paper-thin lead, giving Obama a 50.3 percent chance of winning the Sunshine State.)

Taking the results of multiple “scientific” polls and coming up with an average is relatively simple. But Nate Silver looked back at the accuracy of the individual pollsters in previous elections and gave their findings the weight they deserved. He calculated the typical “bounce” from things like party convention speeches and candidate debates. He factored in economic data — unemployment, GDP, housing sales. He turned skeptics into believers and rendered his critics speechless. Nothing succeeds like clear, supportable, statistical success.

Consider similarly how accurate the predictions of “superstorm” Sandy turned out to be. As Silver himself notes in his new book, “The Signal and the Noise,” hurricane path predictions are three times more accurate now than they were 20 years ago. Haven’t we all come to trust the weather predictions more than we used to? More data, better predictions, less wasted effort.

Few businesses short of Wal-Mart have as many data points to help inform decision-making and planning as Nate Silver and meteorologists have to work with, so instinct and “gut” are still important — especially, it seems, in the softer science of choosing human capital. But the events of the last couple of weeks make us respect the gifted statisticians more than ever.



Please read our comments policy before commenting.