Democratic Data

datamining

One of the biggest stories of the just-passed presidential campaigns was a story about a statistician, of all occupations. Nate Silver of the FiveThirtyEight Blog at the New York Times does statistical analysis. Formerly of baseball statistical fame, Nate’s more recent career finds him analyzing an aggregate of political polls and forecasting probabilities based upon an algorithm that has been tested and proven remarkably accurate through multiple elections.

Nevertheless, approaching the 2012 presidential election, Mr. Silver and his methodology came under attack from supporters of the Mitt Romney Campaign because his analysis showed a high statistical probability that President Obama would be re-elected. Furthermore, by all evidence, Romney campaign officials and Republican pundits and pollsters refused to believe his forecasts, choosing instead to create their own polling models until they predicted the election of their candidate. As it turns out, the FiveThirtyEight Blog predictions were almost perfectly reliable.

Time.com blogger, Michael Scherer published a fascinating recount of how the Obama Campaign used modern consumer-oriented techniques of database management, data-mining, “consumer” profiling, and behavior modeling to turn out likely voters through highly dynamic voter turnout techniques, in real time. It worked almost exactly as predicted by the data and campaign gurus who developed the program and to devastating surprise to the Romney Campaign.

The thing is, data doesn’t care who you are, what you think or, especially, what you want to believe. It’s simply a tool. Data is useful to collect, process, analyze and utilize correctly to answer specific questions or solve specific problems. Those who fail to appreciate what it can do or understand what it says will simply be left behind by those who do.
blog comments powered by Disqus