Regardless of your ideology or political leanings, November 6, 2012, was a triumph. As the longest election campaign in America history wound down, a growing number of professional pundits were trying to undermine the polling data and the statistical analysis based on that data. This hostility was not based on any rational counterargument; rather, they took an ostrich-like approach to facts they didn’t like. Oddly, this anger and frustration coalesced around one brainy statistician, Nate Silver. And, despite their best efforts to discredit the polling process and Silver’s methodology, the polls were right the entire time.
Silver developed a model that aggregates and analyzes the results of state polling data and assigns them weight based on their historical accuracy. Based on this model, Silver was able to predict 49 out of 50 states in the 2008 election, 36 out of 37 governorships in 2010 and was within the confidence interval for the 2010 House races. When his model forecast a 2012 victory for Obama, Silver’s methodology came under intense scrutiny and attack.
While Silver’s formula is far too complex for me to wrap my head around, he insists that his predictions are only as accurate as the polling and survey data he receives. If the polling data would have been inaccurate, then the public trust in polling and statistical analysis would have been undermined. As it stands, a well-crafted survey with solid participation rates can still yield accurate, actionable results.
All is well with the world.
I have never doubted the efficacy of survey data and statistical analysis. However, I have a better understanding of the concept of GIGO — Garbage In, Garbage Out. Your survey data is only as good as the survey itself. The survey itself is only beneficial if it reaches your intended audience and convinces them to complete and submit it. Creating a well-designed, engaging poll or survey, while minimizing error, is fundamental to describing, understanding and predicting human behavior.
Not only did Nate Silver correctly predict the outcome of the 2012 presidential race, but he was also alarmingly accurate. While Silver is not the only statistician to utilize a highly accurate model, his emergence onto the pop cultural landscape should be celebrated.
Public confidence in a number of institutions is eroding. So, even if you don’t fully understand his formula or methodology, Silver’s reliance on accurate polling and survey data is reassuring.