Predicting and Understanding the 2012 Election

Опубликовано 28 сентября 2012, 15:54

For more than 75 years, elections forecasting has been static—ask a random sample of a representative group of voters who they would vote for if the election were held today, and then report the poll result. In this talk, David Rothschild first demonstrates that the same samples could be addressed with other questions to produce a more accurate standard forecast (such as binary winner or expected vote share). Second, he challenges the standard forecast, stating that what most stakeholders really want and need are more innovative forecasts, like probability of victory or even probability distributions. Third, David shows how both standard and innovative forecasts can be made more efficient with new methods that utilize more cost-effective, non-representative samples and, in time, passively generated social media data. Fourth, he shows how Microsoft is going to be a leader in this new innovation. Finally, David tells you know who is going to win the election!