An avid lover of baseball and statistics, Nate Silver began his journey into political forecasting right before the 2008 presidential election. In 2008 his blog, FiveThirtyEight, Silver predicted 49 out of 50 correct states. However, for this election Silver predicted all 50 states correctly.
Silver’s model combines Bayesian probability with polling data. For senate races, FiveThirtyEight went through seven stages to predict winners. Stages include:
Weighted polling average – polls are weighted on recency, sample size, and pollster rating
Adjusted polling average – subject to three additional types of adjustments. The trendline adjustment (detailed evaluation of trends within generic congressional ballot polling), the house effects adjustment (Sometimes, polls from a particular polling firm tend consistently to be more favorable toward one or the other political party), and the likely voter adjustment (polls may be conducted among a variety of population samples).
FiveThirtyEight regression – polling is supplemented by other information suc as incumbency status, Campaign contributions, etc.
FiveThirtyEight snapshot – the adjusted polling average and the regression are combined into a ‘snapshot’ that provides the most comprehensive evaluation of the candidates’ electoral standing at the present time.
Election day projection – represents the most accurate forecast of where candidates will finish on Election Day
Error analysis – determines the degree of uncertainty intrinsic to the forecast
Simulation – provides a probabilistic assessment of the outcome of any one given race