Combining Forecasts : An Application to Elections

We summarize the literature on the effectiveness of combining forecasts by assessing the conditions under which combining is most valuable. Using data on the six U.S. Presidential elections from 1992 through 2012, we then report the reduction in error obtained by averaging forecasts within and across four election forecasting methods: poll projections, expert judgment, quantitative models, and the Iowa Electronic Markets. Across the six elections, the resulting combined forecasts were on average more accurate than each of the component methods. The gains in accuracy from combining increased with the number of forecasts used, especially when these forecasts were based on different methods and different data, and in situations involving high uncertainty. Combining yielded error reductions ranging from 16% to 59%, compared to the average errors of the individual forecasts. This improvement is substantially greater than the 12% reduction in error that had been previously reported for combining forecasts