New York University’s Sanford Gordon has devised a new method for forecasting the 2010 congressional elections that calibrates current expert rankings of individual House races based on the predictive success of those rankings in previous election cycles.
New York University’s Sanford Gordon has devised a new method for forecasting the 2010 congressional elections that calibrates current expert rankings of individual House races based on the predictive success of those rankings in previous election cycles.
Gordon, an associate professor in NYU’s Wilf Family Department of Politics, uses publicly available race rankings determined by the Cook Political Report and the Rothenberg Political Report (e.g., “likely Republican,” “leaning Democratic,” “toss-up”). However, his model also considers the rankings from previous election cycles and their subsequent electoral outcomes.
“The model is a systematic way to aggregate expert rankings to tell us about what is likely to happen in November,” Gordon explained.
“The basic idea is to treat the outcome of each current House race as the result of a weighted coin-flip,” he added. “So, for example, suppose that in the past, Democrats have won seats classified as ‘leaning Democratic’ 93 percent of the time. Then our best estimate of the chance that the Democrat will win in a ‘leaning Democratic’ district in the current race should be in the neighborhood of 93 percent.”
Gordon ran 10,000 simulations using rankings from the Cook and Rothenberg political reports (as of late September) to come up with which hypothetical election outcomes are more or less likely. The full details of the procedures are available here.
Using the Cook rankings, the procedure projects an expected Republican caucus of 215 seats; using the Rothenberg rankings, the prediction is 213 seats. Two hundred and eighteen seats are necessary for a majority. Gordon also ran simulations to calculate the likelihood of the House changing hands—that is, what are the chances that the Republicans will win at least 218 seats?
The simulation using the Cook rankings suggests that the probability is about 37 percent; using the Rothenberg rankings, it is 14 percent.