I have updated the graph from my “earlier post”:https://themonkeycage.org/2010/09/political_science_forecasts_fo.html to include forecasts from “538”:http://fivethirtyeight.blogs.nytimes.com/2010/09/10/g-o-p-has-2-in-3-chance-of-taking-house-model-forecasts/ (specifically, they estimate that the Democrats will retain 210 seats) and Stochastic Democracy (specifically, the histogram “here”:http://stochasticdemocracy.blogspot.com/2010/09/new-projections-republicans-favored-to.html, which states that the GOP should win on average 226 seats).
Those forecasts that rely on polling data — e.g., the generic ballot — could change in the months ahead if the polls change. However, I am _not_ going to be updating this graph. I want to put the models that rely on such data on equal footing with those that do not. We can thus think of all of these forecasts as made at roughly the same time.
I want to emphasize something that Andy said in his “post”:https://themonkeycage.org/2010/09/comparison_of_forecasts_for_th.html. All of these forecasts have considerable uncertainty, so the forecasts that have the Democrats losing 40-50 seats are essentially the same forecast.