This is the first Open Left House forecast for 2010. I currently estimate a Republican net gain of 17 seats, for a Democratic majority of 240-195--Chris
Over the past month, nine polling firms have published surveys on the national House ballot. Looking only at the most recent poll from each of those nine pollsters, the results show Democrats maintaining a decent advantage.
A Democratic advantage of 41.2%-37.7% is a far cry for the supposed Republican wave we keep hearing about. Democrats are, after all, still winning.
These numbers are placed in more context--context which, I might ad, is something that blogs do a far, far better job of providing than any other medium when it comes to electoral forecasting--below the fold. The bottom line this, my first crude House forecast for 2010, is that Republicans are not currently poised to retake the House. Currently, I project a Republican gain of 17 seats, for a partisan balance of 240-195 in favor of Democrats.
Generic ballot numbers in context, and how they lead to a first crude House forecast:
Pollster.com shows Democrats ahead by 1.3%, Real Clear Politics by 3.7%. Pollster.com has slightly more favorable numbers for Republicans than mine, showing the Democratic advantage at only 1.3%. However, due to the nature of their methodology, this far out from an election they will place more weight on those polling firms that release more polls: Rasmussen, Democracy Corps, and Polimetrix. The strong weight toward Rasmussen in particular presents a more favorable picture for Republicans than would be the case if more polling firms were releasing polls.
Generic Ballot reasonably predicts final national vote totals Over the last six House elections (1998-2008), there has been a mean error of 1.8% from the final polling average in the generic congressional ballot to the final House national vote. That really is about as accurate as can be expected from polling and electoral forecasting. I would feel better with an error rate of 1.4%, but 1.8% isn't far off from that ideal.
Republicans do not tend to do better from final polls to final results. Over the last six House elections, Republicans improved on their margin from the final polls to the final results by a mean of 1.6% and a median of 1.5%. However, rather than seeing some sort of great conspiracy or excellent Republican ground game in this, it is likely that this is a statistical quirk due to a small sample size of only six elections.
Research I conducted on the 2004 and 2006 elections showed Democrats gaining an average of 0.5% from the final polls to the final results across 42 statewide elections. In a preliminary study I have conducted of the 33 closest statewide elections in 2008 (Presidential and Senate), Democrats gained a mean of 0.40% from the final polls to the final result. I will complete and publish this study later in the week.
The 75 data points showing Democrats improving from the final polls to the final results by 0.4-0.5% easily outweigh 6 six House-specific data points showing Republicans improving by 1.5%-1.6%. Due to the nature of polling, the type of elections that are being polled do not matter.
Republicans far off from the Democratic position in 2005. In the fall of 2005, there was not a single congressional generic ballot where Democrats led by less than 5%, and the average Democratic advantage hovered around 8-9%. Given that they still trail by 3.5% in my numbers, 3.7% according to Real Clear Politics and 1.3% according to Pollster.com, Republicans are at least 9%, and as much as 12%, behind where Democrats were at this same point in 2005.
Not enough seats for Republicans: Let's say that Democrats win the national House popular vote by 2.8%, which is the average forecast of Real Clear Politics, Pollster.com and myself. This would represent a 6.1% improvement for Republicans from 2008. If the margin in all House campaigns from 2008-2009 shifted 6.1% in favor of Republicans, they would net a total 18 seats (17 from 2008, plus NY-20), and a partisan make-up of 239-196 in favor of Democrats.
As such, in order to even come close to taking back the House, Republicans need a much larger shift than 6.1%, and / or for that shift to be disproportionate to swing districts. Right now, there is no convincing evidence that either the Republican position will continue to improve nationally, or that the gains they have made are concentrated in swing districts. This is not to say this won't happen-just that there is no evidence either way.
Overall, these numbers lead me to a forecast of a 17-seat Republican gain. This comes from the 18 seats in the final bullet point, minus the Louisiana 2nd, which at a PVI of D+25 is a virtual lock to switch back to Democrats (it only flipped in 2008 due to severe corruption from former Representative William Jefferson).
In summary, the current hysteria about Republicans poised to take back the House is simply not borne out by the numbers. Still, due to the lack of district-level detail at this time, there is a wide range of error in this prediction-it could be anywhere from a 10-25 Republican net gain, according to current numbers. As we get closer to the election, eventually I will abandon use of the generic Congressional ballot entirely, and produce the forecast only on a seat-by-seat basis.