While the 2012 House Swingometer may have problems due to redistricting making it impossible to do a perfect seat-for-seat swing, I’m going to try using it anyway to see what it says.
We have two generic ballot polls from last month. Let’s see what they might predict for the House in 2012.
As with last time, the Swingometer focuses on the two-party vote, subtracting third parties and other distractions. In 2010, the raw vote went 51.38% for Republicans, and 44.77% for Democrats, per Wikipedia. For a two party split that comes out as 53.4 R-46.6 D, for a Republicans +6.8 result. The difference from that edge is the ‘swing’ from year to year that I use.
We have few polls right now, so I’m not going to average them all as I did in the runup to the 2010 elections. For now I’ll look at each poll’s projection separately. I report, you decide.
First up is the Rasmussen Reports generic ballot polling 3,500 Likely Voters by telephone (no mention given of mobile phone handling), MoE 2. The result is R 43-D 39. Taking that as a two party split, we have 52.4-47.6, R+4.8. This represents a two point swing toward the Democrats for 2010, and Swingometer, which projects Democrats gaining 8 seats, giving John Boehner a reduced 234-201 majority.
Second we have Public Policy Polling’s generic ballot which, in a change from last November, has returned to polling Registered Voters: 928 RVs, MoE 3.2. In addition to the RV/LV issue, the survey notably has the generic ballot as the sixth question, the first two asking about Rupert Murdoch and John Boehner, but not about Barack Obama or Harry Reid.
Regardless, the result is D 46-R 44, or as a two party split, D 51.1-R 48.9, D+2.2. This poll represents a swing of 9 points to the Democrats. Swingometer says a 9 point swing would win 35 seats for the Democrats, giving Nancy Pelosi a 228-207 majority.
If I did average these as I did in 2010, the LV poll would get double weight, and the weighted average swing would be D+4.3, winning the Democrats 15 seats, and John Boehner a small 227-208 majority.