Generally speaking, the RAND poll seems to have settled down into a relatively stable 49/45 Obama lead, now that we have passed the post-debate volatility. Very little movement the past four days. But as those who follow the poll know, they also show various breakouts of their sample, and today's break out was on three "battleground" states. EDIT - Note the state level results are Predictions on how residents of the states see the race, NOT HOW THEY WILL VOTE. Now keeping in mind that their state populations must be awfully small, there finding are interesting. First, PA, which most no longer think is battleground state has a twenty point plus Obama advantage. Second, Ohio voters, predict O win. But Florida is the surprise - with the least fluctuation, consistently keepin a 10+ point Obama prediction margin. Several thought below.
1. First, if you go to the Link you will note rather wide error band/confidence margins, befitting small samples, but FL and PA are both outside the band, suggesting that at least from a statistical view, they measure a real separation in prediction of outcome
2. FL's stability suggests that issues other than debate performance are important to those voters.
3. The OH margin moved from comfortably outside the MOE to clearly within the MOE, so with perceptions changing so much, upcoming OH polls should be quite instructive
4. Also note that the rate of switching from Obama to Romney and vice versa has just about equalled out, after BIG Obama tilts prior to debate and smaller and shorter lived tilt to Romey after the debates. This again suggests returning stability to the race.
5. Finally, the current 4 point forecast margin is consistent with Gallup - although today (Tuesday) we get their shift to likely voter model. Note that RAND deals with the issue of likely voter by their methodology of asking two questions 1- how likely are you to vote for O/R, and 2- How likely are you to vote. So somebody that says they are 75% lely to vote for Obsma and 75% likely to vote is counted as .56 of a vote for Obama and a .19 vote for Romney. I don't think this methodology has really been tested out, but on the face of it should balance out in the end I suppose