Accurate University Study Predicts Romney Win
- Thread starter pspr
- Start date
I am working on this thesis with respect to this poll and many others.... that just because there may have been more Rs... does not mean the Rs were over sampled. Some polls adjust their poll results to their anticipated turnout.
Unless the poll tells you the ratios in the sample ... we really have no idea of how they tallied their headline numbers.
I agree with you that the Romney lead amongst indies is a big deal, but equally as important are women. That is the whole drive behind war on women campaigning. Obama carried women (the largest voting group) by 13 points in '08, while he only tied with men. In the pew poll, Romney leads by 8 with men and is tied with likely women voters. If Romney gets within 2-points of Obama with women, Obama doesn't have a chance.
But it only matters in OH, FL, and VA. That is the only realistic path to victory for Romney. He needs all three of those and then one of either IA, CO, or NV.
Specifically, this poll gave a lot of detail and provided unweighted numbers. I also think it is fairly safe to assume that the actual voter turnout will be equal at best from a GOP viewpoint. In 2004 it was even and in 2008 Dem turnout was +7%. Even under the current circumstances is it very unlikely that the turnout will swing completely the other direction. IMO, it will still be Dem +2-3%, but I think that indies will be close to 32% this year given that almost every poll for the last two years has shown a decrease in party affiliation on both sides with the number of claimed independents rising.
The CU professor who created the model was just on Fox. He notes that the model prediction in August was about the same predicting Romney and that the big jump from the debate for Romney is just things coming back in line to reflect the micro economics across the courntry.
As we get closer to the election he expects that the polling will continue to work it's way closer to a Romney win.
As we get closer to the election he expects that the polling will continue to work it's way closer to a Romney win.
1. I agree the poll gave the unweighted numbers... but did the poll explain if it use the unweighted numbers of likely voters to tally up Romney at plus 4.
Romney could have been plus 6 unweighted if the poll had more Rs than Ds... and then they rolled it back to match up with say your predicted sample of D plus 2-3.
2. Because Ds have lost far more voters than Rs to the the Is since 2008.
and because most polls are showing Rs are more motivated...
I think the Ds will be at the very best slightly positive but it is far more like the Rs will turn out in greater number than the Ds.
There are a lot of union guys who will switch, there are far fewer Jews who will vote for Obama (15%) there are far less motivated young people. There black pastors who are telling their flock to leave and I suspect there are a bunch of ticked off Catholics..
Romney could have been plus 6 unweighted if the poll had more Rs than Ds... and then they rolled it back to match up with say your predicted sample of D plus 2-3.
2. Because Ds have lost far more voters than Rs to the the Is since 2008.
and because most polls are showing Rs are more motivated...
I think the Ds will be at the very best slightly positive but it is far more like the Rs will turn out in greater number than the Ds.
There are a lot of union guys who will switch, there are far fewer Jews who will vote for Obama (15%) there are far less motivated young people. There black pastors who are telling their flock to leave and I suspect there are a bunch of ticked off Catholics..
1- I see your point. But the verbiage and numbers suggest that any such weighting adjustments would've been minor.
2- Several of those might be the case, but without studies it is speculation. Anyway, that doesn't change their party ID, it only changes their vote. I believe that there are many Ds who will vote R or stay home, but they are still Ds in the polls. So the idea that suddenly there will be more Rs than Ds is unrealistic.
