Originally Posted by Dame_Enda

RCP have now moved Ohio from Toss-up to Leans Obama in their electoral map. Obama now leads 265-191 excluding DK states (100 EVS and generally ones where the gap is less than 5%). The average Obama lead in Ohio is 5.2%.

That's due to blatantly skewed polls which I wrote an extensive post on yesterday, from which I'll quote the key figures:

Obama won Ohio in 2008 by 5%. The D/R/I turnout in the wave year for Democrats in 2008 was 39/31/30 (D+8). The 2010 turnout which was a wave election for Republicans had a D/R/I of 36/37/28 (R+1).

The latest CBS-NYT -Quinnipac giving Obama a 10 point lead that includes 26% Republicans - turnout worse than 2008. Specifically, the D/R/I breakdown is 36/26/33 (and 4 other) giving Democrats a turnout advantage of 10% - not a hope that will happen.

Their numbers for Florida and Pennsylvania are also skewed for the same reason and I may come back to that later with more exact figures for D/R/I breakdowns in those states in 2008 and 2010.

Originally Posted by borntorum

As an aside, wasn't "A Balanced Government - It's Only Fair" the worst political slogan of all time? Anywhere?

Just barely beaten into second place by Charles Taylor's 1997 Classic: "He killed my ma, he killed my pa, but I will vote for him."

Third place goes to "Gilmore for Taoiseach" so while they didn't get the overall prize, two out of three ain't bad.

Originally Posted by livingstone

There are a few problems with this analysis:
1) It is logically inconsistent to say that the reason the WP has Obama +8 is that it oversamples Democrats (+7) when Gravis has Obama at +1 by oversampling Democrats more (+10). Logically, if the reason that WP had Obama +8 was oversampling Democrats, then a poll which oversampled Democrats even more should give him an even stronger lead.

It would be statistically inconsistent if the polling methodologies were exactly the same which they obviously are not.

Originally Posted by livingstone

2) It is a huge leap to suggest that Democrats will go from 39/31 to 36/36. To suggest that Dem turnout will be the same as in 2010 when they didn’t have a still popular President to vote for is bizarre. The margin may change, but Dem turnout will be ahead of Republican turnout..

I never said Dem turnout would be like 2010. In fact I wrote: "...splitting the 2008 and 2010 turnout would give Democrats a 3 to 4% turnout advantage." (at most imo and that's being optimistic for Dem turnout).

Originally Posted by livingstone

3) Claiming Rasmussen as being most accurate without the figures is, frankly, bizarre. You’re basing your faith in them on their blurb saying how great a polling company they are. Ask any polling company ‘do you weight polls accurately’ and every single one will tell you how much effort they put into being accurate. Believing Rasmussen’s weighting is more accurate because they say they put effort in is weak.
4) Also on Rasmussen, you consistently ignore the role that their methodology has in their results. Excluding those without landlines simply cannot give an accurate result, or at the very least you certainly can’t just assume an accuracy out of thin air.
I think you’re right that Obama’s not up by 8 in Ohio. But there is substantial evidence to suggest that he is up by more than 1. I think Purple may be most accurate here – I reckon a four to five point advantage is closer the mark.

Regarding Rasmussen, they have proved themselves to be reliable polls - coming joint first in a study analysing poll accuracy in the 2008 Presidential election.

Originally Posted by Skyrocket

It would be statistically inconsistent if the polling methodologies were exactly the same which they obviously are not.

No, it’s still pretty inconsistent to compare polls, claiming that an oversampling of Democrats overstates Obama’s support while pointing to a poll with a higher sample of Democrats and a lower support base for Obama.

I never said Dem turnout would be like 2010. In fact I wrote: "...splitting the 2008 and 2010 turnout would give Democrats a 3 to 4% turnout advantage." (at most imo and that's being optimistic for Dem turnout).

But you have no basis for that assumption. Any weighting for party affiliation is guesswork, which is one of the reasons most polling companies don’t do it.

The other issue with party affiliation is that despite your claims that GOP turnour will be up on 2008 and Dem turnout will be down on 2008, why then is a poll of likely voters consistently showing Democrats leading in likely voter stakes?

This is who the pollsters are finding, and they are repeatedly finding more Democrats likely to vote than Republicans. One of the problems with weighting for party affiliation is that the question is asked at the end of the poll, so there are some suggestions that respondents give the same affiliation as the candidate they support so they don’t look stupid or inconsistent, so a possible reason that more people are identifying as Democrats in polls because more people are supporting Obama and so give the answer they perceive to be consistent with that.

Similarly the reliability of exit poll data on this is limited for similar reasons. People coming out of a polling station who just said they voted for John McCain are probably less likely to say they are Democrats, even if they voted for Clinton, Gore, Kerry etc, because they consider that they will look stupid or inconsistent.

Ultimately, non-weighting for party affiliation is not a new phenomenon. In 2004, Democrats were arguing that the polls were biased and oversampled Republicans – actually the polls were just finding more respondents willing to identify as Republican because they were backing Bush.

Regarding Rasmussen, they have proved themselves to be reliable polls - coming joint first in a study analysing poll accuracy in the 2008 Presidential election.

Well, firstly, they came first (along with Pew) in the preliminary results based on an assumption that Obama would win by six points. He won by more and in the final version of the comparison, Rasmussen were joint ninth.

But I’ll give you that they came pretty close to calling 2008 broadly right in their last pre-election poll. But they were joint first with, as I’ve said, Pew. Pew has turned out to be one of the most favourable pollsters to Obama and Rasmussen have turned out to be the most favourable to Romney. So while they were on the same page in 2008, we know one of them must be wrong now.

So when you say ‘I believe Rasmussen because they were first in 2008’, it begs the question – if 2008 accuracy were your benchmark for current accuracy, why not believe Pew are accurate? Is it because that suits the angle you’re coming from? That’s fine if so, but you’ve not given any reason to suggest that Rasmussen should be trusted now compared to Pew.

Now I suspect the truth is somewhere in the middle. Pew seems a bit high for Obama for me, but I think that they are at least on the right side of it. Rasmussen had an awful year in 2010 and they’ve not had a chance to redeem themselves in a national election since. So that’s the first reason to suspect that they may not be as accurate as in 2008. Secondly, Rasmussen is out of kilter with almost every other polling company – they repeatedly have Romney doing better than other companies like Democracy Corps, Fox, Opinion Research (all of which came higher than Rasmussen in the final 2008 accuracy comparison).

So if you have a polling company that was accurate four years ago, but which has since had a terrible election cycle of polling, and which is now out of step with most other pollsters (including ones that were more accurate in 2008), it seems illogical to put blind faith in that polling company and dismiss all others. Maybe its not blind, maybe you know something that I don’t, and that Fox don’t, and that Democracy Corps don’t, and that PPP don’t, and that CNN and Opinion Research don’t. If so, feel free to share.

Just looking at the polls on this day in September 2004. Bush was ahead on average 5.5% nationally. These were the state average polls for the day (slightly complicated by Nader's presence which probably increased Bush's margins).
Bush Total = 291 Toss Up = 26 Kerry Total = 221

--------------------------------------------------------------------------------

Battleground State Polls - RCP Average
Poll | Date Bush Kerry Nader Spread
Florida | 9/17-9/22 47.7 44.3 3.0 Bush +3.4
Ohio | 9/10-9/22 49.5 43.7 1.8 Bush +5.8
Pennsylvania | 9/14-9/24 45.5 47.3 1.0 Kerry +1.8
Michigan | 9/14 - 9/22 42.3 47.3 1.5 Kerry +5.0
Missouri | 9/13 - 9/19 49.0 42.3 1 Bush +6.7
Wisconsin | 9/12 - 9/21 49.3 42.8 1.75 Bush +6.5
Minnesota | 9/10 - 9/24 45.5 45.5 2.7 TIE
Iowa | 9/16-9/23 48.6 45.0 1.7 Bush +3.6
New Hampshire | 9/13 - 9/23 46.8 45.5 2.0 Bush +1.3
Nevada | 9/13 - 9/23 49.7 44.3 1.0 Bush +5.4
West Virginia | 9/13-9/20 48.0 44.8 2.0 Bush +3.2
New Mexico | 9/14- 9/16 45.5 46.0 1.5 Kerry +0.5
Oregon | 9/13-9/21 46.3 47.0 1.0 Kerry +0.7
Colorado | 9/14-9/23 49.3 42.3 3.0 Bush +7.0
Maine | 9/10 - 9/23 44.5 45.5 3.0 Kerry +1.0
Washington | 9/17-9/21 42.0 51.3 2.0 Kerry +9.3
New Jersey | 9/12- 9/19 46.3 47.7 1.5 Kerry +1.4

The ultimate outcome was Bush 286, Kerry 252. The only changes were that Wisconsin, New Hampshire and Minnesota went Kerry, while New Mexico went Bush; Virginia wasn't even included as it was seen as solid Republican and indeed it was until 2006. Despite the perception that it was a very tight election (which it was), the result was in fact already decided in late September and the debates and Osama's last-minure intervnetion etc made little difference. Unless there is something drastically different about the polls this time, I think we would need to see a major game-changer to take the election from Obama now. Leave aside that Kerry for all his faults was probably a less repellant (to the other side) candidate!

Obama now +6 in Gallup. This is very interesting as Gallup tends to, along with Rasmussen, be quite favourable to Romney. Also interesting is that he went from +3 to +6 in a single day.

Aside from Rasmussen, Romney hasn't led in a national poll in more than a month. But being down by 6 in Gallup three weeks after the end of the DNC will, I suspect, inspire further panic in Romney's camp.

Originally Posted by livingstone

Obama now +6 in Gallup. This is very interesting as Gallup tends to, along with Rasmussen, be quite favourable to Romney. Also interesting is that he went from +3 to +6 in a single day.

Aside from Rasmussen, Romney hasn't led in a national poll in more than a month. But being down by 6 in Gallup three weeks after the end of the DNC will, I suspect, inspire further panic in Romney's camp.

Huge jump.

Originally Posted by Des Quirell

Huge jump.

But worth hoping that some of the populus may start thinking instead of falling for the gutter "wallet gland" appeal

Originally Posted by Tea Shark

But worth hoping that some of the populus may start thinking instead of falling for the gutter "wallet gland" appeal

You never know, do you?

My frustration in this campaign is that we haven't seen much of Ryan. I suspect that Romney is keeping a lid on him for fear of being overshadowed.

Originally Posted by Skyrocket

It would be statistically inconsistent if the polling methodologies were exactly the same which they obviously are not.



I never said Dem turnout would be like 2010. In fact I wrote: "...splitting the 2008 and 2010 turnout would give Democrats a 3 to 4% turnout advantage." (at most imo and that's being optimistic for Dem turnout).



Regarding Rasmussen, they have proved themselves to be reliable polls - coming joint first in a study analysing poll accuracy in the 2008 Presidential election.

The problem with Rasmussen is that they ignore the 1/3rd of Americans who don't have landline and instead rely on cellphones. Fivethirtyeight.com pointed this out and that Obama does much better with polls that include cellphone users, such as Quinnipiac/CBS/NYT/NBC. Rasmussen also has a 38-37 split which may be exaggerating the GOP identification somewhat. Dick Morris on Fox recently said that in his polling, Obama had a 3% lead and that a number of Independents had switched to a Dem identification - and he's supporting Romney.