« Bad Gallup! No Biscuit! | Main | No Further Comment Necessary Department »

Kerry Leads, Gains in Key Swing States After 3rd Debate

John Kerry leads George Bush among LV's in: in FL +1; PA +6; OR +9 and WA +7. He trails Bush in: AR -5 and NC -3 (stat. tie), according to SurveyUSA Polls conducted 10/15-17.

Comments

I have to say, though, that I don't place much trust in SUSA polls. These were the same people, after all, who called Maryland and New Jersey close in 2000, where Gore then went on to win by 15 points.

Although this isn't the only poll I've seen with Kerry leading in Florida, so I'm inclined to believe he's genuinely ahead.

SUSA has the best record of any polling firm over the past 10 years.

I think the NC poll is the most impressive. We'll see how is really turns out though.

In another thread I noted that it is probably appropriate for pollsters to assume a higher Republican turnout (as a percentage of registered Republicans) than Democratic turnout (as a percentage of registered Democrats). I was surprised that a few posters chose to attack this observation (some quite forcefully). This I believe is a well-documented phenomenon of American politics. It stems from the fact that Republicans tend to be more affluent, more focused on politics, have better access to the polls, etc. Some asked for statistics backing this up, so I went and looked up turnout statistics for some California counties. Obviously this is not a rigorous analysis, but what I found does indicate that Republicans turnout in greater percentages than Democrats. And I've read in many places that this is true nationally - I'll provide cites as they turn up. Some people on this blog just don't like to hear this apparently, and seem to suggest as a point of contention that I'm saying Republicans turn out in greater total numbers - I'm not. The advantage in Democratic party id partially or totally offsets the higher Republican turnout percentage to some extent. As an example, here's results from Contra Costa county, which is heavily democratic and went for Al Gore in 2000:

Registration & Turnout
Democratic
Completed Precincts: 907 of 907
Reg/Turnout Percentage
Democratic Registration 223,155
PRECINCT REGISTRATION 223,155
PRECINCT BALLOTS CAST 99,416 44.6%
ABSENTEE BALLOTS CAST 37,480 16.8%
TOTAL BALLOTS CAST 136,896 61.3%

Registration & Turnout
Republican
Completed Precincts: 907 of 907
Reg/Turnout Percentage
Republican Registration 153,790
PRECINCT REGISTRATION 153,790
PRECINCT BALLOTS CAST 71,717 46.6%
ABSENTEE BALLOTS CAST 32,706 21.3%
TOTAL BALLOTS CAST 104,423 67.9%

Notice that Gore carried the county by ~31,000 votes, despite a registration advantage of 70,000. The reason for the narrower gap was the much higher Republican percentage turnout (68% R vs 61% D). If this is indeed indicative of national turnout habits, and I believe it is, then it is appropriate for Gallup to weight registered Democrats for 8% less turnout than registered Republicans. Alan asserts this is "totally unrealistic" - I disagree.

In my opinion, the big flaw in the Gallup polls is not the turnout weighting, but rather their insistence on oversampling registered Republicans. There just doesn't seem to be a good reason for this. Their assertion that party id is fungible isn't borne out by exit polls from the past several presidential elections. To gain an accurate sampling of the electorate, they should weight for best-known state or national party id distribution, as Zogby does, and also for expected party percentage turnout.

I still say that the Kerry campaign must stop Dukakissing on the Bai article like they did on the flipflop spin. My guess is that, with the same performance in the debates, if he had prodded the press on the flipflop issue along the lines of Chait's article in the 10/18 New Republic SIX MONTHS ago, and dismissed it as a mere spin in the ways I've suggested since, and put forward an NYU+ length speech to counter Bush's monday blowout in NJ on terror, and challenged the Bai spin, and the media that have magnified its already distorted portrayal, he would be running consistently ahead by 4-5 points in the polls, and pulling along Democrats into the House and Senate too. In an honest election, where the Clarke interview with Michael Moore in Fahrenheit 9/11, where he points out 9/12 Bush was only interested in Iraq ACROSS THE COUNTRY -- the Republicans would be in trouble in both Houses of Congress and would have a 92 type futile race for the White House. But that's not the agenda.

I just read that Zogby went out on a limb and predicted that Kerry was going to be the 44th president. He said that there is more to polling than statistics:

'Polling can be uncertain only if you rely on statistics alone,' is Mr Zogby's riposte. 'That's why an effective pollster has to rely on culture, history and sociology. I repeat, polling is the study of human behaviour, not simply a sampling of people's preferences.'

That may well explain his success. His big test will come on Nov 2. In Singapore last Friday, he flatly predicted that Mr Kerry would become the 44th President of the United States.


I sure hope he's right. It's comforting to know that he was one of the 2 polls to get it right in 2000.

Hg

Anyone have a quote for Zogby calling a Kerry win ?

No to mention that the Newark Star Ledger has Kerry up by +13 in God's Country, and the CBS poll, which has the horserace tied, shows the prez approval at only 44%. Surely, if the latter is close to being right, Bush is in big trouble.

Could Ruy or someone else on the site comment on the WaPo and Rasmussen aggregate "swing state" poll numbers? These have been looking good, but I've never seen anything that discusses whether an aggregate result among a group of swing states really tells you anything useful about the EV outcome for those states would be.

Since this is a thread on state data, I thought it would be helpful to have some pre/post debate apple-to-apple comparisons of state polling data. So...I went to race2004.net for a list of state polling data. Before examining the data, I selected surveys dated their from 9/15 to 9/30 as pre-debate, surveys dated 10/10 or later as post-debate. I found 36 polls across 22 states that were represented in both time periods. In 25 of those polls, Kerry gained. Another 6 showed no change. Another 5 showed Bush gaining. The median change was a 2% Kerry gain. The average change was 2.44% toward Kerry. I don't know that this tells us where things stand in the electoral college right now, but it sure does suggest that Kerry gained some ground with the debates. Here's the list, from largest Kerry gain to largest Bush gain:

Ohio, Ohio Poll, Kerry +11
Oregon, SUSA, Kerry +10
Iowa, Rasmussen, Kerry +7
South Carolina, SUSA, Kerry +7
New York, SUSA, Kerry +7
New Jersey, Rasmussen, Kerry +6
New Jersey, Strategic Vision (R), Kerry +5
Georgia, SUSA, Kerry +5
Michigan, Rasmussen, Kerry +4
Arkansas, SUSA, Kerry +4
Ohio, Strategic Vision (R), Kerry +4
Pennsylvania, Rasmussen, Kerry +4
New Jersey, Quinnipiac, Kerry +4
New Hampshire, Research 2000, Kerry +4
Alabama, Rasmussen, Kerry +3
Iowa, Strategic Vision (R), Kerry +2
Wisconsin, Strategic Vision (R), Kerry +2
Michigan, Strategic Vision (R), Kerry +2
Pennsylvania, Strategic Vision (R), Kerry +2
Rhode Island, SUSA, Kerry +2
Illinois, Rasmussen, Kerry +2
South Dakota, Rasmusen, Kerry +2
Washington, SUSA, Kerry +2
Minnesota, Strategic Vision (R), Kerry +1
Arkansas, Rasmussen, Kerry +1
Minnesota, Rasmussen, no change
Wisconsin, Rasmussen, no change
Ohio, Rasmussen, no change
Florida, Rasmussen, no change
Pennsylvania, Quinipiac, no change
New Jersey, FD Univ, no change
Oregon, Research 2000, Bush +1
Arkansas, Zogby, Bush +2
Texas, SUSA, Bush +2
Oklahoma, Wilson Res (R), Bush +5
California, Rasmussen, Bush +5

The MN numbers sure sound like they match on the Lawn sign war!