« Granite State Gains | Main | Edwards, Cheney Split Debate Polls »

Summary of the Post-Debate Polls – Kerry Closes the Gap

Professor Alan Abramowitz of Emory University, one of the leading academic analysts of American politics. contributes this cogent summary of the post-debate polls.

An analysis of six national polls released since the first presidential debate last Thursday shows that John Kerry has made significant gains against President Bush, reducing the president's lead from an average of 6.6 points before the debate to only 1.8 points after the debate.

The six polls included in the analysis are:
CBS/New York Times
ABC/Washington Post
Pew Poll
Zogby Poll

In the six pre-debate polls, Bush's support averaged 49.3 percent compared with 42.7 percent for John Kerry and 2.0 percent for Ralph Nader. In the six post-debate polls, Bush's support averaged 47.8 percent compared with 46.0 percent for Kerry and 1.8 percent for Nader.

These results are based on likely voters when available. Otherwise they are based on registered voters.

Note: as a number of our sharp-eyed readers have pointed out, a typographical error combined the separate CBS/New York Times and Gallup polls on a single line, making it appear that there were only five polls rather then six. The error has been corrected.


Fyi, if anyone is interested in an "objective" analysis of recent polls, read this from Powerline. While Powerline is a Rep blogger, they are also realists.


Oct 5, 2004

More Poll Follies

The latest New York Times/CBS News poll came out this morning, and, like several others, it shows President Bush and John Kerry locked in a 47% to 47% tie, compared to an eight point lead for President Bush after the Republican convention in September.

The poll's internals are easily accessible--which is praiseworthy--so it takes only a moment to determine that the October poll sampled 34% Democrats and 29% Republicans, while the September poll sampled 33% Republicans and 31% Democrats. So it's hardly a surprise that Kerry did better in the October survey. If the pollsters sampled only Democrats, they could show that Kerry was sweeping toward an unprecedented victory.

We aren't going to be able to untangle the pros and cons of "correcting" samples between now and November; suffice it to say, however, that 4% of the population didn't abandon the Republican party for the Democrats over the last 30 days.

I also note that the proportion of liberals sampled in today's Times/CBS poll is the highest they have recorded in any Presidential poll since 1995. Maybe Kerry should be worried that the best he could manage was a tie.

It's noteworthy that all of the polls that over-sampled Republicans in September are now over-sampling Democrats in October. Is this a coincidence, or a deliberate effort to manufacture a Kerry "comeback" to generate momentum for the Democrats? One possible explanation, as least as to the Times/CBS poll, is that their September poll was taken on a Monday through Wednesday, while the poll released today was entirely done on the weekend, when pollsters know they will tend to find more Democrats at home. So was the choice of polling dates deliberate, or coincidental?

For what it's worth, those polls that weight samples to produce a consistent blend of Republicans, Democrats and independents have found little or no change since the first Presidential debate.

Posted by Hindrocket at 08:06 AM | Permalink | TrackBack (2)

I'll make this very short: From what I have read here and elsewhere, polls of REGISTERED voters are considered more reliable than those of LIKELY voters. Why do the above results indicate "Likely." Do we not have data for registered voters? Thanks for all the work you do here. I'm a regular.

The best part of this is to read it in light of the "incumbent 50% rule." Bush may have temporarily boosted his standing by virtue of the convention, but he wasted it all by virtue of his performance last week. It will be very hard for him to get it back, since voter perceptions are hard for an incumbent to move and because his only national forum will include Sen. Kerry on the stage with him to keep him in check.

Kerry needs to focus on keeping his own performance up to acceptable levels, and Bush is toast.

I feel obliged to venture the following commentary on the most recent spate of polls, and in particular on the "party-ID-weighting" controversy.

For a good 4-6 weeks we Dems have been crying foul about the fact that most public polls were too heavy on Republican respondents, and not weighted for party ID. Polls so weighted, or otherwise having party ID breakdowns which we believed were more accurate (Zogby, IBD, etc) showed a much closer race than other polls (Gallup, Time) whose Bush margin seemed directly proportional to the Republican overrepresentation in their sample. We believed that the former polls were the more accurate ones, since it seemed highly unlikely that, after having more registered Dems than Repubs voting in 5 straight presidential elections, the Repubs would suddenly leap up with 4-6% more voters this year. True, the 9/11 attacks might have increased the number of people who want to affiliate with the traditional national security party (e.g. Dennis Miller), but on the other hand the fury borne of the illegitimacy of the 2000 election, combined with the stridently and unapologetically ultraconservative policies of the sitting administration, would seem to be equally powerful motivators for people to register as Democrats.

But now what's happened? As Alan Abramowitz documents, the polls have significantly rebounded in Kerry's favor -- and, based on posts by other in other threads (including good ol' Smooth Jazz, who now wants in on this action), individual polls have also shown significant changes in the number of Dems vs Repubs in their samples, which would serve to explain most of the apparent change.

Does that mean we were right all along? If so, then the race has hardly changed -- just the weighting of respondents to more accurately reflect (what we believe to be) reality.

But that hardly seems right either. After all, on RealClearPolitics you can see that in the month of September, after a viciously successful convention and lots of pro-Bush media spin, there were 29 polls showing Bush with a lead, 2 showing a tie, and only one showing Kerry ahead. Now, after a roundly successful debate showdown, markedly better media spin, and an energized campaign, 3 of 6 polls show Kerry tied or ahead. Can that be explained by all of the pollsters, all of a sudden, getting a more accurate split in the party ID of respondents?

That explanation, to my mind, strains credulity. It seems clear that, since the debate, more people are identifying themselves as Democrats, and saying they will vote for Kerry. This is very relevant to the whole controversy over whether weighting for party ID is appropriate or not. Ruy, Chris Bowers (MyDD) et al argue that this is basically a fixed (or only slowly changing) characteristic of the electorate, very relevant to voting decisions, and therefore crucial for weighting. Others (Gallup's Frank Newport) argue that party ID is malleable and fluctuating, and therefore not appropriate as a weighting variable. I have to say, much as it seems clear that Gallup's previous polls have been inaccurate outliers, that if others' analyses of party ID breakdown in the new polls is true, it would lend significant credence to Newport's side.

Of course, the wild flucuations in Gallup's polls suggest that they don't have the right methods either. Maybe we have to come up with some way to identify "hard core" Dems (most of us on this posting board) and "hard core" Pubbies (Smooth Jazz), and make sure those are appropriately weighted (though I have no idea how), since many in the middle seem to be more likely to identify themselves with a side when that side is doing well.

Mixed polling news - Rasmussen has the candidates tied in Ohio, Pennsylvania, Michigan. Florida is also tied within the margin of error.

Interestingly, Bush's lead in some traditionally strong Red states has shrunk to single digits.

Nationally, the 3-day tracking poll has Bush leading by +1.

Has anyone done a large, truly random poll of just plain citizens? Not LVs. Not RVs. Everyone. And if so, do the results differ from the LVs/RVs? By how much? I seem to remember seeing such a poll once, maybe ten years ago, and was surprised that very little difference showed up.

Zogby polling for the battleground states is published online today at the Wall St. Journal site. The polling was taken BEFORE the Bush / Kerry debate and shows Kerry up over Bush in the battleground states, incl. all the ones he needs like WI, PA, OR, NM, MI, and FL. Plus Kerry is up in AK. Some of these are super tight margains.

If the state polling numbers have moved with the post debate horserace numbers than Bush is in big big trouble.

Scroll a quarter to halfway down and look for 'Battleground Poll.'


How can the Gallup Poll say that there has been a 11-14 point swing in voters?

Kerry's internals still weak:
From the CBS/NY Times:
(Likely voters)
Bush voters
Strongly favor 71%
Like with reservations 21% = 92%

Kerry voters
Strongly favor 51%
Like with reservations 25% = 76%

Say's what he really believes
Kerry: 35%
Say's what people want to hear
Kerry : 60%

So clearly, Senator Kerry is still riding the "anybody but Bush" vote. However, while it's easy to be mad at the president and register a protest with a pollster, doing it at the polls is a much higher hurdle.

Fyi, See article at end of Post. If Sen Kerry is not careful and continues to dis our critical ally Poland, he may lose the support of many Polish Americans, who represent a critical block in the following battleground states - Not to mention potential problems if he's elected and has to work with these allies.

Number of Polish-Americans, by state, from the 2000 census, courtesy of NRO:

Michigan: 854,844 (8.6% of total population)
Pennsylvania: 824,146 (6.7% of total population)New Jersey: 576,473 (6.9% of total population)
Wisconsin: 497,726 (9.3% of total population)
Ohio: 433,016 (3.8% of total population)
Florida: 429,691(2.7% of total population)
Illinois, 932,996(7.5% of total population)


Oct 1, 2004

President of Poland Calls Kerry 'Immoral'
Reacting to John Kerry's omission of Polands efforts in Iraq, President of Poland Alexander Kwasniewski said, "I find it kind of sad that a senator with 20 year parliamentary experience is unable to notice the Polish presence in the anti-terror coalition."

When asked about Kerry's derogation of non-U.S. coalition countries fighting in Iraq, Kwasniewski said: "I don't think it's an ignorance. Anti-terror coalition is larger than the USA, the UK and Australia. There are also Poland, Ukraine, and Bulgaria etc. which lost their soldiers there. It's highly immoral not to see our strong commitment we have taken with a strong believe that we must fight against terror together, that we must show our strong international solidarity because Saddam Hussein was dangerous to the world."

"That's why we are disappointed that our stance and ultimate sacrifice of our soldiers are so diminished", President Kwasniewski commented Kerry's speech during the debate.

"Perhaps Mr Kerry, continues Kwasniewski, thinks about the coalition with Germany and France, countries which disagreed with us on Iraq."

Poland has contributed greatly to the efforts in Iraq. Their troop contribution tops 6,500 and 13 have given the ultimate sacrifice, in order to assist the United States liberate Iraq.

What seems to be a poor choice of judgment, Kerry, so far, has not apologized to the nations that he denigrated that have supported America, during these times of challenge.

BritishCombat deaths: 25
Non-combat deaths: 39
Italy 18
Poland 13
Spain 11
Bulgaria 6
Ukraine 6
Slovakia 3
Thailand 2
Portugal 2
Albania 1
Denmark 1
El Salvador 1
Estonia 1
Georgia 1
Latvia: 1

"The six polls included in the analysis are:
CBS/New York Times Gallup
ABC/Washington Post
Pew Poll
Zogby Poll

Looks like 5 to polls me. What weights are you using?

Two new polls out today confirm a tightening race. Fox has Bush up 2 among LVs in a three-way race (it was 4 in their last poll) and American Research Group has a tie among LVs (among RVs Kerry has a 3-point lead).

Two new polls out today confirm a tightening race: Fox has Bush up 2 among LVs and ARG has a tie (Kerry is up 3 among RVs).

There are two more polls in. ARG shows it tied in LV in a three-way. Fox has Bush up by 2%.

Adding those two polls in, Bush's share has dropped from 48.75% to 47.5%, and Kerry's is up from 43.13% to 45.88%. That closes the gap from 5.62% to 1.62%. 7 of the 8 polls show gains for Kerry. The 8th shows no change.

If you go to RV's, there are 5 polls, as best I can tell. Newsweek, ARG, ABC, Gallup, and Pew. Bush falls from 49.2% to 47.2%. Kerry rises from 42.6% to 45.4%, all in 3-ways. The Bush lead drops from 6.6% to 1.8%.

Couple that with the movement in Rasmussen today (with Bush's lead dropping from 2.5% to 0.9%) and I think it's clear that Kerry gained. And this all suggests that the race is roughly within the margin of error.

Now to keep that momentum going....

The ARG Poll with Kerry up 3 would lower the deficit to maybe 1.0

Don't mean to be 'Debbie downer' here but here's some state polls NOT on that list from SINCE the debates in key swing states that show Bush leading
(cited at the 'Race 2004' website): (1) Ohio -- Bush up by 3, 10/3, Rassmussen; (2) New Hampshire -- Bush up by 5, 10/3, Granite State Poll; (3) Nevada -- Bush up by 4, 10/3, Survey USA, (4) Florida -- Bush up by 2, 10/3 Inside Advantage; (5) Florida --Bush up by 4, 10/3, Rassmussen; (6) Florida -- Bush up by 5, 10/3, Survey USA

These aren't insurmountable leads but if Bush carries all four of those states, and continues to lead in Wisconsin, Missouri, and West Virginia, the race may be closer but not caught up. Since we don't have state breakdowns from the national polls, these statewides could be important.

and I still think the machine agenda is a better predictor than the best read polls, but we'll see

And good news in...the first post-debate Ohio poll, by SUSA, shows Kerry in front there, albeit by only 1%. And an Iowa poll conducted with the debate in the middle also shows Kerry in front, by 1%.

If Kerry can carry the Gore 2000 states, except for Wisconsin, including both districts in Maine, and gets only Ohio of the Bush 2000 states, Kerry wins, 270-268. Changes in Iowa and Ohio coudl be key here, folks.

Can you comment on http://www.electoral-vote.com/fin/oct05p.html? This site projects state polls into an electoral count. It has been consistently showing Bush with a strong lead (projected as of 10/5 is Bush 350/Kerry 178). Although the national %s are close, it does appear that the electoral count is not anywhere close. I'm really interested in understanding the quality of these state polls and because the electoral count is what really matters in the end.

Right now, "Shrub" is tied in the polls or (at worst-) leading by a few points with less than four weeks to go. I would like to think this is enormously good news for the challenger, since fence-sitting voters usually break away from the incumbent. If he hasn't persuaded a clear majority of Americans he deserves a second term yet, it probably will never happen?

Can anybody please post comparable statistics for Nixon, Ford, Carter, Reagan, GHW Bush and Clinton? How were they doing at this stage in the reelection cycle?



Quick stop by this morning. The electoral vote site uses the most recent poll. (Actually, the site has moved to averages of the three most recent polls.) Very few state polls have been conducted since the debates, so the electoral vote site will not change all that fast. You'll note that it is now 285-232 Bush, with 21 undecided, thanks to that new Ohio poll. New Hampshire and Michigan are shown as tied. Right now, it shows Kerry getting Ohio from the Bush 2000 states, and Bush getting Wisconsin and Pennsylvania from the Kerry 2000 states. If Kerry can get Pennsylvania (and a West Chester U poll, the most recent at race2004.net, shows Kerry up by 7%...and several other polls showed Kerry ahead there in late Sept.) and Michigan, where only 1 of about 30 polls since the end of June has shown Bush in the lead, while keeping Ohio, he wins.

Ted K. -

The electoral-vote.com site's "votemaster" has discussed many of the same issues that we see here - an incumbent's inherent disadvantage among undecideds, RVs vs. LVs etc. etc. He's been tinkering with his methods; right now he's basically using the most recent legitimate statewide polls. Much of Bush's "lead" is based on states well within the polls' margin of error. As of today the site shows Bush ahead 285-232, but expect a lot more fluctuation and pray (better yet, WORK) for continued Kerry momentum.

Answering my own question --

(President -- poll date -- Approve -- Disapprove -- Undecided)
"Shrub" 10/1-3/04: 50-48-2
Clinton 10/1-2/96: 58-34-8
Bush 10/1-3/92: 34-58-9
Reagan 10/26-29/84:58-33-9
Carter 9/12-15/80: 37-55-8

These are all Gallup polls, from October (if available). As others have noted, "Shrub" is not doing as well as the presidents who won reelection but nor is he doing as poorly as the ones who lost (Bush pére, Carter, Ford). Note, also, the unusually small number of voters who have no opinion regarding his performance.


I don't recall exact numbers for incumbents at this stage in the election cycle, but from what I do remember or have read (I'm too young to remember some of these elections firsthand), they went like this:

Nixon (72), Reagan (84), and Clinton (96) were cruising to victory (despite a brief bump in the road for Reagan after a weak debate performance, and Clinton suffered a bit from fundraising scandals.)

Ford (76) was behind but gaining, but fell just short.

Carter (80) was even with or a bit ahead of Reagan throughout most of the fall campaign, but things collapsed in the final week.

Bush the Elder (92) was somewhat behind, although not assured of losing (Perot was splitting the opposition vote.)

To round out the cycle, I'll add that Bush I was solidly ahead of Dukakis at this point in 1988, and in early October of 2000, as the debates started, W was overtaking Gore in the polls.

Thanks for the info, Mike.

I guess the current state of the race is a bit puzzling. Is the glass half full or half empty? It is probably fair to say "Shrub" is a relatively weak incumbent since he isn't cruising towards a second term like Nixon, Reagan & Clinton. On the other hand, a more appealing candidate than Kerry probably would have a commanding lead (Carter & Clinton certainly did).
I guess I remain cautiously pessimistic. But Kerry does have an excellent chance, *if* he can win the remaining two debates on points as well. "Shrub's" appeal is entirely due to the "strong determined war leader" myth. If he is frequently seems weak, poorly informed and out-of-touch when debating Kerry, even those voters who rarely follow politics will understand the anti-Kerry ads are mostly partisan scaremongering.

BTW, do you guys & girls think the new format (=townhall debate, answering questions from "ordinary American voters" in the audience) will help or hurt Kerry on Friday? I hear he is better at dealing with unexpected, "tight" situations than "Shrub".