« The Bush Landslide of 2000? | Main | The Race Tightens »

October 10, 2004

The Second Debate

A busy weekend has made me late to the party on the subject of debate reaction. For those who have not seen them, three survey organizations -- Gallup, ABC News and Democracy Corps (a Democratic polling entity) -- did post debate surveys of debate watchers. Each one polled debate viewers they had previously interviewed earlier in the week, Gallup and ABC by telephone and Democracy Corps via a projectable and representative Internet panel (the merits of which, I discussed here).

Gallup described the debate reaction as "a standoff"and ABC News said debate viewers "divided along partisan lines," because the results in both surveys showed that partisans on either side judged their man the winner by nearly identical margins (or at least within sampling error). Vote preferences barely budged during the debate. Historically, that is the typical reaction -- the debate itself tends to reinforce existing preferences.

However, most of the coverage has overlooked the fact that independents perceived Kerry as the winner on all three surveys:

  • Gallup - "Who do you think did a better job in the debate?" Kerry 53%, Bush 37%
  • ABC News - "Who, in your opinion, won the debate?" Kerry 44%, Bush 34%
  • Democracy Corps - "Who would you say won the debate?" Kerry 44%, Bush 33%

The ABC polling unit looked at their own results and noted (in a press release) that the 10-point Kerry margin among independents was "within the margin of error for the number of independents surveyed." True enough. Given that ABC reported a "35-32 percent Democrat-Republican division" in their survey, independents could have been no more than 170 respondents out of 515, for which the margin of error would be at least +/- 7%.

However, the three surveys combined included more than enough interviews among independents to make a 10-point difference significant.* All of this may be fleeting, but for those who actually watched the debate Friday night, Andrew Sullivan's conclusion that "Bush rallied his base, but Kerry won over the middle," seems well supported by the data.

*I know, I know. Question wording and sample frames were slightly different, so from a statistical purist's perspective, lumping these three subgroups together is not strictly kosher. However, the methodological differences in this case seem trivial, and the nearly identical results from all three surveys gives me confidence that they are real. The total number of idependents sampled fell somewhere between 450 and 700, depending on how Democracy Corps defined its independent subgoup.

Related Entries - Debates

Posted by Mark Blumenthal on October 10, 2004 at 06:08 PM in Debates | Permalink


None of the polls ask questions about the demeanour of Bush or Kerry, yet most of the post-debate discussion centres around this. The intangibles of character - and performance during a tough debate - are very significant factors which make voters stick to their choice, or make a choice. In this respect, these polls do not give enough weight to the behavior of the two candidates under the stress of the election. Simple questions such as who won a debate are too simplistic. Are there any "proxies" in the questions asked of voters polled during the past three months or so which could be interpreted in the light of such intangibles? If so, what story is told about trends? The President's subpar behaviour in the first two debates, when contrasted with the bearing of Kerry, are probably more decisive for this election right now than even voters' expressions of who they will vote for. Many people expect a change amongst voters in the next few weeks due to these factors. Can you shed any light on this?

Posted by: CuriosityKilledTheCat | Oct 11, 2004 1:26:13 AM

Mr. Blumenthal,

Does this mean you'll be genuinely surprised as your winner Kerry falls to 6-8% below Bush in the polls this week?

Just asking.

Posted by: Arlen Williams | Oct 11, 2004 3:21:27 AM

Show me a poll that says that and then we'll talk. In the mean time - to ignore the Kerry momentum is foolish - most polls are moving in Kerry favour.

Posted by: Rick | Oct 11, 2004 3:54:58 AM

Alan Keyes also does not have a shot- as much as you also may think so. Delusion leads to grandeur

Posted by: Rick | Oct 11, 2004 3:56:02 AM

Sure looks like the polls are now moving in Bush's favor. Maybe "winning the debate" isn't as important as having debate watchers think that what you are saying is correct.

Posted by: Steve | Oct 11, 2004 6:24:19 AM


One word can make a big difference.

"All of this may be fleeting, but for those who actually watched the debate Friday night, Andrew Sullivan's conclusion that "Bush rallied his base, but Kerry won over the middle," seems well supported by the data."

I disagree, for the inclusion of one word. That word: "over". There is plenty of evidence that Kerry won the debate in the eyes of the middle. None of the evidence you presented suggests that he won over the middle.

A good example of how this can be would be to look back at the National Review's blog for the first debate. They are not middle, but the phenomena is analogous. They were pretty much unanimous that Kerry had won the first debate. I do not think Kerry had won over a single one of them. People can, and often do, make judgements about who won a debate separate from if they are going to change their votes.

Is there evidence that Kerry won over the middle? Looking first at the ABC writeup you linked, I see that in the vote preference among viewers of the debate, Bush actually gained a point. They do not include partisan breakouts on that question, but there are a few possibilities. One is that Kerry won over the middle, but Bush gained Republican support; this is unlikely because most polls have shown the President has little room for growth there. The second is that Kerry won over the middle, but Bush won over some Democrats; it is possible, but seems unlikely. The most likely fit for the ABCNews data: Kerry won the debate in the eyes of the middle, but did not win them over.

The Democracy Corps poll showed Kerry gaining two points overall. Did this come from the middle? There is no supporting evidence.

Did Kerry win over the middle? I suppose it is possible, but the post you made does not give evidence to support the idea. If he did, the tracking polls sure have acted strangely, with them showing not much movement at all (and some showing ever-so-sliught movement towards Bush) since the second debate.

Besides, polls taken shortly after a big event are poor for determining who benefitted from the event. I prefer to see how things are a few days later. Those polls should start hitting us later this week.

Posted by: Gerry | Oct 11, 2004 7:39:37 AM

I added a few more thoughts over at my place.

Posted by: Gerry | Oct 11, 2004 7:48:07 AM

wash post poll coming out today shows bush 51, kerry 46. now that's winning the middle over... for a week.

Posted by: Dave | Oct 11, 2004 11:28:48 AM

Should be noted that the Wash Post poll had the exact same result after the first debate...and before the first debate, for that matter, so it may not be the best indicator of how the polls are moving. Zogby/Reuters tracking poll, which had Bush up 2-3 after the first debate, now has Kerry up 1. Kerry solidified his base voters and reached out to the middle, Bush simply continued to massage the right wing

Posted by: Eamon | Oct 11, 2004 11:35:33 AM

Rick says:

> Arlen,
> Show me a poll that says that and then we'll talk.

Wash Post poll showing 51-46 is five points. Good enough for you?

Posted by: Bruce | Oct 11, 2004 1:11:08 PM

I'm still looking for the second internet the President mentioned on Friday night. I mean, I knew Al Gore invented the first one, and he really should have placed a patent on it. If anyone finds it can you reply to this post?

Posted by: Dan | Oct 11, 2004 2:31:59 PM

Arguing about the day by day gyrations of a bunch of polls asking different question of different groups of people is futile. It's all going to be decided by people the pollsters don't have any clue about. People newly registered to vote, people who haven't made up their mind yet. The democrats have been kicking ass on the republicans this year, registering way more voters (we're going to hear a lot more frivolous voter fraud claims this time around), mostly becasue there are more natural democrates in the big vote states than there are republicans. You add to this the fact that independants always swing to the challenger at the last moment and I don't think it's even going to be close.

Posted by: cw | Oct 11, 2004 2:51:26 PM

Given that the above-referenced WaPo poll has Bush/Cheney rising by 1 point and Kerry/Edwards staying the same, Zogby showing that Kerry/Edwards is up 3 over Bush/Cheney from being down 1 in their last poll, and Gallup has Kerry/Edwards up 1 after being tied with Bush/Cheney in their last poll, and almost all signs show Kerry/Edwards with the momentum, I think that Mr. Blumenthal would be _very_ surprised if "Kerry falls to 6-8% below Bush in the polls this week."

Posted by: Anonymous Blogger | Oct 11, 2004 4:02:20 PM

One of the things polls can't measure is sheer numbers. We don't know turnout yet, how many new registered voters there are on each side and how many of those newly registered will turnout.
Those things will determine the final outcome. Zogby, Gallup, NYtimes, etc. may make for interesting horse race fodder but they have little or nothing to do with outcome. Foot soldiers will win the day. Or maybe the courts not the pollsters.

Posted by: bill | Oct 11, 2004 6:26:01 PM

It seems to me that if you want to find out who won the debate, asking people "who do you think won the debate?" is the wrong way to go about it.

There's an old saying--"When Cicero spoke, people said, 'how well Cicero speaks!' When Demosthenes spoke, people said, 'let us march against Philip'." Which is to say, someone can impress people as being a good speaker (or a good debater) without getting results.

As I see it, the way to measure who won a debate is to see what happens to people's opinions about the candidate--either in the horserace polls, or job approval, or personal approval. But asking 'who do you think won the debate?' is at best a very indirect measure of that. It would be perfectly possible for people to watch a debate, and honestly think "Kerry did a better job debating, Kerry won the debate"--and still get up from the couch more likely to vote for Bush (or vice-versa).

Just as a hypothetical--suppose that Smith gave a lucid, well-argued, intelligent, ruthless argument for raising taxes on all income levels. Suppose that Jones fumbled his answer and looked stupid. Viewers might get up thinking "Boy, Smith sure whipped Jones on that question"--but also thinking "I'm not going to vote for Smith, he'll raise my taxes". In that circumstance, Smith might think he "won" the debate--but if the debate made Jones more likely to get elected president, then Smith didn't win any victory worth having.

P.S. By any measure, Kerry won the first debate, of course. His standing in the general polls went up. But I think the second debate might be different. Pollsters are showing that more people think Kerry won the debate--but Bush has been moving up in the polls since that debate, which is the real test of victory.

Posted by: Andrew S. | Oct 12, 2004 3:05:47 PM

Just a comment on this race and the debates.

It sure is interesting that this race is so extremely close considering that Bush/Cheney have been in office the past 4 years. Regardless of who comes out ahead in this race, there are definately a lot of people that do not support Bush. And that is pretty sad. If the American people thought he was doing so well, this race wouldn't be so close. And it doesn't take a genious to figure that much out. It is my personal opinion that Bush is the flip-flopper and the wishy washy one. He has turned our country upside down.

Posted by: independant from michigan | Oct 13, 2004 5:13:28 PM

'If the American people thought [Bush] was doing so well, this race wouldn't be so close.'

And if the American people thought Bush was doing so *badly*, this race wouldn't be so close. Evidently the American people disagree about just how well Bush is or isn't doing.

Posted by: Andrew S. | Oct 15, 2004 1:54:20 PM

The comments to this entry are closed.