« The Second Debate | Main | Thank you »

October 10, 2004

The Race Tightens

As long as I'm back on the topic of debates, it is worth remembering that media coverage of the debate typically has more impact than the immediate reaction of debate viewers. In the ten days since the first debate on September 30 almost every major public poll has released a new survey, and we are now in a position to see if the voters' preferences have changed.

Although some see conflicting trends, it is now clear that the race has narrowed since first debate. The RealClearPolitics weekly average of the three-way vote (including Nader) had George Bush running ahead of John Kerry by 6-7 points (roughly 49% to 42%) now has Bush ahead by less than two points (47.5% to 45.8%). Of course, RCP simply averages whatever polls were reported in a given week, and their approach is often less than apples-to-apples.

To create more comparable averages, I used results of polls from eight organizations that were in the field just before and after the debates (once between September 17 and 28 and again between October 1 and 7). The organizations were ABC/Washington Post, AP/IPSOS, Gallup/CNN/USA Today, Fox/Opinion Dynamics, ICR, The Marist Institute, Time and Zogby. All eight reported results for "likely voters;" all but Fox and Zogby also reported results for all registered voters.

  • Among "likely voters," the Bush lead narrowed from an average of six percentage points (50% Bush, 44% Kerry, 2% Nader) to just two (48% Bush, 46% Kerry, 2% Nader).
  • Among all registered voters, the change was nearly the same: the Bush lead narrowed from seven points (50% Bush, 43% Kerry, 3% Nader) to just two (48% Bush, 46% Kerry, 2% Nader).

Two other thoughts: First, remember the incumbent rule. If the election were held today, Kerry would likely get most of the average 4% that is still undecided, resulting a photo finish.

Second, the shift over the last two weeks has been a modest 2% net change for each candidate. It may be that only 4-5% of the voters have shifted allegiance, yet the race is close enough that such small changes have great potential consequence. Keep in mind, if they could shift Kerry's way, they could also move back.

----------
Note: The last time I averaged poll results, several readers questioned the wisdom of averaging. From the perspective of pure sampling theory, they have a point. The different polls have different sample sizes (likely voters ranged from n=630 to 1,036), ask slightly different horse race questions, use different likely voter models and different weighting schemes. If we were designing a survey from the ground up and wanted to apply precise statistical tests to the resulting data, averaging is a bad idea.

Having said that I do not agree that averaging in this context attempts to "commensurate what is effectively incommensurable," as one reader put it. I selected polls that were in the field just before and after the first debate, so that each pair of surveys is methodologically comparable. I also averaged results for likely voters and registered voters separately. While different weighting, sampling and likely voter models can produce different results, there is some benefit to pooling the combined "art" of eight different pollsters.

Related Entries - Interpreting Polls, The 2004 Race

Posted by Mark Blumenthal on October 10, 2004 at 11:22 PM in Interpreting Polls, The 2004 Race | Permalink

Comments

I have been campaigning hard for Kerry in the Bible Belt of SW Virginia. For me campaigning hard means wearing buttons, puttin up signs, bumper sticers, manning the office, and engaging as many fence sitters (and often hardcorps republicans in conversational debate). I tend to save my breath for the fence sitters. I am growing increasingly aware of the polarity growing out of the divisive nature of the issues. So I decided to try to write a letter to the editor of our small local rag, with an attempt to build a bridge back to the times we could disagree and then go out for pizza.
Here it is:

As we rapidly approach election day, 2004, the following lines from “The Lord’s Prayer” give me strength: “Thy Will be done – on earth as it is in Heaven.”

I have always believed that God’s Will was that all humans celebrate life on this planet in harmony. I also believe that is what all of us want, even though some of us might not know it. That is my faith in humanity. But we also know that God helps those who help themselves and that we are not empty vessels, devoid of thought. We must struggle in our lives. Lately that struggle, if we are good Americans, has been very much about deciding for whom we will vote in the up-coming elections.

I also believe that people that care strongly about politics in the United States are, likewise, motivated by the same desire to attempt to create an environment on our planet that more closely resembles that individual’s vision of heaven. I am willing to admit that I cannot know that the choice I make is the most expeditious one toward that end, but I truly believe I have done my homework and have made a choice that has the best chance of achieving those goals. No, I am not leading up to why you should vote for a particular candidate, though I celebrate my civil rights often and therefore make no secret of my beliefs. The point of this letter is not to convince anyone of how to vote, but rather to encourage respect for those of differing opinion.

My fear is that after November 2, 2004 the internal tension in The United States will increase and communication between those that have firmly committed to different sides will increase. I am dreading the recounts and claims of voter fraud, should the election be a close one. Therefore I am hoping that if the guy I don’t like wins, he wins by a landslide, which is not to say I will stop campaigning for, contributing to, supporting, or trying to convince anyone I can that my choice is the right one. Obviously, my best case scenario is my guy by a landslide, along with a majority in both houses of congress.

No matter what the results are on election day, I plan to continue to strive to be a good American. I cannot guarantee I will not be involved, at least emotionally, if there are claims that my side was cheated, but I can say, with all my heart, that I will support, albeit with a critical eye, the decision the American voters make, no matter what it is. I will also try to be understanding of the hurt and fear and anguish of those who are convinced it should have been their candidate, if the side I am rooting for emerges victorious. What makes it easy for me to be supportive of “the other side” is not just my faith that it is God’s Will, but also that I personally know and have the utmost respect for many people who have the opposite opinion that I hold about how to best pursue our common goal of Peace on Earth, or a reasonable facsimile thereof.

In the long run I continue to have faith that all Americans are on the same “side”. Being involved is what makes good, patriotic citizens. To say someone can only be a “good American” if they think like you, or that they are “un-patriotic” if they do not agree with you, strikes me as un-American in itself, and infantile to say the least. I am not sure there is a high road or a low road, but just two different roads. Maybe there should be a whole lot more roads. For each of us to have a candidate that perfectly reflected our individual views, would probably require each one of us running for president. In any case, all paths lead to the future and we are all heading there together. Let’s go there with our heads held high, a sense of humor and faith in our hearts that no terrorist can destroy.


Posted by: Laird Baldwin | Oct 11, 2004 12:30:31 AM

"No matter what the results are on election day, I plan to continue to strive to be a good American...I will support, albeit with a critical eye, the decision the American voters make, no matter what it is."

Respecting a differing opinion is not the same as accepting lies, or allowing them to be perpetrated.

If Bush wins, it will be because the voters were duped. Republicans have established a machine to gain control based on lies, character assination, corruption and control of the media. The latest Gallup poll revealed that some 40% of the population STILL believes that Saddam had WMD, which is a direct result of Bush administration/Republican propaganda.

Being a 'good American' is to fight for Democracy, not to accept the consequeces of a corrupt power grab by a proto-fascist, imperialist regime. The Vietnam war was terminated by protest, not acquiesence.

Probably the best thing for the long term is a Bush win. Only in that way will the full measure of his mendacious, corrupt anti-democratic policies penetrate the collective consciousness. That will trigger a Democratic outrage strong enough to put the Republican brand of politics away for a long time.

Posted by: Ken Wheeler | Oct 11, 2004 4:54:36 AM

Most of these poll results are from both registered voters and likely voters. Are there any polls that bring in new registrations into the equation? States like Ohio and Florida are showing record breaking new registrations, and I would tend to think that would favor democrats than republicans. If that is indeed the case, I would think that should pull Kerry ahead by a couple of points. But I'm not the professional. Any idea?

Posted by: Daniel Swartz | Oct 11, 2004 8:03:43 AM

Quick answer, which deserves a longer post: Registeration in all of these surveys is based on a "self report," meaning that registered voters identify themselves. If you registered yesterday, you would be included in the survey of "registered voters" today.

Likely voter models *can* be a different story, as many use self-reports of past voting history or (for Gallup) knowledge of the polling place to define a likely voter. A new registrant would disqualify. I'm working on much more on this topic this week.

MP

Posted by: Mark Blumenthal | Oct 11, 2004 9:18:19 AM

The WaPo daily tracking poll is including 18-21 year-olds who report to be following the election closely as "likely" voters.

Posted by: Eric | Oct 11, 2004 10:58:11 AM

You wrote - Quick answer, which deserves a longer post: Registeration in all of these surveys is based on a "self report," meaning that registered voters identify themselves. If you registered yesterday, you would be included in the survey of "registered voters" today.

Discounting the way new registrants bias poll results is almost certainly misleading for the following reason. Unless the pollers are really incompetent they are using stratified samples. For instance, for a statewide poll of Pennsylvania, a reasonable stratification would be Philadephia, Philadelphia suburbs, northeast counties, etc. In each stratum they use random sampling, but the number of respondents per stratum is determined in advance, ordinarily by the number of actual voters in that stratum in past elections. Thus IF there are differential increases in voters say in Philadelphia as compared to the Philadelphia suburbs, AND these new registrants actually vote, the poll will not include the proper (i.e. actually voting) proportion of Philadelphia votes. Similarly, in a national poll, the strata might be atates or even larger units, but the same principle applies. If there are differential increases in registration in Ohio and Pennsylvania compared say to Texas and Oklahoma, the fixed number of interviews in each state will result in underestimating the effect on the national figures (which of course are meaningless anyway in outr electoral college system) of the new registrants.

By the way the use of stratified samples generally reduces the "sampling error" in fairly significant ways. So the reported sampling errors, which are generally taken from tables that do not include the effect of stratification (for the good reason that each stratification has a unique effect), generally overestimate the sample error, which would be meaningful except for the fact that sample error is such a trivial part of the real error in polls.

Posted by: Mike Rappeport | Oct 11, 2004 9:44:28 PM

It seems that Gallup is today (Oct 11) tracking Kerry 49%, Bush 48%. And Zogby's tracking shows Kerry 47% to Bush 44%.

I guess all those comments yesterday about Kerry slipping 6-8% in the polls this week are a bit further off the mark, now.

Of course, we still have the debates on Wednesday...

Posted by: George | Oct 11, 2004 9:50:57 PM

For those interested, I've posted my latest (10/11) survey of 55 Electoral College tracking / prediction / projection / forecast sites at http://unfutz.blogspot.com/2004/10/electoral-college-survey-1011.html.

Executive summary: Although Bush still leads according to a majority of sites, Kerry has made substantial headway in re-gaining ground he lost to Bush in the last month, reducing a 50 point gap to a 15 - 20 point Bush advantage. Currently Bush has 262 t0 269 electoral votes, while Kerry has 247 to 250.

Posted by: Ed Fitzgerald (unfutz) | Oct 12, 2004 3:05:08 AM

This last minute attack will render Mr. Kerry defenseless. If not countered effectively, the poll will make no sense.

Quote:

TV stations clear schedule for Kerry attack
By Andrew Buncombe in Washington

12 October 2004

The owner of the largest chain of television stations in the US and a huge contributor to Republican causes is to disrupt its regular programming schedules two weeks before the election to air a documentary highly critical of the Democratic challenger John Kerry.

Officials from Sinclair Broadcasting confirmed that the company has ordered its 62 stations - many of them in the swing states of Ohio, Florida, Wisconsin, Pennsylvania and Iowa - to air Stolen Honour: Wounds that Never Heal, during primetime slots next week.

The documentary, by a small production company in Pennsylvania, focuses on Mr Kerry's anti-war testimony to Congress in 1971 and actively links him to the anti-war campaigner Jane Fonda. The documentary includes a number of interviews with former PoWs who claim that their Vietnamese captors used Mr Kerry's comments to demoralise them.

Mark Hyman, a spokesman for Sinclair, said yesterday that the show would contain some or all of the 42-minute film as well as a panel discussion of some sort. He said final details had not been worked out but that Mr Kerry had been invited to participate.

He told The New York Times: "Clearly John Kerry has made his Vietnam service the foundation of the presidential run - this is an issue that is certainly topical."

The Kerry campaign dismissed the documentary while the Democratic National Committee said it planned to file a complaint with the Federal Election Commission arguing that the airing should be considered an illegal contribution to President Bush's campaign.

In a letter to the commission, a group of 18 Democratic senators asked for an investigation. "To allow a broadcasting company to air such a blatantly partisan attack in lieu of regular programming, and to classify that attack as news programming as has been suggested, would violate the spirit, and we think the text, of current law and regulation," they wrote.

Many Democrats will be concerned that the documentary could have a similarly damaging effect as the notorious and largely disproved Swift Boat Veterans for Truth advert, which sought to undermine Mr Kerry's claims about his actions in Vietnam. After that advert was broadcast in many swing states, Mr Kerry saw his numbers in the polls fall markedly.

Posted by: swing | Oct 12, 2004 9:12:55 AM

Does the incumbent rule, most undecideds break for the challenger, apply to House and Senate races?

Posted by: Dan | Oct 12, 2004 11:49:35 AM

If you are going to average the polls, you should use a weighted average. That will at least factor in the difference in sample sizes and produce a more accurate result.

Posted by: Tammy | Oct 12, 2004 1:11:13 PM

Weather forecasters talk about a "ground forecast" i.e. look out the window. I have one of those to report from Western New York. Rochester, New York, is a mid-Western city that somehow wound up in NY instead of Ohio. The surrounding area is and always has been heavily Republican and people there put signs on their lawns. During my frequent trips there, where I grew up by the way, there are far too few Bush signs from the President's side of things. He should be leading Kerry at least 3-2 if not 2-1 or more. In fact, my and my wife's unscientific sample (noted as we drive around the county) is that Kerry is significantly ahead in signs. If our eyes haven't deceived us, I think that does not bode well for the incumbent.

Posted by: Lennie | Oct 12, 2004 3:19:00 PM

Mike wrote...

"Discounting the way new registrants bias poll results is almost certainly misleading for the following reason. Unless the pollers are really incompetent they are using stratified samples"

Believe it or not, most national polls do not statify by actual turnout statistics. Fox/Opinion Dynamics is an exception that I discuss today in my post on weighting by Party, Part III:
http://www.mysterypollster.com/main/2004/10/why_how_pollste_1.html

Most national surveys (and a surprising number of public statewide polls) stratify their samples of adults by *population* statistics only, then let the self-identified registered voters (or likely voters) fall out regionally without any additional stratification or control.

Posted by: Mark Blumenthal | Oct 12, 2004 3:33:29 PM

Do all of the polls report the Kerry v. Bush numbers separated by Democrats, Republicans, and Independents? If so, then why not ignore how (or even if) they weight by party. Instead compute 3 averages for the Kerry v. Bush numbers across all polls: one for Democrats, one for Republicans, and one for Independents. Then, use different party weights as a parameter (e.g., 3%-point more R than D, 3%-point more D than R, or even R and D). That way, we'd get a handle on what the poll results as a function of relative party turnout. That makes more sense to me than simply averaging all of the polls' bottom lines when 1) they all have different party weights, and 2) none of them know what the true party weights should be.

Posted by: Josh | Oct 12, 2004 3:59:10 PM

The polls margin of error is really larger than advertized, as I don't think they take the following into account:

- Does a R or D typically answer the phone?
- Is a R or D more or less likely to want to participate in a poll?
- Is an 18 year old typically the first to answer the phone?
- Can they poll unlisted numbers?
- Would someone with caller ID, works nights, has bill collectors, etc get a poll call?
- If Cleveland has many more people than Cinciannati, do they call in that proportion?
- The cannot call cell phones, those living out of the state they are voting, or out of the country.

These things seem to translate to a huge margin of error, much larger than the difference between the candidate's numbers.

Posted by: me | Oct 12, 2004 4:16:37 PM

Where have the good ol' democrates of the past gone? Democrates today are better defined when referred to as "liberals". I see an increasing number of people moving over to the Republican side due to the far left policies of current Democrates. There will never be one political party that fully represents a persons entire views, but we all know a moderate will never win a presidential election.

Posted by: Zeb | Oct 12, 2004 9:46:37 PM

That's funny because I see the opposite. I see a Republican party that was founded on LESS government control and MORE fiscal responsibility trying to control and spend more.

It's ok to run up defecits as long as the money is returned in tax cuts. It's ok to prevent drugs from coming in from Canada as long as it benefits drug companies. It's ok to have the a religious monument outside a courthouse, force prayer in school, or have companies like Sinclair, Clear Channel, or Fox control so many stations...as long as it promotes the Conservative Christian agenda.

This is the same mentality that exists in the middle east. When you see the koran on your way to a trial, radio and TV stations pushing a muslim agenda, or your kid being forced to pray to buddah in school...ask yourself which country you are in.

Posted by: me | Oct 13, 2004 11:29:30 AM

The race seems to be widening again, with the RCP three-way average showing an average 4-point Bush lead and the head-to-head a 3.8 Bush lead.

This is up from less than a point before the debate. Kerry may have won the debate so where is his lead?

Posted by: Don | Oct 17, 2004 9:57:06 PM

The comments to this entry are closed.