washington, dc

The Democratic Strategist

Political Strategy for a Permanent Democratic Majority

Kerry, Bush Tied in New Economist Poll

John Kerry and George Bush are dead even at 46 percent of nation-wide RV’s, with 2 percent for Nader in a new Economist/YouGov Poll conducted 9/27-9.

47 comments on “Kerry, Bush Tied in New Economist Poll

  1. Roger on

    Yes Smooth Jazz all the national polls are wrong. You have not answered my question. How come Gallup shows Bush ahead by 13 among registered voters nationwide, while Kerry is ahead in Ohio and PA among registered voters? Does this make sense to you? Or do you believe that all CA and NY state polls showing Kerry ahead by double digit margins in those states are wrong? If Gallup shows me in one poll that Bush has a 10-12 point lead in PA, I will believe the Gallup that Bush is ahead by 13 nationwide. Gallup cannot even show that Bush is ahead by 13 in Ohio among likely voters, let alone show it among registered voters. What a bunch of jokers?

  2. Euzoius on

    If Kerry doesn’t spend some time tonight trying to get those 42% of Americans to understand the actual facts about Saddam and 9/11, and force Bush to acknowlege the truth during the debate as well, that will constitute a mistake. There aren’t going to be enough “persuadables” left to come over to Kerry if this level of idiotic misinformation continues.

  3. Smooth Jazz on

    I appreciate those comments; And, no, I’m not BJ. While my comments can be facetious and biting, I try as best I can to stay with the facts. Morever, I have no hatred for Sen Kerry, and my intention is not to demean his supporters.
    I believe his vacillations and equivocations would be a disaster for our country during a time of war, but I have no animus towards his supporters – And respect those supporters that provide cogent arguments to support their belief in him.
    As for which Rep blogs are good for being open minded, my best bet is Polipundit, To be sure, Poli is a very strong GWB supporter and only posts news favorble to GWB, however they are good about welcoming the Dem POV. Poli does allow site visitor postings.
    RealClearPolitics & NRO KerrySpot are also good providing certain pro-Kerry articles and commentary, but again they are pro-GWB and do not allow comments from site visitors.

  4. howard on

    Lawrence, for what it’s worth, we’re all familiar with the study that showed that Fox viewers are the most ill-informed, but a Newsweek poll in early September showed that 42% of americans think that saddam was DIRECTLY (their caps, not mine) involved in 9/11.
    I’m betting that 99% of those people are voting for Bush.
    http://www.pollingreport.com/iraq.htm (scroll down to the Newsweek poll of 9/2-3)

  5. pdb on

    I for one don’t think Smooth Jazz’s posts are at all out of line, or inappropriate to the blog. Not that we can’t disagree with them vigorously when we see fit; but we should acknowledge an obvious good faith effort to comply with the guidelines. The contrast in tone and substance with the old BJ Clinton posts is quite striking. (Does the reference to “other camouflaged aliases” mean they are the same person, BTW?)
    Yes, there are several polls that show a 6-10% Bush lead, and several others that have it more or less a dead heat. In the latter category Jazz fails to mention Rasmussen, who is a Republican in his personal commitments but comes down mainly on the side of Ruy on the polling-methodology we have discussed here (I think he learned from his errors of 2000, when he gave Bush a consistently substantial lead throughout the campaign). But I don’t think the question can be resolved by counting pollsters on each side; either a methodology makes sense or it doesn’t, however many people use it.
    Jazz, can you recommend a GOP blog some of us might be interested in or welcome at? I might be willing to try exposing myself to opposing positions now and then….

  6. Smooth Jazz on

    snicker, Regarding:
    “I’ve never seen you actually rebut any analysis of polls presented here. You do but pooh-pooh. How droll. How uninspired. How utterly boring.”
    If you really believe that, then you must not be paying attention to my posts. That said, I don’t see your point: Is it that I don’t belong here, since this a Dem site, or is it because you believe ALL POSTS sympathetic to Kerry are correct, and there is no other perspective out there – And every other posting or poll outside your cocoon is wrong.
    In other words, CBS, AP, Wash Post, Gallup, ICR, LA Times, ABC & Pew, which all show GWB up 6 – 10 points in the case of BOTH LVs and RVs are all WRONG – And that Kerry is really tied or ahead because the Economist & IBD are correct. Is that really what you are trying to say.?
    I visit many Rep bloggers, and all that I’ve visited welcome inputs from Dems, if only to keep them honest and to get another perpective. Judging from the comments here, I believe that: Some here like Kerry for who he is, but most appear to have a visceral hatred of GWB and couldn’t care less who the Dem candidate is, regardless if he has consistent positions or not.
    Either way, you can certainly petition the board management to ban me if you don’t want to hear anything outside your cocoon. Rest assured, I will not use any other camoflaged aliases to get back on – I’ll just move on.

  7. erg on

    Gallups shows Kerry ahead conformtably in RVs in PA and Ohio, and barely behind in LVs in Ohio. Why should we diss Gallup for that ? Its clear that Gallups LV screen favors Repubs a litltle (not because they’re biased, its just their model). They’re showing 7-8 point swings in RV to LV conversion, which does look a few points too high. As long as kerry is ahead in RVs, and very clsoe in LVs, a GOTV operation can win a state for him.
    FL — the poll may be correct, but any reputable pollster knows that we will need to wait at least a week after hurricanes to get a reasonable poll. Just last week, Gallup was showing FL as a 3 point race.

  8. snicker-snack on

    smooth jazz
    in truth, I’m not quite sure the point of your postings here. I’ve never seen you actually rebut any analysis of polls presented here. You do but pooh-pooh. How droll. How uninspired. How utterly boring.

  9. Dana on

    One point to raise for those who have noted that values (not facts or policies) are being used as the anchor for Bush’s claims. There is a word for this type of rhetoric: heresthetics. Here’s a link to help explain the idea: http://rhetorica.net/heresthetics.htm
    Get to know this concept–it’s an unavoidable part of politics these days.

  10. Gabby Hayes on

    The Economist has them tied. So does CSM.
    The few that have large Bush leads also have badly selected samples which are not close to representative, and therefore mean only that those pollsters have not done the one thing they are supposed to do.

  11. Smooth Jazz on

    Oh, Oh, Time to start Dissing Gallup again:
    GWB LEADS KERRY AMONG LVS IN FLA 52-43, IN OH 49-47, IN PA 49-46
    Bush gains ground in Fla
    Poll results are based on telephone interviews with voters in Florida, Ohio and Pennsylvania. (Related link:Gallup Organization)
    Gallup conducted most of its interviews Friday in the regions affected most by Hurricane Jeanne, which struck Florida Saturday night. The poll’s regional representation closely mirrors those of the other three polls conducted by Gallup in Florida this year.
    1. Now, suppose that the presidential election were being held today, and it included John Kerry and John Edwards as the Democratic candidates, George W. Bush and Dick Cheney as the Republican candidates, and Ralph Nader and Peter Camejo as independent candidates. Who would you vote for?
    Q.4/4A (Kerry vs. Bush vs. Nader) Likely Voters Registered Voters
    FL Sep 24-27 OH Sep 25-28 PA Sep 25-28 FL Sep 24-27 OH Sep 25-28 PA Sep 25-28
    John Kerry and John Edwards (D) 43 47 46 44 49 49
    George W. Bush and Dick Cheney (R) 52 49 49 49 46 45
    Ralph Nader and Peter Camejo (I) 1 1 1 2 2 3
    Neither/other/no opinion 4 3 4

  12. Mary on

    One thing not mentioned in poll discussion is the sheer number of questions people are now expected to answer. When a pollster calls you, they tell you how long it will take to ask all those questions. It can take 20 minutes, and you’re informed of that upfront. So, in addition to pollster calls being screened out by answer machines and Caller ID, anyone answering the phone has to consent upfront to answering 20 minutes worth of questions. I wonder if the “old school” polls over past years asked that many questions or took that much time? Maybe if they streamlined the polls and dropped all the questions, and asked for 5 minutes of your time, or 2 minutes, poll response and participation would become greater. As it is, I do notice in looking at the Acrobat files of complete polling reports that the less questions asked, the better Kerry does. Some of the PDF files you can find with all the questions posed in each poll can run over 20 pages, some run only 5 or so. The longer the poll, the more Republican/conservative the results. You have to have a compliant personality type to assent to 20 minutes of questioning, which would be consistent with conservative ideology. So, I think the tie/dead heat polls are probably much closer to reality. I think there may even be a Kerry lead that’s not picked up. If you “norm” the latest Gallup poll for actual voting patterns over the last three elections in terms of party participation, Kerry actually has about a 7 point lead. Similarly, if you look at the CBS News poll that they conducted with NY Times a few weeks ago, the NY Times breakdown showed that of those polled that voted in 2000, 36% voted for Bush, 28% voted for Gore (the rest either didn’t vote or declined to state). Since Gore actually won the popular vote, and this poll showed Bush with an 8 point lead over Kerry, the probable voting pattern in reality would result in that poll being adjusted to a tie.

  13. Sky on

    Mad Hatter, Many of us Dems supported Kerry from the get-go. I think he is a superb candidate and human being. He’s handsome, articulate, smart, a real war hero; if you think he’s getting it bad from the Repubs now, imagine the field day they would have with Dean.

  14. coldeye on

    Latest LA Times poll has Bush +6.
    Latest Harris poll has Bush +2. Interestingly, according to this poll and some others, Bush does better as education level drops; the more educated the voter, the more poorly he does. He lags a whopping 21 points among those with advanced degrees. Translation: Bush makes an appeal based upon emotion and sound bites; when people actually think about what he stands for and what he’s done, they run.

  15. Lawrence on

    From the ever-amusing Fox News web page daily “interactive” poll, today:
    “Do you believe that Saddam Hussein was part of 9/11 attacks on U.S.?”
    Yes 54%
    No 38%
    I did, but no longer think that 2%
    None of the above 6%
    Argh! Argh! Argh! Argh!

  16. jg on

    What I see under-mentioned in these poll discussions is how many polls had Bush up by wide margins in 2000 going into the election. On election night, I and most of my friends were resigned to the fact that Gore was going to lose because of the majority of the polling that last week/weekend showed Bush winning going away. And “but for” the Florida fiasco, screwy polls would have been a bigger post-election issue.
    I have heard that various theories as to how the 2000 race tightened seeminlgy overnight (must of it resting with Bush’ DUI revelation). Yet, I think the race was always close and it was the polls that were off.
    I am both a pessimist and cynic about politics, but this race is still close and I just get the feeling that Kerry will do better than Gore did. I am not sure why, yet these polls do have a demoralizign effect on Democrats and Kerry supporters who don’t read blogs and obsess over these things. I hope they will recall where Gore was in 2000 when they hear these polls and stay the course.

  17. Bel on

    At the end of the day, this race is not about Kerry and this is exactly what Edwards has been saying in recent times. Its really about George Bush. The debates are not about Kerry, they are about George Bush and I hope Kerry can establish this reality from the first word in each debate.
    You know, its really a pity that we can have a nation of sleepers, who at this time in their lives do not quite understand what Bush and his crew are doing. Its a real pity that approximately 50 percent of the electorate can find something credible about Geroge Bush which would merit giving him an opportunity to even compete neck and neck with Kerry. Its a genuine pity that this same percentage of the electorate seems not to know what is really going on in Iraq, the economy, foreign policy and the likes and are therfore willing to even give bush consideration. For his innumberable failures, Bush should not so well rated by this high a percentage of the electorate.
    After all this time, its hard to fathom how so many people can be asleep in the midst of a war on every possible issue that affects the USA.
    At the end of the day however, its really not Kerry’s to lose but the electorate’s to decide. Kerry has travelled the entire landscape in a very planned mode and delivered himself quite well. He has been very systematic in outlining his positions and plans. He has moved from introducing himself to the people, to outlining his programs to attacking and destroying Bush’ credibility.
    There is very little that Kerry can do to affect the intelligence of the electorate. He has done his job and continues to do an excellent one too. Lets hope that he does well in the debates but pushing them aside, an intelligent electorate should be able to see that bush is light weight, incompetent, a miserable failure and needs to be dumped. If they are not smart enough to see the red light at the end of the street, then whats Kerry supposed to do? He can do nothing.
    I still think Kerry is doing fine but I am not sure about the electorate. Nov 3 will tell me more on their level of intellect.

  18. Dana on

    This reply has little to do with polls, but I have to tell coldeye why we should “stick” Kerry with the cleanup: if it is left to Bush, it won’t be cleaned up because it won’t ever be considered a mess. It will wind up under the rug, period. It’s one of those things that makes us Democrats a little nutty–we can get mad at the more reckless among the GOP faithful, but in the end we try to save them from themselves just the same.

  19. wonkie on

    I’ve had that thought more than once. If Bush is re-elected, he will be the most unpopular president in history in just a few years. Any Democrat (but i hope Barack Obama) will be able to when in 2008.
    But in the meantime he’ll get to appoint judges and make Supreme Court nominations.
    Maybe we should be tracking some of the SEnate races.

  20. Paul on

    Wondering where you heard or saw the CNN?GAllop poll showing Kerry up by three in Ohio. Only good news I heard all day and want to see it for myslef.

  21. warp resident on

    YouGov is the new kid on the polling block. They do Internet polls! That’s how they can get the huge sample size. They were dead on in the 2000 election. Internet polling beats the shit out of land lines.

  22. Smooth Jazz on

    I see a number of posters quoting the new Gallup poll for OH; I just saw Judy Woodruff, a Kerry supporter masquerading as a CNN host, give the results with glee – GWB up 2 LVs after being up by 8 in their last OH LV poll.
    Before you all break out the wine glasses to celebrate Kerry’s “comeback” and rush out to spray on Kerry’s tan oitment in sympathy, keep the following in mind:
    1. The poll was taken during GWB’s trip to OH on 9/27, when hundreds of thousands of Reps turned out for his bus tour, including 50,000 people for 1 event near Cincinatti. It’s conceivable that a lot of Reps were missed in a random sample taken on that day.
    2. Gallup may be feely under attack from Libs, and may have jimmied with the numbers to help Kerry to project a sense of fairness. It doesn’t make sense that they would publish a national poll yesterday with GWB by 8, and generate an OH poll a day later that’s tied – Since OH represents as balanced a party ID state as there is in the country.
    I’m not saying the latest OH is wrong; It may be that the Gallup poll from yesterday was wrong, and this OH poll is closer to reality – I’m just saying wild short term swings, absent any major news developments, should be viewed with skepticism.

  23. wilder on

    If the New Economist should be discounted because it has a Kerry bias, then shouldn’t it follow that ABC/Washington Post and Gallup should be tossed out for their Bush bias?

  24. coldeye on

    Interesting thing about YouGov methodology, if I’m reading right, is that it controls somewhat for the modern difficulty of reaching people for polls. Preselecting the sample group would seem to make it easier to reach these people – after all, they agreed to be in the sample group, so they are more likely to answer the phone when you call (or email or however they do it).
    Of course, preselecting creates it’s own challenges, like how to ensure a preselected group remains representative of the population as a whole and does not end up being skewed due to the fact they were preselected for something. Controlling for that, it would seem superior to phone polling which apparently only reaches 20-35% of those called (I think these percentages are actually high).

  25. Bill on

    To bad the Europeans can’t vote. The polls taken of most of the other countries show Kerry with a overwhilming lead over Bush.I also hear that Americans living in Europe are regestering to vote in large numbers and a majority of them are supporting Kerry.

  26. tony on

    First an “oops”….In my last post, second line from the end, drop “don’t”.
    Second ooops…I’d missed the Gallup Ohio info posted above, which pushes things even more away from the idea of Bush running away with things.
    Third, about the ICR poll…The shift has been from Kerry tied to Bush up 10% in LV, but Kerry up 7% to Bush up 7% in RV’s.

  27. coldeye on

    New ICR poll has Bush up by 10 – yikes! Aren’t these the same guys who had Kerry up by seven a month ago?
    You know, if America wants Bush, let America have Bush. He made the mess in Iraq, in the deficit, in the economy, in Medicare. Why should a Democrat get stuck with the cleanup?

  28. tony on

    And third in a series [I hope those first two took!]: Pennsylvania.
    8/26 Kerry +4
    9/12 Bush +1
    9/19 Bush +1
    9/26 Kerry +2
    8/21 Kerry +8
    9/3 Kerry +3
    9/17 Kerry +3
    [Three other organizations only have two polls to track, in different time frames…]
    9/14 Bush +4
    9/26 Kerry +4
    Strategic Vision (GOP)
    8/28 Bush +2
    9/15 Bush +5
    8/26 tie
    9/7 Bush +1
    This gives the following change, from first to last fo the above sequences:
    Zogby Bush +5
    Strategic Vision Bush +3
    Rasmussen Bush +2
    Gallup Bush +1
    Quinnipian Kerry +8
    median: Bush +2
    Most recent polls:
    Strategic Vision (GOP) Bush +5
    TNS Bush +3
    Gallup Bush +1
    Keystone tied
    ARG Kerry +1
    Mason Dixon Kerry +1
    SUSA Kerry +2
    Rasmussen Kerry +2
    Temple Kerry +2
    Opinion Dynamics Kerry +3
    Zogby Kerry +3
    Quinnipiac Kerry +4
    median: Kerry up 1.5%
    So…in this state, Bush has gained a smidge (just like Ohio and Wisconsin), with Kerry still leading by a smidge.
    Across these three states, if you use the same polls to measure change, you don’t see a median change toward Bush of 2-4%. That’s hardly a landslide….And each of these three is within 3% of the other, to boot.

  29. tony on

    I’m going to do the same with Ohio that I just did with Wisconsin.
    Strategic Vision (GOP)
    8/28, Bush +5
    9/12, Bush +10
    9/26, Bush +9
    8/26, Kerry +2
    9/12, tie
    9/26, Bush +2
    8/21, Bush +5.5
    9/3, Bush +10.6
    9/17, Bush +2.9
    So…in terms of change:
    Strategic Vision, Bush +4
    Rasmussen, Bush +4
    Zogby, Kerry +2.6
    median gain for Bush: 4%
    Looking at most recent polls, all post GOP convention:
    Ohio Poll Bush +11
    Strategic Vision (GOP), Bush +9
    Mason Dixon, Bush +8
    Gallup, Bush +8
    Opinion Dynamics, Bush +3
    Zogby, Bush +3
    Survey USA, Bush +3
    Rasmussen, Bush +2
    ARG, Bush +2
    Lake Snell Perry (Dem), tie
    Garin (Dem), Kerry +1
    median Bush lead, 3%
    If we drop the three partisan polls, we still get a median of 3%.
    That seems an awful lot like Wisconsin. Bush has gained a little bit and is a little bit ahead, but there’s plenty of reason to hope.

  30. tony on

    This is better on a state polling thread, but I’ll put it here since it’s the most recent thread.
    Let’s do an apples to apples comparison of polls in Wisconsin. The three most recent polls to come out all have been conducted in late August, early September, and again in the last few days. [My source is race2004.net.]
    8/26, Bush +3
    9/12, Bush +2
    9/28, Bush +3
    Strategic Vision (a Republican group)
    8/28, Bush +1
    9/13, Bush +5
    9/27, Bush +6
    Moore (another GOP group)
    8/23, Bush +2
    9/9, Bush +2
    9/22, Bush +3
    Now where, exactly, is this Bush surge? The Rasmussen and Moorre polls have been stable for a month. The Strategic Vision poll had a boost from the first time period to the second, but has been flat.
    The perception of the Bush surge comes from other polls that came out in particular, 3:
    Badger, with Bush up 14%
    TNS, with Bush up 10%
    Gallup (!), with Bush up 8%.
    But three other polls this month show a different picture, with Zogby giving Kerry a 2.5% lead, Mason Dixon showing Bush up 2%, and ARG showing a tie.
    Zogby continues in the small change vein
    August 21, Kerry up 4.5%
    Sept. 3, Kerry up 2%
    Sept. 17, Kerry up 2.5%
    Gallup takes Bush from 3% up on August 26 to 8% up on Sept. 12.
    Summary…if we look at the change in polls in Wisconsin, from late August to most recent poll, we see:
    Gallup: Bush gains 5%
    Strategic Vision: Bush gains 5%
    Zogby: Bush gains 2.5%
    Moore: Bush gains 1%
    Rasmussen: zero change
    If we look at Bush’s lead in the most recent of each poll, we see:
    Badger +14
    TNS +10
    Gallup +8
    Strategic Vision (GOP) +6
    Moore (GOP) +3
    Rasmussen +3
    Mason Dixon +2
    ARG 0
    Zogby -2.5
    So, the median improvement in Bush’s position, for the five polls with sufficient data, is 2.5%. His median lead, of the 9 polls since the Republican convention, using only the most recent poll, is 3%.
    Certainly I’d prefer to have gained 2.5% to have lost it, and I’d prefer being up 3% to being down 3%, but let’s not give up hope on Wisconsin, folks. The data there seem quite the mirror for nation-wide polling, with a slight Bush gain this past month, slight Bush current lead, and still a very close ballgame.

  31. Mike in MD on

    While I and others on here have had a field day castigating Gallup lately, and with reason, this just crossed my screen: new Gallup poll in Ohio shows the candidates about even. Kerry in fact was ahead in one of the samples (whether it was LV’s or RV’s I forget.) This of course does not comport with Gallup’s recent national results, and in fact shows Kerry doing at least as well as in any other recent Ohio poll.
    Has Gallup changed, and should we? Or is this just a fluke? Stay tuned….

  32. Mad Hatter on

    Come on folks its like watching your favorite football team behind 20 points in the last 2 minutes and hoping for a miracle. Some times you have to recognize when to say when. Every one got on the Kerry Band wagon cause o’l Dean was “TOO UNSTABLE”. You blew it and now you have to live with it. Now be honest, who really likes Senator Kerry
    CLINTON in 2008.

  33. Joe on

    Note that this is a poll of 2773 voters, if I read it right, nearly THREE times the size of Pew poll. This is an enormous sample. Ruy, you might consider weighting the polls by size.
    Note also that CNN just announced Kerry up by 3 (!) in Ohio, once again underscoring the flaw in Gallups national poll.

  34. bruhrabbit on

    Found this on today’s Yahoo news, as people are going back and forth about polls, this article addresses what I believe is the real problem with polling. I can sum the problem with that old saying- opinions are like a—-le, everyone’s got one. I would add to that (as the article below suggests)no matter how uniformed they. I am not sure in a democracy how we can even have a debate out Republican and Democratic ideas with this kind of electorate:
    More than half of those polled by the National Annenberg Election Survey didn’t know President Bush (news – web sites) alone favors allowing private investments of some Social Security (news – web sites) money. Nearly as many didn’t know that only Democratic candidate John Kerry (news – web sites) proposes getting rid of tax breaks for the overseas profits of U.S. companies.
    Importing drugs from Canada? That’s a Kerry issue, but nearly half either didn’t know or thought Bush also supported changing federal law to allow for drug imports from Canada.
    Making abortions more difficult to obtain? Nearly one-third of those surveyed didn’t know Bush alone supports more restrictions on abortion.
    Eliminating the tax on estates? Two-thirds didn’t know that’s a Bush proposal.
    After two years of presidential campaigning and hundreds of millions of dollars in political ads, many voters remained clueless about those and other policies, according to the survey. Annenberg analyst Kate Kenski blamed the candidates for not stressing their points of view and the news media for focusing on character assessments and the race itself.
    “It’s disappointing that people don’t know where the candidates stand, given how much money’s been spent on the campaigns,” said Kenski, a senior research analyst. “In the absence of good information, voters guess and often guess incorrectly.”
    The poll of 1,189 adults was taken from Sept. 21-26 and has a margin of sampling error of plus or minus 3 percentage points.
    End Quote.
    No matter if you are left or right this has got to be considered as problematic- maybe the problem isn’t Bush or Kerry- maybe its the deliberately ignorant electorate.

  35. Armando on

    TedL call this poll an outlier.
    Actually, there seem to be two grouping of polls –
    This poll, CSMpoll, Fox, Zogby, Rasmussen, WSJ and some others show result between a 1 point Kerry lead to a 4 pt lead for Bush.
    Gallup, CBS, Time and some others show Bush leads of 5-8 points.
    The difference, if I may be so bold is likely Party ID weighting, or not weighting.
    Therein lies the dispute – of ypou think turnout will be similar to 2000, its a very close race, if you think there will be higher GOP turnout, then Bush has a lead.

  36. Joshua on

    There are so many polls with so many different results, that it is hard to figure out where Kerry stands at this point. This poll seems like good news, but who really knows where the country is actually at.

  37. N Winlund on

    ‘Voter modeling’ & ‘likely voter methodology’ are two terms I’m interested in learning more about, specifically their metrics: what they are and how they work.
    When polls are released do they show the questions asked? Who signs off for or authorizes the questions to be asked? Is there review of the standard review of questions before they’re sent to respondents? These investigative questions are important to ask because the answers determine in part how questions are asked in polls, what’s asked when, and why the public sees what it sees.
    “Do you favor John Kerry or George Bush?” is a first tier question with a binary choice, either Kerry or Bush. Questions like: “Do you favor John Kerry or George Bush [while the economy has lost jobs], [while Iraq is worse off than it was when President Bush declared mission accomplished].. supplement the core question.
    Journalists have reported or framed supplemented questions as core questions and nothing else: “When voters were asked what they thought of the presidential candidates..” but voters were actually asked about jobs and Iraq this misleads the context and is a transparency issue in polling and reporting that must be fixed by initiating more community based media projects with local production and authoring solutions.

  38. Ben Ross on

    Why so much variability between polls?
    Here is one possible explanation. It goes (as with my comment posted yesterday on the Gallup thread concerning the Republican bias of education-weighted samples) to the mathematics of weighting in polls.
    The polls weight on several variables to reproduce the demographics of the U.S. population. These variables are correlated — e.g., Hispanics have a different age distribution and different distribution of educations than non-Hispanics. Thus if you increase (let us say) the weighting of Hispanic respondents relative to non-Hispanics to reflect their underrepresentation in the sample, you are also increasing the average weight assigned to younger and less educated respondents. If the original sample had the right number of young respondents, the sample that has been weighted to give the right ethnicity will have the wrong age distribution. To fix this, you adjust the age weighting. That puts the ethnic distribution out of whack, so you have to go back and change the ethnic weights again. What you wind up doing is a round robin where you go back and forth among all the weighted variables. You stop when you feel you are close enough to the right answer, or when you can’t get any closer.
    This is an example of a very common problem in numerical methods, an optimization search. There is a large literature on these problems, which I don’t keep up with. But one of the best known pitfalls of these searches is something I learned about (the hard way) in grad school. When there are too many variables, especially when the data contain statistical noise, this turns out to be what is called mathematically a poorly posed problem.
    What this means is that there is a wide range of numerical solutions which are close to the (unknown) exact solution. Which of these solutions the search winds up at will vary substantially depending on the details of how the search is carried out. (In what order to you adjust the variables, how large an adjustment do you make in each step, etc.)
    In other words, when determining weights is a poorly posed problem, two pollsters who collected the same data, and used the same weighting variables, might assign substantially different weights because one of them adjusted ethnicity first, and then adjusted education, while the other adjusted age first, and then adjusted education, getting to ethnicity only at the end of each iteration.
    For this reason (among others) you pay a penalty when you weight too many variables. You have to trade that penalty off against the reduction in random and systematic error that you gain. There are not (in practical cases) going to be mathematical formulas that tell you what to do — it’s a matter of judgement whether to adjust one more variable. My guess (and from where I sit it’s only a guess, but I’d be very surprised if it was wrong) is that weighting 5 or 6 variables in a 1000-respondent poll is poorly posed.
    I have not looked at journal literature on public opinion polling, but in what I have seen on the internet there does not seem to be a high level of mathematical sophistication. Like many areas of numerical analysis, these weighting procedures are much trickier than they seem at first glance and there are severe pitfalls for the unwary.

  39. pempel on

    Good to be even given the numbers from other polls, but two worries–1. Kerry led the last poll even if only by a point; and 2. the ‘distrust’ number on Bush fell dramatically from 58 to 46%. There’s still lots of volatility, but these points suggest voters are starting to become more tolerant of Bush’s reelection and less willing to dump him for Kerry. Kerry has to begin turning this around quickly before opinions congeal. T.J.

  40. rwc3 on

    Does the YouGov poll weight by party ID, does anyone know? If not, does anyone know what the party ID breakdown is for their poll (or recent polls)?

  41. wilder on

    Pretty nice…
    Also, I have to say that maybe we should be heartened by the pundits’ proclaimation of Kerry’s premature death. When they were prematurely predicting his victory back in July, I knew we were in trouble; and sure enough, the Swift Boats and the “successful” Republican convention arrived. Which means that now a Kerry surge could be in the stars.


Leave a Reply to Mad Hatter Cancel reply

Your email address will not be published. Required fields are marked *

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.