« The "How Can Gallup........" Game | Main | Even MoveOn.org Is Playing the "How Can Gallup......." Game! »

The "How Can Gallup......." Game (Continued)

Well, our correspondent, Alan Abramowitz, has been getting into the spirt of the "How Can Gallup...." game, so I thought I'd share some of the fun he's been having with DR readers:

It's hard to know where to begin when it comes to all of the preposterous results in [the new Gallup poll]. First of all, they've got about a 10 point Republican advantage in party ID among registered voters. I am guessing that this is one of the largest Republican party ID leads in the history of the Gallup Poll. So according to Gallup, what's happened since the Republican convention is something on the order of the New Deal realignment in reverse. If you reweight their data based on the partisan composition of the 2000 exit poll, you get something like Bush 48, Kerry 47.

Going along with the ridiculous party ID results, they've got Bush now leading by 15 points in the Midwest and by 21 points in the West. For Bush to be leading by 21 points in the West he'd have to actually be leading in California by about 10 points and cleaning up in Washington and Oregon as well. This is totally out of line with recent independent state polls showing Kerry leading in California by 15, leading in Washington, and running about even or slightly ahead in Oregon. According to this poll, Bush is doing better in the West as a whole than he is in Montana, where the most recent independent poll has him ahead by only 18. The Midwest result is totally out of line with the most recent independent polls in Illinois (Kerry +15) and Ohio (Bush +3).

Gallup's own state polls are totally out of line with this result. How can Bush be 2 points ahead among RVs in states like Florida and Nevada but 13 ahead among RVs in the entire country?

How indeed? Anyway, be sure to play "How Can Gallup...." at home. Don't let Alan and I have all the fun!

Comments

Anyone know how Rasmussen did in the 2000 prez election?

I firmly believe the reason many and probably most of the republican-favored polls are so "out of wack" from reality is purposeful on their part. There will be wide-spread vote-corruption. It will be in diverse forms...such as what is happening in Ohio (see Dailykos blog) where the Sec of State is trying his damndest to eliminate as many of the newly registered voters through position-of-power privlilege. Absolutely un-American! Why would one American especially one in a position of political power instigate actions that prevent other Americans...especially those less economically able non-the-less citizens... from taking part in perhaps the greatest privilege of being an American citizen...voting to elect our government representatives? A paid assassin?

Gallup is engaged in a see-thru attempt to give a Bush victory an air of inevitability. Then when they steal the election in Florida, or Ohio, or wherever they can, people will be thinking, "well, he WAS ahead." Not to mention the half wits who will vote for Bush just because they think he's winning anyway.

Shouldn't Gallup polls come with a disclaimer detailing Gallup's long time support for the GOP?

How can Gallup do it?

Bad polling.

And yet, it is all over the news, and spoken as a given.

"Well, Kerry is so far behind, he's really got to win the debates ...." they say.

Weighting the likely or registered voter sample based on the last election's exit polls is, I admit, risky business. I don't think anyone expects voter preference and turnout to remain the same for four--or even two--years.

That said, the notion of that large a GOP lead among the general electorate strikes me as preposterous--unless Democratic turnout is extraordinarily (and unforgivably) low. I think the recent Time poll has it closest to right (Bush up by about 4)--and two weeks ago I never thought I'd be praising Time's poll.

It is possible that Bush leads in the national popular vote by running the table in certain states (especially in the Rocky Mountains, Great Plains, and South) which are not being seriously contested, and also by pulling respectable, but not really competitive, numbers in large Democratic states (such as California, New York, and Illinois) where he may be running better than last time but is well short of winning. Even thinking about this, Gallup's RV total still seems off (and why is it closer among likely voters, contradicting earlier results?) This raises the interesting prospect of Bush winning the popular vote but Kerry carrying the electoral vote based on narrow margins in marginal states. There would be a certain poetic justice to that in light of the 2000 result, and I wonder if the electoral college would then be changed or abolished since both parties would have been burned by it.

In any event, the EC's presence means that state by state polls are more useful for determining how the election will ultimately go.

How did Rasmussen do in the 2000 prez election?
Any of you little wranglers know and be willing to do sharezees with that info?

Perhaps you meant "Don't let Alan and ME have all the fun"? Sorry, pet peeve, couldn't let it go by! Keep up the good work.

Fine, the Gallop pole is whacked. But the results in the latest ABCNews/Washington Post Poll aren't that much different.

Please see my blog.
http://www.theleftcoaster.com/

The Gallup Poll that was fed to CNN and USA Today yesterday, as well as the rest of the media is seriously flawed in that the likely voter sample they used assumed a 12% advantage for the GOP (43%-31%) that doesn't exist now and has never existed in the last three elections. The Democratic advantage from the 2000 exit polls was 39% to 35% over the GOP. Yet this flawed poll showed a narrowing Bush lead from their similarly flawed poll of two weeks ago. So if a poll with an insupportable GOP bias of 12% in its likely voter sample, shows an 8% Bush lead amongst likely voters when a poll they used two weeks ago with a 7% GOP bias showed a 13% Bush lead with likely voters, then how can anyone not conclude that Kerry is doing much better than Gallup would believe?

Steve Soto
theleftcoaster.com

See this entry from today's "Left Coaster" blog:

Tuesday :: Sep 28, 2004
Gallup Is At It Again - Yesterday's National Poll Had 12% GOP Bias

Gallup has done it again. After supplying CNN and USA Today with a poll two weeks ago that showed a double-digit Bush lead amongst likely voters that turned out to have a significant bias in its sample favoring the GOP, Gallup did it again yesterday.

Except that yesterday, they not only did it again, they apparently felt that a 7% GOP bias wasn't good enough. So they perpetrated the same fraud upon the media (including their partners CNN and USAT) and voters and this time used a 12% GOP bias in their likely voter screen. I kid you not.

Here is the text from the email I got from Gallup this morning outlining the party ID breakdown in their likely voter samples from their two most recent national polls:

Likely Voter Sample Party IDs Poll of September 13-15
Reflected Bush Winning by 55%-42%

Total Sample: 767
GOP: 305 (40%)
Dem: 253 (33%)
Ind: 208 (28%)

Likely Voter Sample Party IDs Poll of September 24-26
Reflected Bush Winning by 52%-44%

Total Sample: 758
GOP: 328 (43%)
Dem: 236 (31%)
Ind: 189 (25%)

Looking at this, again I have a simple question: how can anyone, especially USA Today and CNN, let alone the rest of the media take a Gallup national poll seriously when Gallup knowingly puts a poll out there for consumption with a 12% GOP bias in its likely voter sample that everyone knows does not exist in the country today or at any time in the last three presidential elections?

Yet this flawed poll showed a narrowing Bush lead from their similarly flawed poll of two weeks ago. So if a poll with an unsupportable GOP bias of 12% in its likely voter sample, shows an 8% Bush lead amongst likely voters when a poll they used two weeks ago with a 7% GOP bias showed a 13% Bush lead with likely voters, then how can anyone not conclude that Kerry is doing much better than Gallup would have you believe?

By presenting these polls with this kind of bias, and then ensuring through CNN and USA Today the farthest possible media saturation, why is Gallup not guilty of engaging in a political disinformation campaign?
Steve Soto :: 7:25 AM :: Comments (4) :: TrackBack (0)
Other blogs commenting on this post

Time to start naming names. WHo are these guys at Gallup and what are their backgrounds? Who bought them out. Are their kids going to Liberty U or interning in D. Haserts office? It appears these guys have been bought and paid for somehow.

FYI Bush is going to get crushed in CA. The best indicator of this is the invisiblity of the Gov. Kerry is going to outpoll Gore in CA and along the west coast.

As Steve Soto points out, 12% is ridiculous. Why does Gallup do it? Well, comments below may explain what we are in for in the polling game this year.

From Electoral Vote Predictor (www.electoral-vote.com), News from the Votemaster:

Some bad news for the polling business. Strategic Vision (R) has a new poll in Ohio showing Bush ahead 52% to 43% there. However, there is also a Lake Snell Perry (D) poll showing the race there to be an exact tie, with both candidates at 46%. It is becoming increasingly clear that the pollsters are producing the results that the people paying the bills want to hear. Even pollsters who were once thought to be above suspicion are now suspicious. Gallup, for example, is now normalizing its samples to include 40% Republicans, even though the 2000 exit polls showed the partisan distribution to be 39% Democratic, 35% Republican. There is scant evidence that the underlying partisan distribution has changed much since then.

http://www.cnn.com/feedback/forms/form1.html?21

put up think link so people can contact them and let them know what is up!!!

Again, it bears repeating: The "mainstream" media do not want people like us to watch/read them. We are wasting our lives and our time bitching about their behavior, because they don't care.

They do not want us to vote or be heard. We have seen time and again large media conglomerates making programming decisions on an ideological, not business, basis. A good example of this is GE/NBC's cutting of Phil Donahue despite the fact that he had MSGOP's highest ratings at the time (admittedly a dubious honor).

The polls are strictly for the internal market of media and the narrow slice of white American consumers, i.e., Bush supporters, their advertisers want to appeal to.

The "mainstream" media is simply the largest partisan echo chamber among many. There is no reason to go on and on bemoaning this fact. I say if they don't want us to watch, then why should we?

Gallup might be guilty of partisanship in its insistence that Party ID not be weighted, despite the absurd overrepresentation of Republicans. Yet equally plausible is that they are simply being rigid, and insisting on imposing standards appropriate only for a bygone time.

The most telling statistic I have encountered about trends in polling is this: a generation ago, 70-80% of people called on the phone would answer pollsters questions; today, that proportion is 20-35%.

The deep impact of this remarkable change in voter habits is that sampling biases that could easily be hidden when the vast majority of people answered questions may now be several times as large in the small pool of voters left who actually answer the questions.

If, say, only 25% of voters answer a pollster's questions, how likely is it that that pool is not in some way rather deviant, with personality characteristics and habits somewhat out of the mainstream? How plausible is it that those characteristics and habits will have ZERO effect on their general political inclinations?

Think of it in terms of a most basic parameter, adaptibility to new ways of doing things. Given that the number of people who are answering the questions is going down dramatically, it's obvious on its face that telephone etiquette and habits ARE indeed changing across the population. Who will be among the residue of those NOT changing those habits? Well, by standard definition, more conservative people -- conservative culturally in any case.

Are we to believe that people who are more conservative in this cultural sense are in no way more likely to be conservative in the strictly political sense? And, if they are conservative in the political sense, are they not more likely to be Republicans?

I don't think Gallup wants to come to terms with this basic change in the circumstances in polling. But that may very well be because, ironically, it is far too conservative in the cultural sense itself (and not necessarily because its conservative in the political sense).

I've been following Ruy since his days over at TomPaine. It's great to see him continuing his thing over at this site.

Let's face it, though; as much as we'd all like to see Bush lose and become the GOP pet goat, he is ahead. The Gallup poll clearly has some issues, but other polls are also consistently showing him ahead. If you average them out, he's probably up by five. This is a very weak position for an incumbent "war-time" president (I have a hard time saying that without inducing a gag reflex), but it's still enough to comfortably carry the electoral college.

Bush and Rove will surely ratchet up the terrorism fear level in October, with many public displays of uniforms and color alerts. Paul Krugman had a good article a while back about the collective denial affect we are seeing regarding the "war" on terrorism and the Iraq horror-that-we-do-not-speak-its-name; people just don't want to believe their leader is a lying, incompetent charletan who is leading them down the road to ruin. See Germany circa 1943.

The Republicans so slavishly backing these con men had better watch what they wish for. A Bush re-election may only serve to hasten the internal rot that has infected the Republican party; when this corrupt bunch of extemists fall, they are going to fall spectacularly.

Quoted from the post: "Going along with the ridiculous party ID results, they've got Bush now leading by 15 points in the Midwest and by 21 points in the West."

I have looked everywhere I have access to find the back-up to this statement; could you please give a link or cite it.

I don't doubt the criticism on the Party ID, but I'd feel better in seeing the data from Gallup regarding regional breakdown.

Sleepless asks a question I've wondered about too, as I've been folllowing the Rasmussen poll for several months and have referred to his results a couple of times on this weblog. So I did a quick google and found something quite fascinating: in 2000 Scott Rasmussen consistently and assertively overestimated Bush support, by a wide margin. His final poll gave Bush a 9 point lead!
Check out the following. They make interesting reading this year.
brian.cannell.com/articles/2000/11/000015.html and
www.worldnetdaily.com/news/archives.asp?AUTHOR_ID-14&PAGE=2

BTW, Rasmussen's Daily Snapshot today again has Bush leading by less than 2%. Maybe he upgraded his methodology after being burned in 2000?

In general, I think there should be a Truth in Forecasting law requiring pollsters to indicate their track record every time they issue a poll. It should apply to meteorologists too. But that's another story.

This might be off-topic, but Cooks' political report:

http://www.cookpolitical.com/races/report_pdfs/2004_electoral_sept27.pdf

has things neck and neck in the electoral college. He has Bush leading 208-207 in states that are at least leaning, with 11 states as toss-ups. Colorado, Florida, Nevada, New Hampshire, Ohio, and West Virginia are tossups that went for Bush last time. Iowa, Minnesota, New Mexico, Pennsylvania, and Wisconsin are tossups that went for Gore last time.

Repeating my mantra...this will be close. Go out and do what you can to drum up support for Kerry, particularly in those 11 states listed above.

Two questions, and forgive me if they've been answered elsewhere:

1) is it unquestionably a fact that Gallup is run by an evangelical who has vowed to do what it takes to get Bush elected?

2) I've been seeing various comments on this site and elsewhere stating that the Gallup stuff is part of a broader and obviously more insidious plot to steal the election. My question is (a) is there any hard evidence of such a coordinated effort/plot, or is it just "obvious" based on what these people are caoable of (and are already doing in Ohio, FLA, etc.)? (b) assuming that such an effort is underway (the effort to supress the vote in key states, steal the election, have polls in place which "show" Bush ahead, etc.), what is the remedy?

I guess another way to say this is: even though this particular conspiracy theory smells a little paranoid to me, it is in fact making me paranoid, and I would like to know what reasonable people think about it. Anything to allay my paranoia is welcome.

I will even start the ideal response for you. It should begin, "even if the Right were conspiring to do exactly [the scenario mentioned above], it would never work because..."

It seems pointless to dismiss any poll you dislike. You may think some of these organizations are skewing samples to help Bush, but would they really destroy their credibility for such short term gain? After all, there is life after Nov 2nd and these organizations do need to be taken seriously in the future.

I absolutely believe that substantially more voters could ID themselves as Republican in this election. Party ID may be sacrosanct to some, but its just a name to others. If voters are worried about terror and see Bush as more aggressive in WOT they may consider themselves more "Republican" in ID. In the next election cycle, they may declare themselves more "Democratic" if different issues take precedence.

I do agree that we are in danger of the mindset that "Bush will win" becoming a self fulfilling prophesy and depressing turnout. No one is going to go vote if they think the outcome is already decided.

Aaron Brown sure had no compunction(s?) about pimping this poll - and he had an almost Chimp-like smirk whilst pimping. I wrote him at newsnight@cnn.com, but I don't think it will matter.

Does anyone have any constructive ideas on how to make news organizations more responsible with how they report these polls?

If Gallup or any other poll shows Bush ahead 13pts the day before the election and Kerry actually ends up winning, does anyone think anyone will be held accountable for these half-assed polls?

My wife, an actual practitioner, claims that Gallup and Gallup-alike surveys do not measure what we intuitively understand as "Party ID" with their questions on that subject. She cautions against reweighting to adjust their results. Further, she claims you should pretty much ignore any one- or three-day poll results, since they will strongly skew for people who are easy to survey during that period -- here, whatever partisan group happens to be in a good mood about their candidate.

For more than that, I'd have to get her to post.

I used to wonder why Gallup kept coming out with these pro-GOP polls...

Then I remembered: Anti-union employers have been using Gallup heavily in the last several years to run surveys of employees prior to union organizing drives.

In many ways, Gallup is becoming the Peter Hart -- of the right-wing.

Just speculation, but I wonder if Gallup doesn't have a vested interest now in the outcome of the election.

Here's one possible reason for polls that have an unrealistically high number of Republicans in their sample. I'm afraid this is going to be a little bit mathematical -- please bear with me if you're interested.

I note that the CBS/NY Times poll, and apparently many others, weight respondents for education level, while not weighting for income. Under certain circumstances, which seem very likely to me, this effort at improving the representativeness of the sample would actually increase a pre-existing bias toward oversampling Republicans.

Consider this model - some of these propositions are clearly true and all seem plausible:

1) Education & income correlated
2) Higher income voters more likely to vote for Bush
3) After controlling for income, more educated persons more likely to vote for Kerry
4) Likelihood to respond to poll positively correlated with income
5) After controlling for income, likelihood to respond to poll uncorrelated with education

If you weight this poll for education, you will increase the weight of high-income, low-education Bush voters who are overrepresented in the sample and decrease the weight of low-income, high-education Kerry voters who are underrepresented in the sample.

This effect does not require (4) and (5) to be true in the strong form stated above. All that is necessary is that likelihood of responding to the poll is correlated with income when you control for education.
This seems to me to be very likely to be true.

I hope everyone got a chance to see MoveOn.org's great ad criticizing GAllop in the NYTimes (Spet. 28) , page A5

or at:

http://www.moveon.org/content/pdfs/Final-Gallup-Ad.pdf

Gallup has very posh offices in DC, employees PhDs by the score and does millions in gubmint contracts.
Newport donates big bucks to the GOP.
That is all ye need to know about Gallup.

Fun and games with Gallup is all well and good but how about some analysis of this Election, Ruy???

Some weeks ago, refering to your analyses of poll internals and their favorability for Kerry, I noted that at some point the internals meet the externals, ie they converge with the horse race

Now I have too much real world work to do but I noticed an MSNBC piece on the internals of the ABC/WaPo Poll and in all key issue areas, they now favor Bush...


I'd appreciate it if you would pick up that thread again..

Thanks

I'm not, nor ever have been of a conspiratorial mindset, but the totality of circumstances allows for at least a tenative inference that Gallup, aided and abetted by cable news in particular, is trying to influence the election. Gallup and his PhDs have GOT to know that their polls are outliers by a wide margin and have GOT to know that it's their methodology.

My sense is that we have some really bright persons in this forum. They know that groups with a vested interest in some outcome need not sit in a room and collude so as to meet some legalistic definition of conspiracy. Such collusions or cabals can be totally unspoken.
I'm making -- and perhaps overstating -- this point because an almost daily RNC "talking point" is that complaints of this type are to be mocked and dismissed as "nut conspiracy theories". Just listen to Tucker Carlson read that part of the RNC script in his best mocking, school girl tone. And it's relatively easy to do because such collusions most likely are unspoken and so the complainer gets put on the defensive. No one wants to be seen as a nut.

Speaking for myself only, I'm persuaded at this point that Gallup is trying to influence the election with the aid of headlines in those "liberal" outlets like CNN, Fox, WaPo, NYT et al.
It's utterly shameless, but what's a mother to do except hope we'll have the last laugh on the morning of Nov 3?
On the other hand, maybe another 4 yrs of this awful administration will work to give the electorate a bellyfull of the GOP so as to put them back into minority party status for a long time.

As you all know, most polls have shown the race within 2 points. Rasmussen's daily tracking poll has (except for a string of a few days last week) been within this range for weeks, as have all other national polls other than those done by Gallu or the broadcast networks. Even Fox News has it as a 2 point race. A new Investor's Business Daily poll released today has Kerry up by 1.

Investor's Business Daily/Christian Science Monitor/TIPP poll that just came out:

KERRY LEADING AMONG LVs:

Among likely voters: (9/22-27/04)

Bush Kerry Unsure
45% 46% 9%

Gallup is also selecting the people participating in the second debate.

Did anyone read Gallup's Editor blog on this issue? They claim that party ID is a flexible identity. In other words, people may identify themselves as Democrats one day and Republican the next. They said that this is common. Any have any more info on that? Here is the link

http://www.gallup.com/poll/content/?ci=12292

Just to show how out-of-whack some of these polls are, TIPP/Investors Business Daily has a new one out today showing Kerry ahead by 1. Finally, a poll that seems to be somewhat reflective of reality!

http://pollingreport.com/

Interesting info on the Gallup sample at www.theleftcoaster today. It seems they used 43% Repubs and 31% Dems. When I adjusted for the 2000 39% Dems and 35% Repubs 2000 turnout, I figured they overrepresented Repubs by 23% and underreped Dems by 20.5%. I got an adjusted poll of Kerry 53%, Bush 40%. Something must be wrong with my adjustment unless Kerry had a big push lately with the Iraq issue, but I like my results :-)

I looked for Newport on Fundrace.org and Opensecrets.org, using both his name and Gallup. Nothing came up on either, so if he's giving to the GOP, could someone please let me know where to find that info - and still waiting for a cite re poll internals/regional breakouts.

OK, Forget Gallup & Wash Post & CBS - They're all biased in favor of President Bush anyway (Tongue firmly planted in cheek). But how about this from Pew. Pew & it's head, Andrew Kohut, has historically been Dem leaning, and you guys were touting them last week as a credible poll.

I take it you'll will have an explanation for the most current Pew results too. Maybe it's not time to panic yet, but heaven forbid if ARG & Zogby, the ultimate Lib polls, start showing these same trends - Then, I suspect, it will be time to panic.

GWB 48%, Kerry 40%

http://people-press.org/reports/display.php3?ReportID=226


Pew Research Center for the People & the Press
Survey Reports

Bush Margin Widens Again, Despite Vulnerabilities

Released: September 28, 2004

Summary of Findings

George W. Bush has reopened a significant lead over challenger John Kerry over the past week, even as voters express less confidence in the president on Iraq and he continues to trail Kerry on the economy. Two successive nationwide surveys of nearly 1,000 registered voters each show Bush's margin over Kerry growing steadily since mid-September (Sept. 11-14), when the two men were tied at 6%-46%. Bush's slight 45%-42% advantage in the Sept. 17-21 survey has grown to 48%-40% in the current poll (Sept. 22-26).

The poll finds that Bush's gains in support are being driven more by perceptions of Kerry's weakness especially on leadership and other personal traits than by improved opinions of Bush. Fewer voters favor Bush over Kerry on handling Iraq than did so earlier this month (46% now, 52% Sept. 11-13). But Kerry's rating in the head-to-head evaluation on Iraq is no higher (38% now, 40% then). The Democratic challenger continues to inspire more confidence than Bush with regard to improving the economy, which 60% of Americans believe is in only fair or poor shape. But even here, the percentage favoring Kerry has not increased since the Sept. 11-13 survey (46% now, 47% then).

Despite Bush's lukewarm evaluations on the issues, he maintains a significant advantage on most personal traits. Kerry has slipped slightly on some key personal assessments, including honesty and empathy. Overall, 32% of voters say the phrase "honest and truthful" better describes Kerry than Bush, down from 36% a few weeks ago (Sept. 11-13). Bush's rating in that period is unchanged at 41%.

A similar pattern is evident in comparative evaluations of the candidates' empathy: 42% say the phrase "cares about people like me" better describes Kerry, compared with 46% earlier this month. Bush's rating on this measure also has not changed (38%).

Kerry's image weakness is most evident on questions of leadership and his willingness to take on unpopular positions. Just 28% say the phrase "strong leader" better fits Kerry, virtually unchanged from earlier this month, while 54% say it applies to Bush. And 23% believe Kerry takes on unpopular stances, compared with 63% who say that about Bush. In both cases, Bush's ratings are down slightly from earlier this month, but Kerry has gained no ground.

Bush is telling O'Reilly that all the options are on the table in Iran. Sort of like all the options were on the table in Iraq. Scary.

I didn't read all the comments, so pardon me if this is a repetition, but I was wondering if Ruy et. al. have thought to check Gallup's party ID in other races. Do they have party ID models for each state? If so, what percentage of voters in CA. do they think are Republicans, etc? A careful examination of how Gallup is polling other races might reveal whether we are seeing poor modeling or a real ethical problem, on Gallup's part.

To Ruy:

Please post something about the ABC/Washpost poll and the AP poll, which are the other big polls with large media distributions which show Bush ahead by more than 4-5.

Bob Novak just used the ABC/Washpost poll to say Bush is way ahead. He made a point of saying it was not just the Gallup poll which he aknowledged was in dispute. But there are other polls which agree with that direction. In particular he noted the Bush was at 51% in that poll.

So our debunking has done its job. Is there more debunking to these other polls? If there is and it is solid and true then that's good.

However I not a troll; and I participated in some of the initial poll dissection in the blogoshere since the RNC. I want us to be realistic. So I think one important question to ask is - historically what have been the largest swings in party ID in both short and long terms?

Hoping to hear from you asap.


I

Andrew Kohut is quoted as saying the race is highly volatile..

So too Ron Fournier of AP

This is what I would like Ruy to address if he has the time..

Wavering Voters Anti-Iraq, Wary of Kerry

7 minutes ago Top Stories - AP

By RON FOURNIER, AP Political Writer

WASHINGTON - In an election where most voters have already chosen sides, the presidency could be decided by a small slice of America in the mushy middle wavering voters who are more likely than others to question President Bush (news - web sites)'s honesty and think the war in Iraq (news - web sites) was a mistake.


AP Photo


Latest headlines:
Wavering Voters Anti-Iraq, Wary of Kerry
AP - 7 minutes ago

Supreme Court Refuses to Put Nader on Oregon Ballot
Reuters - 10 minutes ago

Soros Steps Up Efforts to Oust Bush
AP - 34 minutes ago

All Election Coverage

An Associated Press study of 1,329 "persuadable" voters, conducted by Knowledge Networks in advance of the presidential debates, suggests these people are deeply conflicted about change in the White House. While they have problems with Bush, they also have doubts about Democratic Sen. John Kerry (news - web sites)'s leadership skills and believe Bush is best suited to protect the nation.


One in every five voters is persuadable including about 5 percent who tell pollsters they don't know who will get their vote and about 15 percent who say they are leaning toward one candidate but could switch to another. In past elections, as much as a third fit that description, but most of the nation was quick to pick sides this year in the aftermath of the disputed 2000 election.


It's possible many persuadable voters will stay home Nov. 2 out of frustration with their choices, but there are enough of them floating in the political center to alter the race for the White House.


"I don't want to see Bush get in, but I don't want to vote Kerry just to keep Bush out," said Grace Elliott, a 70-year-old retiree from Portland, Ore. She opposes the president's conduct of the war but says of Kerry: "He just makes me feel uneasy."

How can Gallup do it?

I'm reminded of the helicopter gunner in FULL IN METAL JACKET, when responding to the question "How can you shoot women and children?"

"It's easy," he replied. "You just don't lead 'em as much."

How can Gallup do it?

They are representing their interests, which are heavily big business. Gallup went from the pure polling entity Dr. Gallup built to something completely different, all in the past 20 years.

How can Gallup do it?

By selling out, by going HUGE, by getting into partnerships or joint ventures with 200 other organizations. They are guns for hire in the business community, far removed from the the commitment to fairly call a presidential election.

No, their loyalty is to all those checks they cash every month, checks from big business, checks that won't be there if they are perceived as "too liberal." This action, like the 2000 election call by Gallup, is a marketing tool for Gallup.

They are hired guns, and selling their NAME is what they are about. Being publicly identified with showing a huge Bush lead is their way of being the Fox News of polling.

Gallup is rolling the dice, getting exchatalogical about this election. Lots of evangelicals think this election is of biblical significance -- as in let's Bring on the rapture! But once it becomes clear (after November 2) that
a) Gallup's predictions were WAY off and
b) The spaceships aren't landing to take the believers up to IO or wherever, Gallup may go the way of Arthur Anderson. Without credibility, they've got nothing to sell.
-grebberg

p.s. this may violate your new comment policy, but Ruy, come on, it's "...don't let Alan and ME," not "...don't let Alan and I."

Let I entertain you! Let I free!

Party ID is stable. We don't have to go back to 2000 and assume that things have not changed.

this excellant article looks at party ID since 1987
http://www.mydd.com/story/2004/9/18/454/00913#

Pew and Harris both did surverys to get at party ID . Pew did one in June and July 04 that had 19,000 people in their sample (.05%MOE) and Harris did one in 2003 that had 6000 (1% MOE). VERY large smaple sizes, nad note that Pew did theirs in an election cylce.

the results for 2001-2004. note that in the Pew 9/11 cause a change in party id but that by 2004 it was back to 2000 levels

pew

democrats repub indep gap
2001
pre 9/11 28 35 37 7
post 9/11 31 32 37 1
2002 30 31 39 1
2003 30 31 39 1
2004 29 33 38 4


harris

2003 28 33 5
2002 31 34 3
2001 29 36 5

So the best surveys we have of party Id with a combined sampole size of 25,000 people (that is an extremely large sample 25x bigger than sample in most current polls show that Democrats still outnuber republicans.

er... that's "eschatalogical." (I'm thinking about glass houses and my just comeuppance for throwing a stone!)

More on why party ID does change significantly over short periods of time:

http://people-press.org/commentary/display.php3?AnalysisID=97

This is a report by Pew, and nobody has claimed they are biased. Now matter how much people may deny it, a double-digit change in party ID over a period of months is NOT historically rare. In fact, Ruy Teixeira was claiming the same thing a few months ago when the Dems were ahead, and conveniently changed his opinion once the Republican convention proved his original opinion was correct (http://mysterypollster.typepad.com/). The liberal blogs are consistently opportunistic when it comes to the facts about polling.

To me, the comments by frankly0 concerning voter response tendencies in the era of radically changing telephone habits, represent the most interesting and important line of inquiry. (The impact of changing commmunications media on society is also my area of specialty.) Although we don't have much beyond anecdotes and speculation, I think we can dissect this issue a little further:

The group most likely NOT to adopt new telephone tools and habits, such as cell phones, answering machines and Caller ID, is the elderly. The group most prominently adopting them is the young. Contrary to some posts, cell phone numbers typically ARE included in the universe of numbers called by pollsters, but the chance of getting a live answer is certainly much lower than with landlines. For these reasons, I would strongly expect that actual voter samples for most polls are skewed heavily toward older respondents, and away from the youngest demographic. If they are then re-weighted to reflect anticipated age distribution of the electorate, the potential variability increases, since you're extrapolating from a much smaller sample (young voters).

Take an example of a 1,000 RV poll. Assume that, for the reasons stated, they end up including, say 300 voters over age 50, and only 100 under age 25. If the more appropriate alignment should be 200 in each group, and the poll is re-weighted to reflect that distribution, then you will be doubling whatever bias is built into the smaller 100 young voter sample. The MoE figures that the pollsters present never take account of these internal bias-increasing effects, but they should.

None of this answers directly the question of how changing telephone affect the inherent left- or right-bias of polling. I suspect that there is not a major systematic bias in that regard, although I'd like to think about it some more. The problem, however, is not so much about systematic bias as it is about unreliable data due to non-representative sampling.

Finally, this analysis doesn't prove or disprove anything about Gallup vis-a-vis other pollsters, as they should all face the same sampling problems. I think the most interesting inside information would be not merely how each organization weights its samples and measures likely voters, but how they actually attempt to reach voters in the first place: what times do they call, how do they first introduce themselves to respondents, how persistent are they to obtain answers, etc.? I could imagine a rational argument that Gallup, as perhaps the most recognized name in polls (especially among older citizens) might actually obtain responses from certain people who would hang up on unfamiliar organizations -- assuming that they identify themselves when they call. Could that fact alone account for differential sample makeup?

You're right, brother. Washington State is not going to go for Little bush, and I am confident that Oregon will also see the light.

More good news from Washington State: We are on track to send 8 Democrats to the House (out of 9 seats), including two that are now rethugnican seats. Wish us luck, OK? I dearly want to say the words "Speaker Nancy Pelosi."

Grebberg,

I agree with you that Gallup's figures are way off, but on your other point... well, just maybe the spaceships _will_ come and take the believers away.
We can always hope...

Smooth Jazz:

Don't get too excited about the Pew Poll. It turns out that the Pew poll, no matter how highly I feel about Andrew Kohut personally, doesn't weight their samples by party ID either. Worse yet, Pew doesn't do their own polls. Instead, they use Princeton Survey Research to do their polls. What's wrong with that?

Princeton Survey Research is the contract pollster for Newsweek, who already put out a questionable poll earlier this month that also oversampled significantly for GOP respondents.

Do we know that Pew is weighing its responses the same way, though? One could assume that it is, but other polls, like SurveyUSA, don't have a specific weighing system. Secondly, people could be basing their opinions on the constant gloom and doom, downer treatment of Kerry and his prospects, but given sunnier prospects, would float over to his side.

Or maybe this is a crappy sample.

Plus, all of the attention has been on foreign policy lately and Bush, even with his glaring flaws, wins in that category. Kerry has to be able to attack him there, but then change the focus back to everything else, if he can.

http://www.pollingreport.com/wh04gen.htm
You don't see this major poll (devoid of statistical chicanery) being advanced in the media today. Why? Because Kerry is actually AHEAD. The media has to set up the scenario that Bush is ahead so in the next month when the polls "tighten up" they will have more of a story. Edward R. Murrow....WHERE ARE YOU????

Ramdan-

Large samples don't necessarily mean good samples. I'm wondering those surveys trying to pin down party ID numbers got a higher participation rate than some of the current surveys.

Smooth Jazz-

Two polls today. IBD at a tie, Pew with Bush up 8%. My take would be about the same as it was at the beginning of the day, and about the same as it's been for a good while now, that Bush is up about 4%. That makes it difficult for Kerry, but also means it's quite winnable. And the debates and GOTV will be important.

Interesting comments here, but let's not forget that there's one thing that Gallup's big media clients care far more about than helping the GOP, and that is making the election as close as possible so that people stay glued to the TV in the days prior to the election, and weeks after that with the legal and recount battles that would certainly follow. That's the way to sell newspapers and get the ratings that bring in ad revenue, folks. So don't be surprised if in the coming weeks Gallup engineers the samples to show a "surprising" Kerry comeback and keep the suspense for as long as possible.

A short while ago I heard an millionaire "pundit" on Lou Dobbs dismiss the MoveOn ad in the NYT as so much sour grapes. Dobbs himself held up the ad as if he was holding a dead rat.

Few minutes later, Chris Matthews, another millionaire "pundit", announced another poll showing Bush "lengthening his lead."

So much for our pundits on cable tv. They know on which side is buttered their bread.

Wrong. Party ID is NOT stable over short periods of time, according to Pew:

http://people-press.org/commentary/display.php3?AnalysisID=97

Does anyone know what the party breakdown was for the Pew poll?

I've never dealt with an org the size of Gallup, but I've seen smaller consulting firms lose all credibility in the 70s and 80s because they earned a reputation for cooking the books and telling clients what they thought the client wanted to hear. Clients tooled for products with little or no market or made investments in staff or services, all based on foo-foo dust provided by the consulting firm based on data firmly rooted in sand. I've seen conclusions that were patently false and disprovable and mgmt pretend to buy into them rather than admit they had been snookered by snake oil salesmen.
If Bush wins, Gallup will have been vindicated for having been on the winning side, but if he loses, most of us will remember Gallop in 2008 and laugh at his 2004 and 2000 predictions like so much soothsaying based on the movement of planets.

Yes I do recall Ruy sayin somethin about party ID volatility back when he was fueding with Dowd.

But serious folks...can't we talk about what is REALLY going on in the poll???


I am speaking of course of Pew/Gallup v. TIPP...

Kerry Katapults into Lead
Bush Crashes 9% in One Day!

I think Bush is ahead -- but by 4-5 points, a range which it should be possible to make up.

Do party Id's change over time ? Probably. Nonetheless, the last Gallup Poll result is still bizarre. It shows far more Republicans than Democrats, and a huge swing in 1-2 weeks where little else happened. So the answer is midway between the 2 extremes.

Kerry's last two weeks have been good. He was strong on Iraq and very strong on pointing his finger in Bush' face.

I dont expect to see him leading Bush this week but I feel pretty confident that he will show a slight lead in about 10 days or so.

If he gets his bearings right during the debate on Thursday, then I am even more assured of a lead in 10 days.

Keep trucking guys.. we need a change in the white house.

Was there Party ID breakdown for the WaPo poll? I am firmly convinced that Party ID does NOT change rapidly. There are many people who freely admit to being Democrats voting for Bush or vice versa. One reason Party ID favors the Dems is that in the South many people still call themselves Democrats even if they don't vote Democrat on national races. In Oklahoma, for example, almost 50% of the electorate is registered Democrat. Now whether "registered" Democrat means "identifies" Democrat, I don't know. But Oklahoma hasn't voted for a Democrat for President since Johnson. Anyway, there are some Republicans who never vote Republican on national races as well, mostly in New England. (In 1936 Vermont and Maine were the only two states to vote against FDR. Jim Jeffords came from that long line of liberal Republicans going back through the 1930s to Senator Justin Morrell during the Civil War. I don't have figures but I bet Party ID in Vermont is close to 50-50.) But there are more Dems for GOP candidates than vice versa, which is why a Dems holding a four point lead in Party ID this summer makes the race even, and a Party ID sample with a GOP lead would give Bush a huge lead.

Thanks for pointing that out Tony (IBD shows a tight race). But given that Gallup, Pew, Wash Post/ABC, CBS & AP have all come out with polls in recent days that suggest GWB has a 7-8 point lead, will you concede that IBD/CS MAY be an outlier?

Even more troubling for Kerry is that Pew shows him behind in the battle ground states 50-37. That's a huge margin to make up, if he is in fact that far behind in the key states.

Kerry's ahead, and the reason we know it is the number of Republican operatives that here trying to convince us otherwise.

Here are truths I believe the election will bear out:

1. The turnout will be at least 5 million more than voted in the presidential election of 2000.

2. Almost no one who voted for Gore in 2000 will vote for Bush in 2004.

3. Almost no one who voted for Nader in 2000 will vote for Bush in 2004.

4. Most of the additional voters will vote for Kerry.

5. Kerry will win by several million votes.

The poll which gives Kerry the biggest lead on election day will be the one which is most right.

Smooth Jazz,

If IBD/CS is an outlier, then what about Harris, Zogby, Fox, Rassuman, and ARG? Seems to me that either side can argue that theirs are the correct figures, and we Democrats have a sound basis for our arguments.

Presidential poll performance history

http://www.ncpp.org/1936-2000.htm

Smooth Jazz:

When individual state polls in places such as Michigan, Oregon, Washington, Pennsylvania and other places show that Bush is ahead in each of these states by over 10, I will believe the polls that show he is ahead by 8 nation wide. But, I need proof, and today's state poll, Quinnipac shows Kerry ahead of Bush in PA...and that is not proof to support Pew's claim that Bush is ahead in the Battleground State by 13 points. All National polls are garbage. Elections are won by electoral college. So just follow the state polls.

www.race2004.net

ruy, ruy, ruy,

it's alan and me - don't let alan and me have all the fun - not alan and i. ok pet peeve about grammar. cheney says "revert back" therefore must go. intriguing, encouraging website. thank you.

dnta,

You make some interesting points in your post.

It's very likely true, as you say, that young voters would be far less well represented in telephone polls today than older voters, requiring significant weighting to get them back to the proper proportions of the electorate.

In your example, if only 100, instead of 200, voters are included, then indeed far more unreliability is introduced on that count alone. But it's actually far worse than this. If, say, 30% of voters respond to pollster's questions, then this number would imply that only 15% of young voters would. So not only would the small number of young voters introduce great unreliabilities, the class of young voters who respond is far more deviant than for, say, older voters, greatly increasing the likelhood of sampling bias.

The more I think about all the ways polls today can get things far wrong, the more it impresses me that declaring anything with confidence based on a poll with fairly close results is sheer chutzpah.

Smooth Jazz-

"May" is not a term that gets used with outliers. A data point is or is not an outlier. I'm dashing through this morning, but just pulled up polling report. I see, in their overall list of surveys, and in particular of two-ways (it's the first set that I came to), a 2% Bush lead by Fox, a tie from Democracy Corps, and a 3% Bush lead from Zogby, in the last 10 days. Were IBD the only poll in recent days to show it deadlocked, it would be an outlier. It is not alone, so it is not an outlier.

Rather, you are cherry picking your polls to make the point you wish to make.

BTW, I've stopped monitoring those threads, but when last I checked, you had not responded to my query about the source (obviously pro-Bush) of your critique of polls nor to my notes demonstrating that when you wrote that "all" polls were showing a particular pro-Bush stance that there was ample evidence that it was not all polls.

'Even more troubling for Kerry is that Pew shows him behind in the battle ground states 50-37.'

But individual statepolls are a little different. Kerry looks solid in Michigan and Washington. He has slight, but consistent leads in PA, OR, MN. Wisconsin, I'm afraid, looks like a consistent Bush lead. NM and Iook like toss-ups.

Among the Bush states, NH looks like it may go Kerry, NV is close. Ohio and Missouri look pretty solid for Bush right now. FL still looks like a tossup, or a very slight advantage for Bush.

So Bush has a very definite advantage, but its not insurmountable at all.

By its own terms, Gallups polling is inconsistent.

Consider: in September Gallup has conducted 11 State Polls. If you compare these numbers to 2000, they so the follow:
Bush 1.01 % above his number in 2000
Kerry -2.85 below Gores number

This translates into the following National Number:
Bush 49.1
Kerry 45.5

Less than a 4 point race, with Bush below 50.

BTW - of you take the average of all September State polls and compare them to 2000, you get the following:
Bush -.78 %
Kerry - 2.77 percent, which translates into Bush 47.4, Kerry 45.6. , Undecided 5.36 %.

The state polls say this is a 2 point race with Bush under 50 and in trouble.

erg-

I'm a bit more optimistic about Ohio than you. I don't see it as solid Bush.

I draw on race2004.net for my state poll summaries.


The latest poll does show Bush up by 9%. But it's a GOP poll. The latest Dem. poll, by contrast, shows it even. The three latest nonpartisan polls have Bush up by either 2% (Rasmussen and ARG) or 3% (Opinion Dynamics). I'm inclined to peg Bush's lead there at the 2-3% range, which is easily surmountable. And that's a good thing, because I think a "win Ohio and NH while losing Wisconsin" strategy might be Kerry's best bet. That would give him 274 to Bush's 264. (Kerry doesn't really need NH in that scenario...)

BTW, I'm not sure I'd put in Wisconsin as solid Bush, too. The latest poll there, a GOP poll, has Bush up by only 3%. We'll see.

Fladem-

Interesting take. Thanks. Two questions...first, did you weight the polls in the Gallup states by population? Second, where did you get your estimate of the fifty different states?

Thanks.

Tony -- there have been 2 recent polls in Wisconsin, one by ABC and the Badger poll showing Bush up by 10+. While probably outliers, this and other polls to me indicates a Bush lead of around 4--5%, which is hard (but far from impossible) to overcome.

OH: The Ohio poll had a 10 point game. Again I think thats too much but a 4-5 point lead is hard to break.

Ruy and Alan, I asked http://www.dalythoughts.com/index.php?p=1872#comment-10894 Gerry Dale, proprietor of Daly Thoughts and Dales' Electoral College Breakdown 2004 http://dalythoughts.com/ about ya'll's analysis/ critique of Gallup's recent polls. Gerry is a freeper and a right-leaning gentleman but he has a very informative and comprehensive (hence valuable) compilation of and commentary on all individual state polls and maintains his own EC running total. He was kind enough to respond by starting this thread: Teixeira on Gallup. http://dalythoughts.com/index.php?p=1884 I thought it only proper to provide you with a link to his take on your article and provide you with the opportunity, if you wish, to respond.

I find the whole topic of polling methodology and analysis fascinating, confusing and addictive. Throw Gallup's recent polling results into the mix and I am even more dependent upon your, Chris Bowers' http://www.mydd.com/ and yes, folks like Gerry's help in achieving some semblance of a reasoned understanding of the subject. In this regard, may I say without any hint of flattery, your's is my "go-to-first" site and I greatly appreciate your efforts.

Hardly original, my own tongue-in-cheek response has been to post my INFALLIBLE 2004 Election Predictor: http://usapassport.blogspot.com/2004/09/infallible-2004-presidential-election.html

Please keep up your valuable work!