Why You Should Ignore The Gallup Poll This Morning - And Maybe All Of Theirs

September 17, 2004 2:05 PM

Why You Should Ignore The Gallup Poll This Morning - And Maybe All Of Theirs If you support the Dems, you might want to pass this on!
"This morning we awoke to the startling news that despite a flurry of different polls this week all showing a tied race, the venerable Gallup Poll, as reported widely in the media (USA Today and CNN) today, showed George W. Bush with a huge 55%-42% lead over John Kerry amongst likely voters. The same Gallup Poll showed an 8-point lead for Bush amongst registered voters (52%-44%). Before you get discouraged by these results, you should be more upset that Gallup gets major media outlets to tout these polls and present a false, disappointing account of the actual state of the race. Why?"...
posted by Postroad (53 comments total)

Presidential candidates have won after trailing by similar margins. One was George W. Bush himself. In 2000, he was behind Al Gore by 10 points among registered voters in early October and then prevailed in the Electoral College, though he lost the popular vote.

In 1980, Ronald Reagan was down 8 points in the Gallup Poll in late October but won in a landslide after doing well in the only debate held with President Carter.

"Sen. Kerry is like Seabiscuit: He runs better from behind," says Donna Brazile, who was Gore's campaign manager. But she acknowledges that "backbenchers" in the Democratic Party "have begun pushing the panic button."
posted by thomcatspike at 2:12 PM on September 17, 2004

79% of Mefites think polls are shit.

My sample size is 1.
posted by mr_crash_davis at 2:12 PM on September 17, 2004

ALREADY POSTED! DOUBLE POST! SOYJOY EXCLUSIVE! MUST CREDIT SOYJOY!

Naw, but seriously, Postroad, your bar for what constitutes a good, solid FPP seems to be... well, lower than mine.
posted by soyjoy at 2:19 PM on September 17, 2004

Crash, what 21% of you buys into polling?

I can guess, I just wanna hear you say it.
posted by chicobangs at 2:28 PM on September 17, 2004

Polling, as a science, is questionable at best. There was an NPR program about the polling process (I think it was on TAL) that was really, really insightful, although I can't find the link.

Basically it showed how pollsters pressure the pollees to confine their answers to the narrow options that the pollsters present..
posted by SweetJesus at 2:33 PM on September 17, 2004

chico, that may be a door you just don't want to see opened.

Polls are meaningless for the most part, especially these wildly fluctuating polls. Someone's already said it better but "Who in their right fucking mind is undecided at this point in the game?"
posted by fenriq at 2:42 PM on September 17, 2004

"Sen. Kerry is like Seabiscuit: He runs better from behind,"

Was that supposed to be encouraging? Because it's not.
posted by 2sheets at 2:45 PM on September 17, 2004

SweetJesus: that is a perfectly fair commentary on polls that ask questions like 'what should we do about medicare'. But it's not very relevant to polls where there are specific, narrow options, like, say, 'bush or kerry.'

[And it's worth noting that there is a large body of research into how to construct questions that are relatively neutral/unloaded; any 'report' that dismissed polling because a pollster can construct loaded/biased questions was either discussing a specific subset of polling, or was itself pretty screwy/biased.]
posted by louie at 2:46 PM on September 17, 2004

Jesus, next time a media organization wants to commission a poll, they would do more good to have the pollsters just phone people up and read them the newspaper.
posted by Space Coyote at 2:46 PM on September 17, 2004

What soyjoy said. But the issue of whether or not to normalize for party affiliation has long been argued. This has been discussed this election season going back to the controversial LA newspaper poll (IIRC).

My opinion is that normalizing for an assumed correct party affiliation ratio is in general a bad thing because party affiliation does change over time and, anyway, it's adding into the mix a new uncertainty. Texeira, I think, has very recently (the Newsweek poll suffered from the same problem) advocated implementing something like a rolling-average of reported party affiliation in a polling firm's recent polls as a normalization of individual polls. That sounds like the best all-around solution.
posted by Ethereal Bligh at 2:50 PM on September 17, 2004

SweetJesus: that is a perfectly fair commentary on polls that ask questions like 'what should we do about medicare'. But it's not very relevant to polls where there are specific, narrow options, like, say, 'bush or kerry.'

I'm not so sure you're right. I found the link, and it's specifically about Zogby (second segment). Listen and judge for yourself.
posted by SweetJesus at 2:52 PM on September 17, 2004

The 55% lead is a joke. A poll that has a margin of error as wide as 4% is not worth anyones consideration.
posted by fleener at 3:19 PM on September 17, 2004

I'm glad this finally came out--if all the organizations doing the polling released their oversampling--and assorted other tricks (like asking for the man of the house, and not giving an undecided option...), we might just begin to stop seeing them all reported as actual news (but i doubt it.)
posted by amberglow at 3:19 PM on September 17, 2004

Aren't the more important polls the state-by-state ones, since electoral votes are what matter in the election, not nationwide totals?
posted by gyc at 3:36 PM on September 17, 2004

Polls are useless and actually mean nothing anytime before the day of the vote.

That said, every other polling place has Kerry leading by around 1-3%. Me think Gallup is corruptus.
posted by destro at 3:50 PM on September 17, 2004

A poll that has a margin of error as wide as 4% is not worth anyones consideration.

There are polls with much smaller margins? I'd actually think 4% is pretty good.
posted by weston at 3:58 PM on September 17, 2004

And one more thing. Here's why you should really ignore the polls:

They shouldn't affect your vote. If you've already decided that Bush or Kerry is the guy with the better set of principles for governance, then the only thing I think polls can tell you is how many people think differently. And sure, if that number is significant, you probably should bother to try to find out why they have a different point of view, but once you've done that and made up your made about what's right, why should you even care about polls -- other than as an incentive to try to persuade others?

In short: paying attention to polls is for strategists and activists. And, I suppose, sheep who like to just make sure they're on the winning team, but if you don't fit into any of those categories, I don't see how polls have much value to you.
posted by weston at 4:13 PM on September 17, 2004

Polls aren't useless, as long as you understand the limitations. A four point (8% +/- 4%) is pretty close for any statistician. I think they should be polling people on the likelihood that they will vote. Personally, I am a little skeptical that people who use cell phones will constitute a discrete group of voters.

My man on the street intuition-- which doesn't mean much considering that I live in Canada and vote by mail-- is that Americans are getting jittery. Usually by this time I feel like voting is a matter of personal expression. But holy shit, motivation to vote might actually decide this election.
posted by gesamtkunstwerk at 4:44 PM on September 17, 2004

In short: paying attention to polls is for strategists and activists. And, I suppose, sheep who like to just make sure they're on the winning team, but if you don't fit into any of those categories, I don't see how polls have much value to you.

'Cause I want to know how the election is going to turn out?

I mean, there's a whole bunch of stuff that interests me even though it has no impact on how I choose to act. Isn't that kind of the raison d'etre of Metafilter?
posted by mr_roboto at 4:47 PM on September 17, 2004

> Why You Should Ignore The Gallup Poll This Morning

It shows Kerry losing. Do you need any more reason than that?
posted by jfuller at 5:05 PM on September 17, 2004

4 to 8% is close for any statistician? WTF? Anything above 3% is worthless. They're admitting to a variety of error inducers. I ignore any poll with an error rate above 3%.
posted by fleener at 5:27 PM on September 17, 2004

Personally, I am a little skeptical that people who use cell phones will constitute a discrete group of voters.

Really? I would think people who only have cells would be more likely to be younger & live in urban areas. Of course American politicians care fuck-all about the young & urban, though I can't help but wonder if their opinions were reflected in polls . . . (and yes, if they voted more).
posted by dame at 5:35 PM on September 17, 2004

Why should I care whether this kind of poll is right or not? Polls are meant to report my existing opinions, not influence them, so it doesn't matter to voters whether they're right or wrong. Nobody would change their vote just to be in the majority! That would be stupid! Right? Right?

[Silence reveals the sound of the wind. A tumbleweed rolls across the frame from left to right.]
posted by Hildago at 5:35 PM on September 17, 2004

That story is unbelievable. It's absolutely astounding the sort of shit that's shoveled to the American public without the media putting up so much as a fight--in fact, they seem to support it more than anything. You'd think with this sort of information out there that cable news outlets everywhere would be denouncing the gallup pole, but that certainly doesn't seem to be the case.

Oh, and the much vaunted (hah!) electoral-vote seems to have taken the bait, putting GWB over 300. Hopefully these shenanigans don't influence the outcome of the election.
posted by The God Complex at 5:51 PM on September 17, 2004

"Anything above 3% is worthless." I meant a 4-8% lead with a margin of error of 4% is not significant.

Reading polls is difficult, and it's hard not to project your own suspicions. They do collect quantifiable data, but respondents don't necessarily think quantifiably. I do hope that cell phone users will change the demographics. But statistically large groups generally reflect something close to the mean.
posted by gesamtkunstwerk at 5:56 PM on September 17, 2004

but part of the problem is that it's never large groups polled--always 2000 people or less, i find. How is that extrapolatable to our country of 280+ million?
posted by amberglow at 6:04 PM on September 17, 2004

'Cause I want to know how the election is going to turn out?

You will need time travel for this. Time series analysis of past polling data might substitute, giving you an idea of how likely a specific election outcome is given the recent stream of polling data. But if you really want to know how the election is going to turn out, you are going to have to travel to election day at 60 minutes per hour like the rest of us.

That said:

Oh, and the much vaunted (hah!) electoral-vote seems to have taken the bait, putting GWB over 300. Hopefully these shenanigans don't influence the outcome of the election.

I don't know if I'm comfortable with people dissing Electoral Vote. I'm sure the guy running the site has his failings, but everything I've read there leads me to believe that he's shooting straight, just trying to report the data he thinks is credible. In fact, read his notes from today:

Jimmy Breslin of Newsday had an column yesterday that, if true, makes this website irrelevant. Breslin claims that pollsters do not call the 168 million cell phones in the country. Since many younger voters do not have a land line and just a cell phone, they will be hugely underrepresented in all the telephone polls. Since younger voters lean more towards the Democrats than the average voter, the polls may be greatly underestimating Kerry's strength. Between missing all the people who have only a cell phone and no land line and the 5 million overseas voters, the polls maybe missing a very large section of the electorate.

The polls say what they say, and may be accurate or inaccurate, and may or may not predict tomorrow's outcomes. Those who don't like the results should get busy doing what they can to change them.

Or, one can be like the freepers who responded to Zogby's polls.
posted by weston at 6:06 PM on September 17, 2004

Me think Gallup is corruptus.

Funny you should mention that. Gallup is a very interesting organization, based in Omaha, Nebraska. I applied for a job there once. They make you take an online test that you're only allowed to take once: it has a java-based system that prevents you from using your BACK button, which should give you a clue as to the type of person they're looking for when hiring. The questions, as well, must be answered on a scale like "Agree Strongly -- Agree Somewhat -- No Opinion -- Disagree Somewhat -- Disagree Strongly." They're all psychological-type questions that seem rather innocuous, (e.g., "I'd rather go out and play ball with my friends during my free time vs. I'd rather stay in and read a book").

I'd heard that the kind of person they look for was the go-getter, corporate ass-blanket type. The kind of guy that went to college, joined a frat, got drunk with a bunch of assholes every weekend, barely skimming by, then went to work with said assholes at Investment Banks and other "Succeed!" kind of jobs where everyone goes around cracking slightly racist/sexist jokes and patting each other on the ass about a "job well done."

I decided to take the test with this mentality, while my GF, an intellectual bookish type with a penchant for Eastern European poets and a calm, quiet demeanor decided to answer the questions honestly. I ended up answering almost all of my questions either "very" or "not at all" -- no middle ground whatsoever. No compromise. No surrender. I like sports. I hate reading. I hate history. I love making money. I love thinking about all the stuff I'll have in the future. I hate being practical. That kind of crap.

Now, you'd think pollsters would want the sort of people that didn't have strong political opinions (if they want accurate results). You'd think that they'd want people with a broad understanding of history and culture to be able to analyse results and formulate questions.

Turns out, I got the interview, my GF got the "thanks, but no thanks" email. In the end, the commute from Lincoln to arrive at work at 8 a.m. would have killed me, so I'm not working there.

I mention this not as an inditement of the poll results, which as jfuller points out ("It shows Kerry losing. Do you need any more reason than that?") are somewhat subject to personal bias. I just wanted to throw in a small anecdote about the company in question, and like the poll results, let you draw your own conclusions.
posted by Civil_Disobedient at 6:13 PM on September 17, 2004

"Horse-pucky, pure horse-pucky ... Zogby, take your biased polling and stick it. By the way Zogby does the term "landslide" invoke fear? Well, get used to it come November when Bush kicks Kerry's butt every way but loose in the general election. Explain your polls then AH."
posted by weston at 6:36 PM on September 17, 2004

I have seen a number of online team games like Halo or Quake 3, et al where some joker will switch sides a minute or two before the end of the game just so they can be on the "winning" team. It makes no sense to me. But I bet those are the people who vote (if they vote) according to who is expected to win. I think polls do matter as a sort of self-fulfilling prophecy.
posted by stevis at 6:51 PM on September 17, 2004

4 to 8% is close for any statistician? WTF? Anything above 3% is worthless

No, it's not. It just has wider error bands. You can't make as many fine distinctions, but they're still useful.

You have to look at how margin of error scales with sample size. If you want a 4% margin, you only need about 600 respondents. If you want a 3%, you're up to 1000, if you want 2% you need about 2500, and if you want 1% you need about 10000; there's not much point in going beyond 1000 people (3% margin) unless you want to do inference about subpopulations.

But, you might be able to learn more from 15 polls of 600 people than you can from one poll of 10,000. You might be able to learn more from 4 samples of 600 than 2 of 1200. So going for a max sample size, or a 3% margin, isn't always the best thing to do. We see a lot of 3% margin polls for no better reason than that about 1000 people is where the bang-for-buck starts to get really small.

They're admitting to a variety of error inducers. I ignore any poll with an error rate above 3%

These have nothing to do with each other. For a proportion, margin of error is a function of sample size and the proportion itself (extreme proportions have smaller error bands than 50/50 splits do). The error inducers you're thinking of are basically departures from simple random sampling, which you might or might not have a good reason to do depending on what you want to find out.
posted by ROU_Xenophobe at 6:52 PM on September 17, 2004

I've dealt with Gallup in depth and directly; forced to participate in one of their mega-buck surveys through my workplace. To my mind, their methodology in sheer BS pop psychology. They were essentially feeding my employer back what they pretty much wanted to hear, not unlike a scam fortune teller. They play off corporate insecurity. I see no reason to accord these latter-day water dowsers any credibility or attention.
posted by RavinDave at 8:26 PM on September 17, 2004

this was in the wsj today, "What If the Polls Are Wrong?"

Gallup explains it has what it considers a time-tested formula for determining most likely voters. It asks eight questions, such as current intensity of interest, past voting behavior and interest, and whether you know where your voting place is.

[...]

But there is reason to suspect those criteria are outdated, especially in an election where both sides say the intensity level is much higher than four years ago and get-out-the-vote organizations are considerably better than ever -- few people on Nov. 2 will be in the dark on where the voting polls are.

"A formula that made sense years ago may not recognize all the changes in society," notes Mr. Hart. "It gives more credence to past behavior and too little to current interest."

"For low-turnout elections those old models work well," suggests Bill McInturff, a Republican, and the other WSJ/NBC News pollster. "But in today's presidential election those models tend to [tilt to] a little older, a little more white, a little more affluent and a little more Republican voters. They may miss some of the extraordinary activity going on in African-American and Latino communities."

The registered-likely voters dichotomy also is evident in some of Gallup's state surveys including last week's Ohio results." Among registered voters in the Buckeye State, Bush-Cheney had a 48%-to- 47% edge, a dead heat. Among likely voters, however, this poll had the Republicans up 52%-44%; that garnered all the attention, followed by a spate of stories suggesting this key battleground state was moving to the president.

hmmmm, zogby uses 'likely' voters too...
posted by kliuless at 9:11 PM on September 17, 2004

Thanks Civil_Disobedient - nice addition to the thread.
posted by meech at 10:41 PM on September 17, 2004

Well, I gotta admit it's turned out better than I expected.

...so (mumbles uncomfortably) ....thnks ....pstrod...
posted by soyjoy at 11:03 PM on September 17, 2004

PEW poll 45-46
margin of error 3.5%

Gallop poll 56-45
margin of error 4.0%

margin of ERROR?!/!?!?!"


BTW, Dan Rather = Whore.
posted by HTuttle at 1:08 AM on September 18, 2004

amberglow, take a statistics class, the math is all worked out to prove it works (if a certain set of reasonable assumpetions are met).


Ok, so we all now know that the Gallup is over-representing republicans by 8% over 2000 exit poll returns. The question is -- why are they doing this? I mean, do they think their formula for LV is really gonna be waht turns out? Is this a big fluke? Like whats the explanation for the internals, havent seen that anywhere yet.
posted by nads at 1:17 AM on September 18, 2004

but part of the problem is that it's never large groups polled--always 2000 people or less, i find. How is that extrapolatable to our country of 280+ million?

Easily, so long as everyone in the target population has the same probability of being in the sample*. Differences start cancelling each other out, so the margin of error depends, basically, on the absolute size of the sample.

It turns out from the math that population size isn't very important except for very small populations. In fact, it's usually easier to just assume that your 2000 people are out of an infinitely large population.

There's a calculator up here, if you trust its math. If you want 95% confidence and a 3 point margin, then you need 92 people from a population of 100, 516 from 1000, 964 from 10000, 1056 from 100000, 1066 from 1000000, or 1067 from 10000000000000000000000000000000000000000000000000.

*it's doing this that's the trick.
posted by ROU_Xenophobe at 1:57 AM on September 18, 2004

Amberglow, apparently it is proven that the magic number is 2000 people polled to have an amazingly low margin of error. However, this assumes, as Rou and nads point out that the sample is REPRESENTATIVE of the the population. All 300 million of them. Evidently, the Gallup poll is deeply flawed because they are overpolling Republicans. Therefore, it is correct to assume that their margin of error is in fact much much higher than 4%.
posted by sic at 3:14 AM on September 18, 2004

We're assuming that Gallup and others are overpolling Republicans. And we're probably—almost certainly—right. But we don't know this.

I'm really trying to break the habit of arguing from counterfactuals but...well, it's hard for me to imagine that if Dems were being "overpolled", and Kerry were 8+ points ahead, that people wouldn't here be arguing that normalizing the results by adjusting party affiliation to an assumed correct amount would be BS. They'd say, "Hey, people are excited about the platform and Kerry and tired of Bush and so more people are identifying as Democrats. This really reflects what we'll see on Nov. 2."

Now, as has been discussed, the truth of the matter is that large changes in party affiliation in a matter of weeks or months (or even a few years) are completely unheard of and therefore extremely unlikely. So normalizing for party identification is probably safe to do.

But the very first time such a normalization goes against one's partisan interest, one will likely seriously question whether it was a good idea.

To the degree that you can safely adjust your data against things like true party affiliation and true likliehood of voting (versus being merely able to vote), the more desirable it is to do so. But the big question with these things is how safe is it do so? How good are the assumptions and models? Do they introduce a larger error than they are attempting to emlinate?

I think that aside from LV data immediately before the election, history has shown that LV polling is less accurate rather than more (relative to RV polling).
posted by Ethereal Bligh at 3:41 AM on September 18, 2004

"We're assuming that Gallup and others are overpolling Republicans. And we're probably—almost certainly—right. But we don't know this." (EB) - Yes, and I'm assuming right now that it is raining outside of my window. I can see and hear the rain. But it could all be a clever illusion, or I could be hallucinating.

"....it's hard for me to imagine that if Dems were being "overpolled", and Kerry were 8+ points ahead, that people wouldn't here be arguing that normalizing the results by adjusting party affiliation to an assumed correct amount would be BS." (EB) - Perhaps you're right. Let's hope we'll have the luxury of running that experiment in real time, eh ?

But, as things stand, some of these poll results we're hearing seem intentionally slanted to skew the election by suppressing the Democratic vote, encouraging Republicans to vote, and - crucially - through the "perceived winner" effect (many people will switch allegiance at the last minute to back a perceived winner).

I have to assume that the Gallup polling organization is not staffed entirely by idiots (though Civil_Disobedient's story of Gallup's online Java-based "interview" was telling) and that there are people at Gallup who are aware of the methodological underpinnings of such poll bias but don't or can't correct for the bias : I can also imagine orders from the top of this organization preventing such correction of known bias, for political reasons.

But, who knows ?

All I know is that even as I type, Public radio's pundit prince Dan Schorr is mumbling about this issue and citing one "neck and neck" poll and two "Bush clearly in a strong lead" polls and so - in my opinion - Schorr and Public Radio are acting as another cog in this machinery designed to pull off a Bush win by creating a perception, now, that Bush is in fact ahead.

Consciously ? Who knows. I could invoke the "Idiots are everywhere" hypothesis (and it does hold some water) but I have to assume some measure of intelligence on the part of the chattering media classes and so I choose to assume a measure of intent.

The largest newspaper in my area - not clearly left or right, I'd say - is reported the Gallup/AP claim (propagandistic, I'd call it) verbatim - that Bush has a wide lead.

I call BIAS (see comment below)

"That story is unbelievable. It's absolutely astounding the sort of shit that's shoveled to the American public without the media putting up so much as a fight--in fact, they seem to support it more than anything."(The God Complex) - the media slants right. Even Public Radio.

Here's the king of American print media, the NYT, on the story : "Varying Polls Reflect Volatility, Experts Say" (NYT, September 18, 2004) "The Pew poll released Thursday was based on a survey of 1,972 registered voters in two waves between Sept. 8 and 14. It found that Mr. Bush and Mr. Kerry were tied at 46 percent among registered voters while Mr. Bush held a statistically insignificant 47 percent to 46 percent lead among likely voters by the end of the second stage of polling, from Sept. 11 to 14. The first stage, Sept. 8-10, showed Mr. Bush leading Mr. Kerry by 12 and 16 points in those groups, respectively - a clear sign of voter volatility.

A Gallup Poll released Friday, on the other hand, found Mr. Bush with 52 percent to Mr. Kerry's 44 percent among registered voters and a 55 percent to 42 percent lead among likely voters in a survey taken Sept. 13 to 15. The New York Times/CBS News poll conducted Sept. 12 to 16 had Mr. Bush over Mr. Kerry by 50 percent to 42 percent among registered voters. Mr. Bush's edge increased slightly - 51 percent to 42 percent - among likely voters. Other national polls have reflected a closer contest."

Notice that the NYT does not mention even the names of any of the organizations which have conducted polls showing the race to be in a statistical dead heat (other than the Pew Research Center's poll).

That's a beautiful example of bias insertion : give the views you want to promote more air or print time/space, and - while mentioning them in a vague way - minimize the views you want to downplay.

Having read this NYT piece, it's clear to my what Dan Schorr was lazily regurgitating, on "Weekend Edition" this Saturday, from his morning coffee and newspaper breakfast reading session.

UK Guardian : Race remains wide open
: The race for the presidency appeared wide open yesterday as contradictory opinion polls suggested that the electorate is increasingly volatile......Voter volatility was best summed up by two polls for the Pew Research Centre. The first, between September 8 and 10, gave Mr Bush a 16-point lead. The second, conducted between September 11 and 14, had Mr Bush with a statistically insignificant one-point lead among probable voters and the two candidates were deadlocked among registered voters.

A Harris Interactive poll, conducted from September 9 to 13, gave Mr Kerry a one-point lead - the same organisation gave Mr Bush a 10-point lead in June. A Gallup poll published yesterday and conducted from September 13 to 15 showed Mr Bush widening his lead with a 13-point advantage over Mr Kerry. - Unlike the clearly biased NYT story, the UK Guardian does not cite unnamed polling organizations, and it notes that the Gallup poll results are a clear anomaly.
posted by troutfishing at 7:00 AM on September 18, 2004

Thanks all, but i know that people run and set up these polls, and are always biased in one way or another--show me a truly objective poll--in language, people chosen for the calls, etc, and i'll buy it. There are going to be many millions of (scared) first-time voters this year (this is the first election for the Britney/baby Jessica in the well generation), but i've yet to see a poll that mentioned them. Are college campuses polled? There are more college aged people in the population now than have been for about 25 or more years, so why would a pollster use past college voting patterns (during peacetime, and when they were a smaller percentage of the pop) to extrapolate, especially at a time of war?

And if we're going to buy what polls say: Polls have shown an even split all year--why would that change, without anything big in the world happening to affect voter perception? (there's no Saddam capture, or invasion about to happen, no giant scandal yet, etc) I contend that it's the language/setup of the polls (as in that bs convention bounce poll) and the people being called (this oversampling thing, etc). I could put together a mefi poll right now and get the results i wanted, but that wouldn't be objective.
posted by amberglow at 7:25 AM on September 18, 2004

Skallas - As with the demonstrated bias of Gallup New, "The magazine Editor & Publisher has been tracking newspaper endorsements of presidential candidates since 1932.  Contrary to the myth of the liberal media, in only two elections since then -- 1964 and 1992 -- have more endorsements gone to the Democratic candidate than the Republican.  The 2000 election was no exception, according to a survey E&P (11/6/00) commissioned of newspaper executives: 48 percent said their paper would support George W. Bush, while only 23 percent were picking Al Gore.  Personally, 59 percent of publishers said they planned to vote for Bush, v. just 20 percent inclined to Gore."  (Extra! (FAIR), January/February, 2001). (passed on via this site, and also see here

Here, almost in it's entirety, is the Rosetta Stone of modern American Editorial and Publisher bias, this, pre-2000 election E&P survey :

"George W. Bush is heading for a surprisingly comfortable win in Tuesday's presidential election, according to a new E&P/TIPP survey of the nation's newspaper
editors and publishers, completed Tuesday."

So, Gallup - along with US print media - has slanted right since the mid 30's : or, from that era when some in the US - deeply concerned about perceived inroads being made by the American Communist Party at the time - began a concerted PR campaign to thwart those advances which degraded later into a self-serving effort to thwart any and all progressive political movements in the US.

A study of ABC World News Tonight, CBS Evening News and NBC Nightly News in the year 2001 shows that 92 percent of all U.S. sources interviewed were white, 85 percent were male and, where party affiliation was identifiable, 75 percent were Republican. "

[ background material : Metafilter 35411, The Noise Machine :The Republican propaganda mill, a brief history It's bigger than Bush vs. Kerry. It's about billionaire funded thinktanks (AEI, Heritage) paying columnists to sit around and make stuff up or legitimize crackpot theories (blacks are genetically stupid, japanese internment was okay). Furthermore its about radio, internet, blogs, tv news and publishing houses working in concert to pummel memes onto the American public. When this stuff infects your culture and is no longer the domain of the loons but now as mainstream as apple pie and Wal-Mart, what do you do?
Also see Metafilter 35443, "The bias of balance : new study of how media "evenhandness" distorts truth" ( subtitled "The Moon - Satellite...or Cheeze Wiz ?" )""Media covers massive D.C. (and world) Anti-War protests, discounts numbers - Backflash: NPR and the NYT later issued apologies for their drastic undercounting of the Oct. 26 D.C. Anti-War protest" "You could get a good call girl, for a couple hundred dollars a month" (on Project Mockingbird) "George Lakoff tells how conservatives use language to dominate politics - Why do conservatives appear to be so much better at framing? - Because they've put billions of dollars into it. ]
posted by troutfishing at 8:03 AM on September 18, 2004

Thanks, trout.
posted by Civil_Disobedient at 12:54 PM on September 18, 2004

I find a good rule of thumb to be don't trust any poll unless the numbers reflect that the population disapproves of the current administration and will elect Kerry in November. Otherwise, it's clearly crap data.

More crap data.
posted by David Dark at 1:02 PM on September 18, 2004

troutfishing for President
posted by matteo at 1:43 PM on September 18, 2004

matteo for first lady.
posted by David Dark at 1:51 PM on September 18, 2004

Great thread all!

Your data mining is truly a treasure for all MeFis trout so I also wish to thank you.

In order for Diebold to deliver the vote for Bush it is first necessary for the Mighty Wurlitzer to deliver the perception that it is legitimate. Gallup and many media outlets are busy laying the groundwork. At least Putin is honest and forthright about usurping any semblance of democracy.
posted by nofundy at 3:33 PM on September 18, 2004

nofundy - thanks. Also - I think Putin is a lot more honest than are most US politicians (which is not saying much).

matteo - sorry, but I've got too many blots on my blot of a non-record.

David Dark - Why not just chill and eat a Habanero pepper from my garden ? It's not the hottest but still packs a decent buzz.

Special Thread Bonus : The CIA on Campus!
posted by troutfishing at 8:24 PM on September 18, 2004

I just got a survey in the mail from the Post Office, and guess who's conducting it? Gallup.
posted by amberglow at 8:38 PM on September 18, 2004

I bet Gallup is flush with cash for funky polls this month.
posted by troutfishing at 8:53 PM on September 18, 2004

Who's really ahead in race for White House?

George W. Bush has either established a commanding lead over John Kerry or he has lost most of the momentum generated by his party's convention and fallen back into a tie with his Democratic rival. Perplexed? Don't feel alone. So, it seems, are the pollsters. How else to explain the fact that two national surveys released on consecutive days last week by two highly respected polling firms, Gallup and the Pew Research Center, came up with such contradictory results? And in the individual states where the presidency will be decided, things are even stranger. Polls show some states regarded as uncompetitive are suddenly close, others thought to be battlegrounds falling out of that category -- and in one swing state, Minnesota, four surveys with widely differing results. Amid this uncertainty, one thing seems clear: After months of being frozen in deadlock, the race for the White House has become highly volatile.

Bush Seen Vulnerable to Kerry Among Independent Voters

President Bush, who holds a sizable lead in some polls, still appears to be vulnerable to Democrat John Kerry among independent voters whose shifting loyalties could determine the winner of the November election, pollsters say. Polling results from the Pew Research Center, the Christian Science Monitor and the Gallup Organization suggest independent voters are favoring Kerry as concerns about the economy and Iraq re-emerge as top campaign issues, despite a surge of support for Bush following the Republican convention.
posted by y2karl at 10:11 PM on September 18, 2004

They did it again! Gallup Is At It Again - Yesterday's National Poll Had 12% GOP Bias
posted by amberglow at 5:36 PM on September 28, 2004