It is awfully suspicious. It's clearly not a good sample, given that there is a preponderence of self-proclaimed Democrats in the United States. However, if a large number of Democrats have changed their minds in the last 2-4 years, and now call themselves Republicans, this may be a legitimate poll.

So the question is whether a large number of Democrats have had sufficient reason to switch parties in the last few years. I'm not sure they have. But then I haven't, so what do I know?

Obviously, Gallup polls are bad but now every time some gop pundit quotes them, you can say "Even Gallup shows Bush losing support."

They're guilty of the disinformation campaign because they hate Democrats.

Bush can lie to start a war, take the deficit to $422 billion, produce not-since-Hoover net job losses, allow 9/11 to happen on his watch, ignore health care, piss on workers and help out the rich, who were already in so much pain.

So? The fag-loving god-hating weak Democrats are out there. Crush them, even if you have to transparently cheat and lie to do so!

[sigh] Some of this comes from an inability of the Democrats to fight back. But almost all of it comes from the twisted GOP logic that it's un-American to disagree with them.

If you don't believe that America was set up to be a theocratic corporate oligarchy, you are simply the enemy. The whoring press agrees with them and helps the GOP out, right along with Gallup.

I dreaded this election for precisely this reason and it's happened. ABC News (sic) reported yesterday Kerry has an approval rating of 37%. No one outside of the Democratic base likes him. Becuase the only choice Bush had was to ruthlessly trash Kerry with whatever disgusting lie he could find while the press let him.

We're screwed. Again. For the ranting psychosis of a few fascists who have hijacked the Republican party the press is willing to throw the country away. That's just life in America. Our reward is to watch the country slide into repressive oblivion for the rest of our lives. Thank you so much, ABC News (sic).

Sorry. That's the answer. There is no other outcome if Bush wins.

Giant Corporations control the Media and the GOP is the friend of Giant Corporations. Pretty simple, huh? What did you folks think was gonna happen? the GOP would just lose the election and slink away with a frown on their face? No, that is not the way it will happen. They will fight with whatever necessary to keep power.

Check out the ad in today's New York Times that really puts it to the Gallup organization. The polling company is now run by Son-Of-Gallup who's a born-again evangelical and believes he's "doing God's work" with his polling . . . i.e. getting GWB reelected, I suppose.

Hmm, the poll could be accurate considering how much Republicans will try to supress Democratic voters.

Would anyone else here, besides me, be tempted to...say...mislead a pollster as to what your party affiliation is and who you are planning to vote for just to skew their results?

I've never been called (New York, no cares about us non-swing states) but I must say I would be tempted.

I don't understand why you blog guys don't do what Freepers do and put the email address and phone number of who needs to be contacted at CNN to complain and then ask us to bombard them with emails.

It gets things done...and I doubt the decision making people at CNN are reading Left Coaster with any regularity.

Any URL associated with the NYT ad?

This poll is so ridiculous it cannot be believed. Of course, it will be believed, by the media as it tries everything in its power to make a Bush reelection seem inevitable. Kerry's strength is much much higher than recent polls have indicated. Remember, in 2000, only 2 of about 2 dozen polls right before Election Day had Gore up, when he won the popular vote. I think Bush is doing suprisingly well, considering how bad a president he is. The pubs are stubbornly loyal and the media is in their pocket. But Kerry has broader support than is being measured. And when you look at the surge in voter registration -- all the new voters are not included in the Likely Voter polls -- the tide is for Kerry.

As Ruy Teixeira points out at his site, to believe these numbers, you'd have to believe that the Republican convention -- not September 11th, not the Iraq war -- caused a switch in party identification greater than any experienced in polling history. Right after September 11th there was a small such migration, but nothing on this level, and it soon normalized back to status quo ante. Now, a few speeches by Schwarzenegger and Guiliani are supposed to have been the re-aligning equivalent of the New Deal? Please.

These numbers are indefensible in terms of reflecting any genuine reality. And, as some numbers-crunchers have shown, if you normalize Gallup for party ID, you get about the same result as the many other polls showing a tight, both-below-50% race -- one which history suggests favors Kerry.

Gallup is rapidly making itself a joke, and CNN along with it.

spage@usatoday.com

Also, Take Back the Media has those email addresses.

born-again evangelical and believes he's "doing God's work" with his polling . . . i.e. getting GWB reelected, I suppose.
I will deliver Ohio's 20 electoral votes for Bush. There's another nut.

For the ranting psychosis of a few fascists who have hijacked the Republican party the press is willing to throw the country away.

It isn't just a few fascists, it's a large chunk of America that has bought into the "we're victims of liberalism" bullshit the Rethugs have been shoveling for decades. America is in trouble, guys. We're facing the strongest challenge to Constituinal values in our history, from a group of people that foam a the mouth with warped idealism. They will do whatever is necessary to win, and to hell with the rule of law. They did it in 2000 and they will certainly try to do it again in 2004..

Useful on this topic is this post (Wednesday, September 22, 2004) by Robert Waldmann on
ARG reports polls for 50 states and DC.

It had Kerry/Bush at 46/47.

check out what the votemaster has to say today over at www.electoral-vote.com

the money bytes are:

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.

D and T, well, there was a NYtimes article, I'm sure it was posted on Leftcoaster as it was on many blogs, that showed Dems kicking GOP ass in voter registration for this year in Ohio and Florida. So you have a point, but if so many felt that way, I dont see a large number lining up to register Dem, which is what is happening.

Red States!

Our blue tax dollars aren't enough!

They wanna control how we live our lives as well!

Better dead than Red!

The polls are skewed because the Republicans want to ensure the public will not be surprised when Bush steals the election.

If the polls were honest they would show Kerry to be far ahead. A Bush "victory" without the setup of crooked poll information would be revealed as theft.

What kind of numbers would a fair re-weighting of this poll yield?

Anyone?... Anyone?... Something-doo election polling.

I assume this is part of the set up, defining the margin by which the bushmafia intend to "win". The thumbprints of election fraud are everywhere -- the fix is in. We must just pray for overwhelming voter turnout, dumb luck, and their plans to go awry. My consolation is that the Republicans always overreach, and , thus, are the agents of their own undoing. Fear not.! "Mars, bitches"

I wonder if the false results might have the unintended consequence of making the Rethugs overconfident?

IMO, that seems to be the case with bush. His arrogance is over the top these days ('mission accomplished' - I'd say it again, etc).

Yesterday on Inside Politics, Judy Woodruff seemed baffled when Tad Devine (from the Kerry campaign) called these polls into question. He said they were skewed heavily with republicans but she said something like 'well I don't know about that we'll have to check into it'. Let's see if they follow up on that.

If I have said it once on various blogs, I have said it a million times.
BUSH MUST LOOK LIKE HE IS IN THE LEAD WHEN THEY STEAL THE ELECTION.
It's critical that people believe that Bush is winning this election.
That way, when they try to steal the election in Ohio, Florida and other close states, people will shrug their shoulders and say: well, what's right is right. Bush was ahead in the polls, he probably is the clear winner. We're all supposed to roll over and take it.
I don't believe any of the polls except Zogby. Even then I am skeptical. The article in AP the other day said that Democrats have registered en masse. Of course, the headlines didn't say that, because they are hoping people don't read to much in it. But Democrats came out en masse in the primaries, they have registered en masse to vote. They want Bush out, but those same voters aren't included in polling because many of the newly registered voters aren't considered "likely" since they haven't participated in past elections.
It's all smoke and mirrors here. We are supposed to believe that this election is Bush's to lose. That way, when the SCOTUS hands the election to Bush, we won't be shocked. The polls are complete and utter BS.

I sent numerous e-mails to CNN and USA Today today.

What about the rest of you?

Are you going to just bitch or help work the refs?

Somebody will pick up on this story. Perhaps we should be targeting competitors that use polls other than Gallup?

Daily KOS has a project going on right now to expose CNN for this by blasting their America Votes 2004 email bin. Here is the link:

http://www.dailykos.com/story/2004/9/28/113514/523

I think Erica is right. The polls are merely a cover for theft of the election. Basically, what we have here is something very akin to an organized crime family that has taken over the U.S. government.

And we shouldn't be suprised if, as the election gets closer, and especially after the election, there are huge problems with the internet that somehow only affect left wing blogs.

I thought I remember reading Gallup's "defense" of their methods being that they don't ask for party registration but rather "what party you affiliate most with." They then claimed that this wording pushes people to self identify with whichever party's candidate is actually winning. "Bush really has a big lead, thus, people are self identifying as Republican. The idea of even using this method seems highly dubious to me, not to mention that I don't know if I agree with their analysis. Just thought I'd throw out the other side's "defense."

I thought I remember reading Gallup's "defense" of their methods being that they don't ask for party registration but rather "what party you affiliate most with." They then claimed that this wording pushes people to self identify with whichever party's candidate is actually winning. "Bush really has a big lead, thus, people are self identifying as Republican. The idea of even using this method seems highly dubious to me, not to mention that I don't know if I agree with their analysis. Just thought I'd throw out the other side's "defense."

Just sent the following email to all branches of CNN and to all the anchors (and Greenfield). Suggest you do something similar!

I hope that when you have a report on the latest Gallup Poll of likely voters, you mention that the poll sampled 12% more Republicans than Democrats (43% to 31%)?
How about an opening line like "Bush leads by only 8% in a poll with a 12% Republican sample bias." That sure puts a different spin on the story! You better look closer at Gallup's methodology. It doesn't reflect reality.

Just sent the following email to all branches of CNN and to all the anchors (and Greenfield). Suggest you do something similar!

I hope that when you have a report on the latest Gallup Poll of likely voters, you mention that the poll sampled 12% more Republicans than Democrats (43% to 31%)?
How about an opening line like "Bush leads by only 8% in a poll with a 12% Republican sample bias." That sure puts a different spin on the story! You better look closer at Gallup's methodology. It doesn't reflect reality.

fucking gallup

Corporate ACC shows Kerry in the lead 49% to 43%, rounded. The other 8 points are muddled, less than three to Nader, 3 lean towards Kerry, etc.

I wonder how much the polls are being skewed by that group of democrats who self identify as independent (about 15% of the entire population), yet who are registered Democrat, and more likely to vote Kerry?

Actually, I'm not supposed to tell anyone this, but Gallup is actually working for the Democrats.

They're skewing the polls to show Bush with a lead so that traditional conservatives (small government, individual liberties) who have been annoyed by Bush will just stay home on election day, thinking they don't need to vote. Kerry wins in a landslide.

... or was that a dream I had?

could Gallup be taking their estimate of vote fraud into account?

I don't understand how anybody can take seriously a poll that makes a priori assumptions about who's going to turn out. How widespread is this?

What's the real ratio of registered Republicans to registered Democrats in the USA? Anybody know?

What kind of numbers would a fair re-weighting of this poll yield?

Kerry 49% - 43% for both polls, rounded. There appears to be 5 points still up for grabs in the undecided category. The moral of the story: Get the vote out.

That was not a dream, buddy. Kerry will win in a landslide. After the election, all of us true blue folks who want to live in a free America, should storm all major media TV stations and hold anchors hostage with the threat of beheading...

Rather than kick up a shitstorm, maybe we should consider the possibility that once again the Republicans are doing a better job of getting their base excited and committed to this election. This upswing may not last, but it correlates with an explosion of Bush signs around my small town that were invisible a month ago.

Mystery pollster and others have done some informative debunking of the idea that surveys could successfully weight by party ID. Right now, republicans look a lot more excited than democrats. That may swing by Nov 2, but it's a problem to deal with not shout down.

Great post, Steve. I posted a part of this piece on Democratic Underground so it would gain more readers. Also plugged this site over there (as I have before). You guys deserve more readers!

Add up the categories:

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

That's 101%, even before you include "don't know."

I didn't do much with math beyond 11th grade, but that doesn't sound right...

KOOL-AID HERE!!

GET YOUR FRESH COLD KOOL-AID HERE!!!

You people are PATHETIC!! Making up excuses already! Has it ever occurred to you that.... get ready now..... listen close.....the majority of the people REALLY DON'T LIKE KERRY!! ooohh my god! Could it really be true!!

Problems with polling...

1) Many 18-25 year-olds (college students) have only a cell phone. Just as with telemarketers, polling organizations do not call cell phone numbers due to legal issues. (By calling you a cell phone, you are forcing the cell phone owner to incur charges, etc.) It is common knowledge/conventional wisdom that 18-25 year-olds are more liberal than 35-50 year-olds. (This can be see in polls taken measuring support of gay marriage. 18-25 support was 43% in poll I saw while 35-50 support was 18%.) So liberal voters are being under sampled (as evidenced by Gallup's polling data.)

2) Older American are more willing to participate in a phone poll. A greater percentage of younger Americans have caller ID than older Americans...as a result, many younger Americans will not answer a call from a polling organization, whereas older Americans pick up the phone whenever it rings. (Again, see Gallup's data.)

3) Many democrats/liberal-leaning people are "closet" Kerry supporters (including myself.) I work in an extremely pro-Bush, right-wing organization (military contractor, majority of employees retired military.) My co-workers know I don't like Bush, but I moderate that with criticism of Kerry so as to not be the "target" of their anti-Kerry attitudes/remarks. I live in Virginia (very republican state) and I have seen more bumper stickers and yard signs for Kerry than I have seen those for Bush. That alone is a good indicator...there may be pro-bush people out there, but they aren't "pro" enough to advertise the fact. If I were called and I decided to answer the phone, I would say that I was voting for Bush...if for no other reason than out of spite.

The reason the numbers don't add up to 100% is because they are rounded to the nearest point.

That being said, the issue here is what is known as "weighting by political party". Or more accurately, not weighting by political party. Political polls typically weight by geography, sex, race, and age in order to achieve a more accurate prediction.

However, the jury is still out whether a political poll should weight the responses by political party.

There is an excellent article on the pro's and con's of weighting here:

http://www.hs.ttu.edu/hdfs3390/weighting.htm

Some pollsters, such as Rasmussen and Zogby, and die-hard "weighters". Others like Pew and Gallup do not weight by political party. Which is more accurate? We'll know on November 3.

--Stephen

Lets hope Kerry doesn't shoot himself in the foot during the debates.

The reason the numbers don't add up to 100% is because they are rounded to the nearest point.

That still doesn't work, unless all three were rounded up, which is not the generally accepted practice in rounding.

It seems obvious to me that the purposeful pro-GOP weighting is done for three reasons:

1) To make Kerry look weak (and Bush look stronger than he is) prior to the election and make the media put him on the defensive.

2) To provide cover for attmepts to steal the election.

3) To provide cover for an immediate, massive, all-out assault on the legitimacy of a Kerry presidency should 1 & 2 fail.

*sigh* It saddens me it has come to this for America...the end of democracy is far closer at hand than I ever thought it was.

What's the real ratio of registered Republicans to registered Democrats in the USA? Anybody know?
There were more dems than R in 00. Apparently, there was a small swing after 9/11, but it was returned to the 00 level. The NYtimes had an article(posted here I believe) that the dems were kicking ass with voter registration in OH and FL. So I cant see more registered republicans than dems.

How did you get this e-mail. Can I have a copy of it. I'd like to write an editorial in my local paper about the Gallup polling company and their questionable methodology.

Thanks

Who says their self declared party affiliation doesn't change? Maybe folks just change their response as they go along and it's a valid poll question and not a "rock of Gibralter" that you can measure against.

Party affiliation isn't tattooed on our foreheads. It probably swings with the other questions.

Therefore, I would guess you cant completely adjust for it and get the real answer.

Ruy Texiera did us a favor by "adjusting" Gallup's sample partisan breakdown to what it actually IS, nationally, for their last poll. That yielded, instead of Dub being ahead by 13, the race tied at 48 apiece. Wonder what such an adjustment would yield this time...?

CNN PHONE NUMBER
404 827 2600

HIT THEM WITH THE PHONES TOO!

It's news like this that has decimated Gallup's legitimacy in my eyes for the foreseeable future. At this point, if I see a poll from Gallup I just immediately discount the results as due to poor or biased sampling.

The issue is not the polls per se, but how the VRWC media uses it to bludgeon Senator Kerry and the Democrats. It's just another tool in their colostomy bag of tricks.

That's why you have to challenge this on all fronts. Even if it works to keep goppers comfortably at home on election day, we can't afford to not counteract this. We need to work the refs too, if not more so.

When I was leaving the house this morning the CNN blurb was telling me how I was going to miss the story of how one of Kerry's voter blocks needs shoring up, and shoring up fast. That would've been women.

IT'S ABOUT PERCEPTION, we can't let this stand unchallenged. So write those damn letters. I did, and I'll do it some more.

Some pollsters, such as Rasmussen and Zogby, and die-hard "weighters". Others like Pew and Gallup do not weight by political party. Which is more accurate? We'll know on November 3.

Actually, we found out in 2000. Zogby was much more accurate at national polling than Gallup.

Posted by Erica at September 28, 2004 08:44 AM
Posted by Eric Kingsley at September 28, 2004 10:25 AM

Wow. So many great comments! Yes I hear you. I'm doing my bit. I wrote all the local TV stations to tell them to look out for this bias.

One did a stupid "the tale of two polls" that was worthless and never even addressed the Gallup problem. It DID throw doubt on Polling, so I guess it was a marginal good thing. I'll write again so that they can drop in a word that Questions the Gallup. Maybe if they say, In a CONTROVERSAL poll by the Gallup organization...
I think Gallup should be sued. That is one of the few things that the media notices these days.
If only the Gallup poll had been accused of kidnapping and killing their pregnant wife, then we would see some REAL in depth coverage!

The actual party affiliation in 200 was
39% DEM
35% GOP
26% IND

If you reweight the findings of these two polls to reflect the overweighting and allocate independents 2-1 against the incumbent, I get
Kerry Wins 52% to 44%, which is probably too wide.
But still nice to look at. ;)

Tim: "D and T, well, there was a NYtimes article, I'm sure it was posted on Leftcoaster as it was on many blogs, that showed Dems kicking GOP ass in voter registration for this year in Ohio and Florida. So you have a point, but if so many felt that way, I dont see a large number lining up to register Dem, which is what is happening."

There was a story on NPR (Day to Day with Alex Chadwick)as well yesterday regarding the Dems 'kicking GOP ass' in voter registration. Kind of
a quiet campaign.

The reason it adds up to more than 100% is that you can't ask 1.3333 people a question.

The problem is that it's not just Gallup. Most polls lately show Bush in the lead. Why do you think when you go to www.electoral-vtoe.com you see a whole lotta red and pink, and nearly no blue? So don't only blame the pollsters.

It's probably accurate. The number of people who identify with one party or another without accurately registering is incredibly volatile. Big John has done a woderful job of barely quietly opposing a little bit on some obscure technical diplomatic nuance a man who believes that God and America stand for something and are worth going out and killing and dying for. Vision thing. Quit whining about Swift Boats and the Media and all that Corporate crap. Kerry has failed to lead the nation anywhere. Dukakis redux.

According to USA Today (1/22/04), Registered Democrats make up 35.7% of the electorate, Registered GOP are 27.3%, and Independents are 20.8%. Unless Gallup is assuming dramatically different turnout rates for Dems and Repubs, they are overweighting Repubs by a factor of 1.58, and underweighting Dems by a factor of .87.

One thing still puzzles me. Gallup says its poll of "likely voters" (which is based on the above strange weightings) favors Bush over Kerry by 52% to 54%, but the poll of "registered voters" (which should not include those adjustments) shows an even more lopsided result -- 54% to 41%. Anybody got an idea how Gallup is corrupting the data here?

According to USA Today (1/22/04), Registered Democrats make up 35.7% of the electorate, Registered GOP are 27.3%, and Independents are 20.8%. Unless Gallup is assuming dramatically different turnout rates for Dems and Repubs, they are overweighting Repubs by a factor of 1.58, and underweighting Dems by a factor of .87.

One thing still puzzles me. Gallup says its poll of "likely voters" (which is based on the above strange weightings) favors Bush over Kerry by 52% to 54%, but the poll of "registered voters" (which should not include those adjustments) shows an even more lopsided result -- 54% to 41%. Anybody got an idea how Gallup is corrupting the data here?

To Just Sayin':
The reason www.electoral-vote.com shows Bush doing well is that they give a state to the winner of the last poll taken, even if the poll comes from a completely biased Republican pollster like Strategic Vision.

To Just Sayin':
The reason www.electoral-vote.com shows Bush doing well is that they give a state to the winner of the last poll taken, even if the poll comes from a completely biased Republican pollster like Strategic Vision.

I think it is great that Gallup is front loaded the sample with the GOP....great job Gallup. Then the pundits, CNN's Wolf"i would rather work at Fox"Blitzer can whine along in the haze of stupidity and have a great surprise in November.

According to USA Today (1/22/04), Registered Democrats make up 35.7% of the population, Registered GOP are 27.3%, and Independents are 20.8%. Unless Gallup is assuming dramatically different turnout rates for Dems and Repubs, they are overweighting Rebubs by a factor of 1.58, and underweighting Dems by a factor of .87.

One thing still puzzles me, though. Gallup says their poll of "likely voters" (which is based on the above strange weightings) favors Bush over Kerry by 52% to 54%, but the poll of "registered voters" (which should not include those adjustments) shows an even more lopsided result -- 54% to 41%. Anybody got an idea how Gallup is corrupting the data here?

12%?? The bias in the Gallup poll might better be described as 40%! There were 328 GOPs and 236 DEMs in this poll (40% more)... that is a SUBSTANTIAL difference in numbers, based on the total sample. In other words, nearly 40% more GOPers than DEMs responded to this poll ... so, all things being equal, Bush should be nearly 40% ahead!

...the pundits, CNN's Wolf"i would rather work at Fox"Blitzer can whine along in the haze of stupidity and have a great surprise in November.

I wonder if this is not a scheme to mute protests and outcries when shenanigans are used to distort the Nov. voting results, just as such shenanigans were used in Nov. 2000. By making it look plausible that Bush has a chance in hell, another coup becomes more possible.

Since the conventions, Democrat voters have become less enthused with Kerry, hating Bush has not been enough. The poll is consistant with why Kerry has been presenting a more 'Dean like' message since NYU. The poll is based on 'likely' voters, not just 'registered' voters. That small group of swing voters previously thought to determine the outcome of this election will not matter if the Democrat base of likely voters is shrinking.

Gallup is flawed and we should make a point to ruin their reputation following this election. They are showing to high a Bias and they can make no excuse for it. There are more registered Democrats in this country than registered Republicans. Democratic Registration has been up more than Republican Registration. Taken seperately, these facts don't say much. But taken together they prove that there is no massive movement in the electorate (or Republicans would also have an increase in registration too).

Theres only one reason for this I can think of, They are using 2002's turn-out numbers (which the media pretended didn't exist, even though they did) And they are weighting from a year in which Republican turn-out was very high and most Democrats stayed home.

A little global find and replace gives us this.

Since the conventions, Republican voters have become less enthused with Bush, hating Kerry has not been enough. The poll is consistant with why Bush has been presenting a more 'Reagan like' message since NYU. The poll is based on 'likely' voters, not just 'registered' voters. That small group of swing voters previously thought to determine the outcome of this election will not matter if the Republican base of likely voters is shrinking.

And

Bush has failed to lead the nation anywhere. George H.W.Bush redux.

Polls don't matter--message does.

At this time, neither candidate has anything honest to say about the state of the union.

As Lewis Black would say, Kerry has no ideas, Bush has bad ideas.

You look at the polls only to see whether the rest of the nation agrees with who you happen to like less on a particular day.

Polls don't matter--message does.

At this time, neither candidate has anything honest to say about the state of the union.

As Lewis Black would say, Kerry has no ideas, Bush has bad ideas.

You look at the polls only to see whether the rest of the nation agrees with who you happen to like less on a particular day.

Polls don't matter--message does.

At this time, neither candidate has anything honest to say about the state of the union.

As Lewis Black would say, Kerry has no ideas, Bush has bad ideas.

You look at the polls only to see whether the rest of the nation agrees with who you happen to like less on a particular day.

Remember that in Republican fantasy world it's always opposite day!

When they rant about the democrats they are often projecting. See if the rant could be equally applied to them and you will see how true this is.

It's a fun party game too!

fine, shoot the messenger.

But consider this. Why is it if there are nearly twice the number of registered Democrats to registered Republicans the races are so close? Becuase 80% of Republicans vote on election day, compared to 40% of the Democrats. Pollsters try to determine who's actually going to vote and this poll reflects their findings.

And it's not that Republicans are 'suppressing' the Democrat vote, its apathy among the Democrats. Thus Kerry MUST re-energize his base of voters although risk loosing those in the middle, then try to get the swing voters in the final week(s) and he only has a month to go!

Cynic:

Who says that there are twice as many registered Democrats as there are registered Republicans? And again, where have the GOP ever had a 12% edge over the Democrats?

Cynic

I'm not shooting the messenger. I'm just saying the message that you are conveying is wrong. It might not be your job to analyse it. You are simply a messenger.

They say this in the stock market ads all the time.
PAST PERFORMANCE IS NOT NECESSARILY INDICATIVE OF FUTURE RESULTS.

This is not the year 2000. This is not 1996.
I like some of the things Zogby has said and The Left Coaster posted about how many different variable they are trying to juggle and if they get a few of them wrong they will have significant differences.

I spoke to a guy from Bell Labs once. He said, "We always talk about stuff 5-10 years out. If we get it wrong NOBODY remembers!"

I used to work with market researchers. Rarely did anybody hold up their past predictions and say, "You got 80 percent of them wrong!" Yet people kept buying the reports to support their own ideas.

Now with the search capability of the internet we have the ability to go back and see, yep They got it wrong. But when we do that, it is considered "A historical document". And they can play the same game I did, "Well this year it is different."

Aw... the post is irrelevant. If Gallop stops 767 people and 40% turn out to be GOP and 30% turn out to be Democratic, why is that some sort of Gallop plot? It's the POINT - it's a POLL. Are they supposed to poll on the curve? They didn't get enough Dem's in this sampling of 750 so people so let's drop a few undecided and GOP's to even it out? Or, poll an infinite number of people until, hopefully, by chance they get an even number of Republicans and Democrats? What would they expect the result to be then? Hmmm...40% for Bush, 40% for Kerry, 20% for Nader or undecided? This article is, unfortunately, a greater evidence of bias than the media. Enough already! There's enough real issues out there without having to manufacture more. It only looks bad on us. The only thing that would be evidence of manipulation was if Gallop polled within specific populations to influence their results. However, whatever you think of Gallop and their methods, they are in the position they are in because they DO attempt to return reasonable results - They'd only the ones to look stupid if they didn't. (I'm not an employee.) However, my guess is that this was done randomly by phone. That's what we did when I worked for them when I was in college.
Not EVERYTHING is a PLOT. Sheesh.

There's a new TIPP poll out today (Tues.) that has kery up by 1 head to head. Check it out!

There's a new TIPP poll out today (Tues.) that has kery up by 1 head to head. Check it out!

Someone seriously needs to start a website devoted to getting people to stop watching CNN. There could be like a pledge that people sign electronically and then email to CNN. It should be a splashy campaign designed to educate viewers on the evils of the channel, just like News Hounds is devoted to calling Fox out on the their crap.
Just as a fat woman goes on the Atkins diet and swears off Twinkies, I would swear off CNN for good and I would be the first to sign an electronic pledge not to watch the network. Emailing such a pledge would be icing on the cake. Rubbing it in their faces would be of great joy and comfort.
Anyone want to start such a site?

John D: You don't understand polling.

Gallup didn't just call up 767 people at random then announce the raw results. They normalize the data using weight factors to make sure their sample is representative of the population.

The problem is either that they are intentionally using a faulty weighting percentage for party ID, or they are intentionally NOT using any weighting percentage for party ID.

Whichever is the case, they are intentionally overstating Bush's support. To be blunt, they are lying.

It's appalling how bad some of you are at simple arithmetic. And it's amusing that some of you think you are smarter at the business of polling than a company that has been at that business for fifty years or more.

You might want to have a friend lock up all of the sharp objects in your home sometime before November 2.

I have a question. Why is it that whenever a poll favorable to the President is released, you guys call it "suspicious" while, at the same time, you excitedly tout the methodology of those polls that are favorable to John Kerry?

Oh, wait a minute. What was I thinking? There have been no polls favorable to John Kerry. Sorry, my mistake.

Greendale:

Explain why a credible poll would use a sample of likely voters that has 12% more Republicans than Democrats.

Scottbrattle: Same question to you. And where have folks here been "excitedly tout(ing) the methodology of polls" favorable to Kerry?

We just want simple answers to simple questions.

Has anyone compared Gallup's polling of the presidential race with that of other races?
For example, does Gallup have individual party ID models for each state which they use to weight the polling data for Congressional races?
If so, it would be very interesting to know what percentage of voters in CA. Gallup thinks are Republican, for example.
Now, even supposing Gallup doesn't weight congressional races like that, one can still compare their polling of various state races to their presidential polling. If you wanted to crunch numbers, you could probably make an educated guess as to the answer of my first question.

What I'm getting at is can you look at all Gallup's polling for Congress and the presidential race to determine if Gallup is merely guilty of using poor modeling, or has crossed over to outright unethical behavior as a means of boosting Bush?
Thoughts?

Pew came out with a new poll today. Instead of weighting by party registration, they get people to describe their beliefs (conservative R, moderate R, indep, moderate D, liberal D). It's an even split of Rs to Ds. The results are not pretty for our man:

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

Polls can be instructive, as long as you're not so angry by their results that you willfully blind yourself to what they're saying and instead try to discredit the messenger. That's an old Fox News and Repub tactic. No, go get your fellow Kerry supporters motivated and stay motivated, volunteer if you can. Because we are looking down the barrel of a gun here.

Anyone else think that this might blow up in the GOP's face, that it might convince joe republican that his vote isn't needed, if Bush is so far ahead?

Of course the converse is what the GOP is hoping for.

here's what I find disturbing. Soto had to specifically request the party ID information from Gallup; it's not anywhere to be found on their 3 pages of results at gallup.com. That's more than a little dishonest, in my opinion.

I work in psych research and for every study we do, we are required to include a section that reports the demographics of our sample. Why is Gallup not required to publish age/party ID/etc information after their results (especially when they don't weight this information to adjust their results)?

I do not believe any of the polls,but the media muist do something to generate interest or to make headlines. However,the media has been wrong so many time it seems they are not doing there homework for any number of reasons. I know this they will all look bad if what they wrote turn out to be wrong like a lot of other things.

Jeez, Smith, if the media needs to do something to generate headlines, they could try reporting the news and doing some investigating - ok, I know that's not gonna happen.

Repeat this as often as necassary or whenever you find yourself scratching your head in disbelief at whatever the latest media indignity is:

The media is not my friend.
The media is not my friend.
The media is not my friend.

Got it?

The media is not, repeat NOT, going to help Kerry. Find a copy of Outfoxed and Fahrenheit 9/11 and watch them a few times.

The media wants to proclaim Bush King and get back to the Peterson trial, in between their more lucrative commercials.

Meanwhile, all of the polls are all missing several vital pieces of information:

They do not include ANY of the millions of newly registered voters, and many of those are people voting for the first time because they are appalled by Bush.

They do not include anyone who uses only their cell phone. Cell phones are not part of any polls. How many young adults only have cell phones?

They do not include people with caller ID who do not want to be bothered.

And, some, like Gallup, deliberately skew for certain biases.

The bad news is that some people will believe polls like Gallup's and want to vote for whoever they perceive to be the 'winner'. Stupid way to pick a candidiate, but there you are.

The good news is that this race will be decided, not by the polls, but by whoever gets the most votes out, because of all of the new registtered voters. It ain't over yet.

Cynic: Democrats may be less enthused with Kerry, but their hatred for Bush is stronger. I would vote for anyone right now if it meant unseating the evil man who sits in the WH. That same attitude is held by everyone I know.

And it's amusing that some of you think you are smarter at the business of polling than a company that has been at that business for fifty years or more.

Silly comment. You assert that just because a company has experience means that it is impervious to error? Easily disproven. Consider, there are now vast gaps in the results of different, highly experienced polling companies sampling the same population. Clearly some sort of fundamental errors are occurring.

For the record, Gallup's reputation has suffered greatly during the past decade or two. Once the pioneers in the field, they've failed to keep up with current innovations and lost the aura of "independent arbiter" that they once had..

Furthermore, some of us here have lots of experience evaluating statistical analyses. Even in professional works, fundamental statistical errors are all too common amongst those who should know better. Even more so if one has a bias, as one is tempted to choose data to fit the desired results. Which is why the system for peer review exists.

I do not know what the problem is. However, I have no clue why you guys are obsessing with the national polls. Just focus on the state polls. In CA Kerry is up by 13, same margin in NY, about the same in IL, up by 40 in MA and RI. The only state that is close from the previous solid ones is NJ where Kerry is up by 3. If Kerry wins CA by 16, NY by 20, IL by 15, MA by 40 and NJ by 5 and carries WA by 8 and MI by 8, he will still win the popular vote even if Bush carries WI and IA narrowly and OH by the same margin. You guys need to ask Gallup to justify why Bush is ahead by 1 in FL while ahead by 13 nationally. Does it make any sense?

One other sign of sample skew: Some pollster (and I thought it was Gallup) also asked who got their vote in 2000. Their sample showed a significantly (near 10%??) greater count of prior Bush voters. Since we know that the actual popular vote wasn't anything like that, something smells like its past its sell-by date.

-dp-

Jeff - it's tricky judging a sample by past vote recall.

In the UK pollsters don't weight by party indentification, instead some of them try to make sure samples have a fair balance of political views by weighting by past vote.

Experience has shown them that when people answer polls they are very bad at remembering their past vote - in short, the more socially responsible answer is to say that one voted, so people who didn't actually vote tend to lie and say they did, people tend to say they voted for the winner, people who vote for third party candidates tend to forget they did, and people tend to say how they'd liked to have voted with hindsight rather than how they actually did (the solution here is to weight to an adjusted past vote, taking account of false recall.)

The same thing happens in the USA - the classic example was post-1963. In reality JFK got 49.7% of the vote, but after his assasination two-thirds of people surveyed claimed they voted for him!

Samples reporting recalled votes that are out of line with historical reality are not necesarily an indication of sample bias, most of the time it is people "misremembering" how they voted.

A new poll puts Kerry ahead by 1 poll.

So, why is CNN swinging right? Because they are rapidly losing viewers so they're trying to appeal to the conservatives.

Anyway, we're trying to point this all out at US Nation. 1 million uniques per day.....

http://www.atlantanation.com Atlanta
http://www.cobbonline.com Marietta, GA
http://www.cherokeenation.net Canton, GA
http://www.fultonnation.com Alpharetta, GA
http://www.gwinnettnation.com Lawrenceville, GA

http://www.clnation.com Chicago
http://www.cincynation.com Cincinnati
http://www.nynation.com New York


http://www.ncnation.com North Carolina
http://www.jaxnation.com Jacksonville
http://www.tbnation.com Tampa

http://www.dallasnation.com Dallas
http://www.houstonnation.com Houston
http://www.forthworthnation.com Fort Worth

http://www.pnation.com Phoenix
http://www.lasvegasnation.com Las Vegas

Hmmm...
The Washington Post has it Bush 51% vs. Kerry 44% http://www.washingtonpost.com/wp-srv/politics/elections/2004/charting.html
Gallup has it Bush 52% vs. Kerry 44%
http://www.gallup.com/election2004/
That bunch of right-wingers at the Washington Post must be in league with Gallup.
What a larf… you all are nothing more than a bunch of left-wing conspiracy theorist. Look out... the trilateral commission might be watching. Hahaha

Say what you like, the bottom line is it's Kerry and his mismangement of his campiagn ie hiring the wrong people, taking bad advice. Taking different postions on a weekly basis. You Dems have no one to blame but Kerry. ALL polls show Kerry losing ground on key issues. Esp. keeping this country safe. It's not so much people like GW it's they can't take to Kerry. esp. when he comes out in that monotone voice and sounds like he's talking down to people even though i'm sure he's not. The bottom line is John Kerry is just not a likeable person. Except for Ted K. he doesn't have any friends. I'm sill convinced the power brokers for the DEM party will make sure Kerry loses so the Clintons can move back in in 2008. If Kerry wins Hillary cannot run in 2008.

You really ought to research how polls work before posting this. By trying to get people upset about Gallup without knowing their actual methods, you're doing a worse disservice than you claim Gallup is. You seem to be clinging to the mythical "call 1000 people, ask them who they vote for, then divide by 10 to get percentages" method.

You know when you read the results of a medical study and they say "group X is more likely to get disease Y than group Z, even after taking into account factors A, B, and C"? Statisticians take these factors into account. That's what they do. (No, it does not matter that they got a few more republicans than democrats in the poll. No, it does not matter that many college students have only a cel phone.)

Women are more consistent voters than men, but you don't know or care about the male/female ratio of these polls, or any other factor. You're just playing politics here: not really upset about the accuracy of their polls, but just trying to push your agenda by getting everybody worked up.

I want the conservatives out of the White House, too. But don't resort to unfounded accusations and lies to do it. Then you're no better than the republicans are.

I have 2 thoughts on all this:

If the polls are really a cover for stealing the election, then if Kerry wins the election despite all the attempts to make him lose, the Democrats will be accused of trying to steal the election just as they were in 2000. After all, the polls all told us that Bush was the people's choice!

(And if Kerry takes office, of course, the Republicans aren't going to go sulking back into the night. We're living in interesting times.)

But what if all this talk about disinformation is wishful thinking? What if Bush takes office by getting elected? Are we prepared for what to do after that?

d.e.
then how do you explain this
Actually, it's what Gallup doesn't do that is at the heart of the debate. The polling firm does not adjust its "pool" of voters to add or subtract Republicans or Democrats in an effort to mirror those parties' estimated make-ups.

Apparently, they do call up 1000 or so people and divide by 10.

Rush uses Gallup consistently. You can usually spot Rush in a lie within 15 minutes of his show startup. Sometimes not even that long.

Other permutations:
Sample size way too small for representative national sample.
Skewed Repug. constituency -- too big.
Skewed Undecided constituency -- too big.
Skewed Likely Voter -- left out cell phones.

Soooo. . . Gallup is just a FOX poll arm. You can trust Gallup about as much as you can trust FOX -- which is, not at all.