Gallop poll methodology

[Poiuyt]

Has Gallop ever given their rational for oversampling Republicans? Do they feel that Repubs are more likely to go and vote in November than Democrats?

[rocktivity]

1. Their methodology is simple

They give money to the Bush regime. The Bush regime gives money and business to them. So it's in their best interests to do all they can to keep Bush in office.

:headbang:
rocknation

[MallRat]

2. They say that self-identification of party is drifting towards GOP.

Self-identification of party affiliation is always in flux. There's always going to be a handful of people who are... well, for lack of a nicer term, they're frontrunning twits.

If Kerry is up, they'll say they're Democrats; if Bush is ahead, they'll call themselves Republicans. There are also independents who come off the fence and go right back on the fence during an election year.

But that said, a 12-point GOP weighting advantage is absolutely insane, especially considering the 39-35 Democratic advantage in the actual vote in 2000.

I'd love to see someone from Gallup justify these wacky numbers. 10% of the electorate didn't suddenly decide to start calling themselves Republicans.

-MR

[tritsofme]

5. They're not even talking about party registration,

they only ask this question at the end of each poll:

In politics, as of today, do you consider yourself a Republican, a Democrat, or an Independent?

Their view is that there is a subsection of the voting population that will identify with the party that is viewed as "ahead" at the current time, and that party ID is not set in stone and can fluctuate throughout the campaign.

They don't weight their polls for party ID, only for information that can be checked against census data.

[On the Road]

13. I Heard Zogby Say That Polls Should NOT be Weighted

for party self-identification, which makes me think there's a legitimate reason there. During this election, however, the distortion from weighting for party ID seems outweighed by the problems of NOT weighting.

[tritsofme]

18. I don't think Zogby would have said that

being that he is the king of weighting and special sauce.

[On the Road]

20. I Heard Zogby Interviewed on NPR a Year or Two Ago

and when he said this, it struck me. He absolutely believes in weighting. He just didn't think party self-identification should be one of the weights.

I am perfectly willing to believe that this might have been a good idea in the past. This year, however, the polls results clearly don't reflect reality. So it's possible he's either changed his mind or favors some other adjustments to weighting that would have a similar effect.

[dolstein]

6. Relying on 2000 to predict 2004 turnout does't make sense either

Your chief argument against the composition of the Gallup sample seems to be the partisan make-up of the electorate in 2000. Please explain to me why the composition of the 2000 electorate is a better predictor of the composition of the 2004 electorate and the self-identification of Gallup's 2004 polling sample.

[MallRat]

9. Is there anything that warrants a +16 swing in favor of GOP affiliation?

I'm not in favor of strictly applying 2000's final Dem/GOP vote numbers to raw data collected in a 2004 poll. Maybe the numbers are closer to 37-37, or 36-38, or whatever.

But is there anything, anything at all that would explain a 16-point swing in 4 years? With the electorate more polarized than ever, with Democrats more motivated to kick out a Republican incumbent, why should there be a 16-point swing?

-MR

[lancdem]

15. Historically, more Dems vote in presidential elections than Repubs

In 2000, they had a 4-point edge, in 1996 a 5-point edge. Is there any reason to think 2004 won't be similar? Hell, in 1994 when the Repubs made huge gains in the mid-terms, they didn't have anything close to a 12-point party advantage. Gallup also was way off in 2000.

[On the Road]

3. They Don't Oversample Republicans

They just don't normalize the raw data for party identification. There's a certain rationale for doing this, namely that party ID fluctuates. More people identified themselves as Republican while Reagan was most popular; more people identified themselves as Democrats when Clinton was most popular.

The problem is that it's difficult to believe that this phenomenon accounts for the the poll results. There are other things going on, such as who is reachable by telephone, and they are having a bigger effect than ever on the results.

[rocktivity]

7. What you and Mallrat say would be make sense

If Gallup's numbers were consistent with everybody else's. At http://www.pollingreport.com, there's only one poll besides theirs that had Bush leading by more than five points.

:headbang:
rocknation

[On the Road]

12. Gallup May Use Different Ways of Getting in Touch with People

even something as simple as time of day might make a difference. People at home in the middle of the day are a different group and would poll differently. Don't know which way, but Gallup isn't necessarily cooking the numbers just because their results are outliers. They may just be using dated methodology and not adjusting for it.

[GainesT1958]

4. My understanding (from a REAL poll-junkie colleague) is that

They claim to base their voter political demography on the 2002 election's turnout--which had an energized Repub base. (Thus making it a complete fallacy to replicate for this election, of course).

What gets me about this latest "survey" of theirs is that they INCREASED the Repub sample to 43%, and DECREASED the Dem. sample to 31%. Ruy Texiera had a very good "adjustment" of their last sham--er, poll:eyes:--by realigning the per centages to the actual party registration breakdown nationwide. It turned a 13-point Dub lead into a 48-48 tie. Hope Ruy does it again---wonder what it would show for their LATEST result!?:eyes:

His site, of course, is www.emergingdemocraticmajority.com

:kick:

B-)

[tritsofme]

8. If you want to reweight back to 2000 exit poll data, its a tie.

(which I don't believe is particularly valid)

bush 48.93% to Kerry 47.21% among registered voters.

Here is the partisian data from the latest poll.

Kerry Bush
R 6% 93%
I 46% 48%
D 85% 10%

Assuming D39% R35% I26%


I don't know what your buddy was talking about with the 2002 elections, Gallup uses a random sample and does not reweight for party ID.

[wadestock]

10. If you 've been listening to Randi on Air America....

she's been saying that we will see a HUGE increase in total Democrats this time around....owing to specific efforts the dems have made to bring in new voters in poor neighborhoods, etc.

The problem is...the trumped up polls have been recently changed, with the expressed intent to show a MOMENTUM for Bush. There is probably solid statistical evidence to support doing this right now for the sake of promoting a win in Nov. Perhaps those doing the fixing have actually left this "cushion for manuevering" in order to do this exactly when they needed to.

This is particularly where our media has failed us. In the old days, the good ol 60 minutes would be on this in a flash. But now the country is confused, the media is corrupt, and the power base is pushing W along with every tool they can find.

[dumpster_baby]

11. I dunno, the media has always been corrupt

At least with respect to foreign policy. I mean, look at how they cooperate with govt to give a false picture of leftism abroad.

But regarding the polls, in 2000, notice how they showed Bush leading until the last day or so, when they showed Gore closing on him. I sure would like to see a really good analysis of the timeline of poll results and what Gore was saying during those final days of that campaign.

[CityDem]

14. Best approach / methodology

Ignore Gallup polls.

[foo_bar]

16. ignore all polls

baseball stats are more useful in conversation.

[uhhuh]

17. Here's an Idea

Gallup goes through all the busywork of calling people an recording the surveys, and then pulls whatever numbers they feel like out of their ass and submits those as the results. They have one or two people make the numbers add up, and present the RW spun results as statistically representative of the current will of the voters.
Maximum of three people needed for this scam, maybe less.

Who verifies their results? Do they have each survey reviewed by an outside source, or just throw the numbers out there?

I went and got my answer from the source:

"Gallup generally provides written analysis of our own polling data. But we also provide ample opportunity for the press, other pollsters, students, professors and the general public to draw their own conclusions about what the data mean. The results to all Gallup surveys are in the "public domain" - once they have been publicly released by us, anyone who chooses may pick up the information and write about it themselves. The survey results are regularly published in the major media, in theGallup Poll Monthly, and on several electronic information services such as Nexus, the Roper Center and the Internet. We also make the raw data available to researchers who want to perform more complex statistical analysis. In addition to the exact question wordings and current results, Gallup reports trend results to all questions that have been asked previously so that even the casual observer can review the current results in context with public opinion in the past."

http://www.gallup.com/help/FAQs/poll1.asp


They let people review their RESULTS, but they don't have the results reconstructed by an outside agency.
I thought so.

[depakid]

19. The bottom line is that Gallop is dishonest

and agenda driven. Their purpose in presidential polls and others that they do on social issues is mean TO DRIVE PUBLIC OPINION, not to reflect it. They have a long and sordid history of this. So it wouldn't matter what "rationale" that they gave for their consistently skewed results. It would be like Judy Woodruff and Leslie Blitzer saying "we consistently bad mouth Democrats and progressive causes and promote right wing ones because...."