The Final Nail in the Gallup Poll's Coffin

[jefferson_dem]

Total Weighted Sample: 557 Likely Voters
(2000 exit poll actual results in parentheses)

By Political Ideology:
Conservative: 41% (29%)
Moderate: 41% (50%)
Liberal: 18% (20%)

Party ID:
GOP: 39% (35%)
Dem: 35% (39%)
Ind: 25% (27%)

Income:
Over $75,000: 32% (28%)
$50-75,000: 16% (25%)
$30-50,000: 26% (24%)
$20-30,000: 11%
Under $20K: 9%

And if this wasn't bad enough,

Race:
White: 85% (81%)
NonWhite: 15% (19%)
Black: (a subset of NonWhite) 8% (10%)

Not only does Gallup not weight by Party self-identification, a characteristic that is debatably a demographic, they do not weight by race or income either, characteristics that are unquestionably demographics.

Even without the Party ID debate, Gallup is clearly using an unrepresentative sample. To use such shoddy methodology, have such a tight connection to major media outlets, and simultaneously show Bush ahead by more than any other poll, is nothing short of an irresponsible in-kind contribution to the Bush campaign. Shame on you Gallup. Shame on any news outlet that still uses their polls.

http://www.mydd.com/

[kikiek]

1. I think there will be a lot of conservatives staying home this year.

[kostya]

13. Except where those hateful anti-gay measures are on the ballot

That will draw the right-wingers out by the droves.... though... on second thought most of those have such big yes margins that they may stay home figuring they'll pass without their vote and they can't stomach voting for Bush. :puke:

[tritsofme]

2. Gallup does weight its raw sample against

figures that can be checked against the census.

[jefferson_dem]

4. race, gender, age. But surely failure to try to correct for obvious bias

on any political indicators (party affiliation) will, inevitably, produce biased results.

[tritsofme]

7. Its the same method they have always used

And they make the claim that party identification is fluid and that people identify more with the "team" that is percieved to be winning.

They say that their random sample should be representative of the entire population.

If what they have seen over the past few months is true, then we could have a very scary situation on our hands that many pollsters miss completely because they just reweight to "correct" the bias that they see in their sample.

Look at this historical data from Gallup. They seem to see a huge shift in party idenification over to the GOP through the past few years, and these numbers have never been measured any near this hight before, the only comparable election is 1988 when Gallup measured self party idenification at 32% each. And they have asked the exact same question for decades in terms of party idenification.

This is not my experience, this is not what I have seen around, but this is what Gallup has been consistantly measuring over the past few months.

This is latest data I have access to:

September Wave 1
Question: D9 Field Date: 9/24/2004-9/26/2004
In politics, as of today, do you consider yourself a Republican, a Democrat, or an Independent?
Mean: N/A Total N: 1006


% N

1 Republican 38.70 389

2 Democrat 31.36 315

3 Independent 26.27 264

4 OTHER PARTY (VOL) 0.18 2

5 DON'T KNOW 1.85 19

6 REFUSED 1.65 17



This is their last poll from 2000:

General Election Tracking Poll Week 9
Question: D7 Field Date: 10/30/2000-11/5/2000
In politics, as of today, do you consider yourself a Republican, a Democrat or an Independent?
Mean: N/A Total N: 8197


% N

1 Republican 28.66 2349

2 Democrat 34.38 2818

3 Independent 28.80 2361

4 OTHER PARTY (vol.) 0.83 68

5 DON'T KNOW/REFUSED 7.33 601

Last poll from 1996:

General Election Tracking Poll 1996 week 9
Question: D7 Field Date: 10/28/1996-11/3/1996
In politics, as of TODAY, do you consider yourself a Republican, a Democrat or an Independent?
Mean: N/A Total N: 6114


% N

1 Republican 28.79 1760

2 Democrat 35.80 2189

3 Independent 27.74 1696

4 OTHER PARTY (vol.) 0.86 53

5 DON'T KNOW/REFUSED 6.80 416

Last poll from 1992:

Presidential Election October Benchmark
Question: D10 Field Date: 10/23/1992-10/25/1992
In politics, as of TODAY, do you consider yourself a Republican, a Democrat or an Independent?
Mean: N/A Total N: 1602


% N

1 Republican 29.58 474

2 Democrat 36.39 583

3 Independent 31.07 498

4 OTHER PARTY (vol.) 1.13 18

5 DON'T KNOW/REFUSED 1.84 29

Last number from 1988:

Voter Perceptions of the Presidential Campaign
Question: qn19 Field Date: 10/20/1988-10/21/1988
In politics, as of today, do you consider yourself a Republican, a Democrat, or an Independent?
Mean: N/A Total N: 1641


% N

1 Republican 32.88 539

2 Democrat 32.79 538

3 Independent 31.56 518

4 Other 0.30 5

0 Don't Know 2.48 41


Last numbers from 1984:

The Gallup Poll #244G
Question: qn902 Field Date: 10/26/1984-10/29/1984
In politics, as of today, do you consider yourself a Republican, a Democrat, or an Independent?
Mean: N/A Total N: 2711


% N

1 Republican 33.68 913

2 Democrat 37.60 1019

3 Independent 25.84 700

4 Other party 1.63 44

0 Undesignated 1.25 34

[jefferson_dem]

9. I appreciate the argument about fluidity but i just think it's more valid

to see that the sample most accurately reflects the "best guess" of what the partisanship of the population actually looks like. As you say, that's their goal with the "random sample." Problem is that in going with pure randomness, they are missing the boat. They are approximating "likely voters" and exit polls over the past half century indicate that more self-described Democrats turn out on election day than Repukes. Recent polls show that Dems still outnumber Reps. Look at their own digits prior to this year.

Anyway, i don't personally buy into the "agenda" theory with Gallup, only that their methodology is flawed given today's environment (cell phones, whatever).

[tritsofme]

10. But has so much changed in four years

to make their 60 year old model invalid?

[jefferson_dem]

11. When their model no longer reflects the same 60

years of political reality (i.e. actual partisan turnout on election day)...apparantly so. I believe it's time to re-tool.

[tritsofme]

14. We won't know whether or not it is reflecting reality

until election day.

But I think at least one school of thought is going to have egg on their face.

[sonicx]

16. do you think election day turnout

will be 39% republicans and 31% democrats?

[tritsofme]

17. It doesn't seem logical to me

However, coming from someone like Gallup is enough for me to sit up and take notice.

But even some of Gallup's historical final election polls that measured party indenification came very close to accurately predicting the final status of the race did not correspond to the party affiliation measured by exit polls.

[The Magistrate]

3. That Is More Than A Nail, Sir

That is sufficient shovelsful of earth to fill up the grave....

"LET'S GO GET THOSE BUSH BASTARDS!"

[underpants]

5. sssssssssssshhhhhhhh

the RW mouthpieces are trying to, at once, tell the faithful that W is a WINNER and not to worry AND tell them to go vote-walking a thin line.

[jefferson_dem]

6. Right. We must keep all this on the way down low. That will make it so

much more fun on election day to watch their utter dismay, as reality sets in! Buh by chimp-boy.

[atre]

8. Good God

The median per capita income in this country is currently a shade less than $30,000. In this poll, the bottom 50% of the population accounts for 20% of the respondents. Likewise, the top 50% account for 80% of the respondents.

I understand more wealthy people are more likely to vote, but there is no way that the median salary of a voter is twice as much (the median here apparently falls between 50,000 and 70,000) as the median salary of the average American.

[Awsi Dooger]

12. I should have recent figures but I can't find them

So many darn old Zip and Jazz disks but I can't locate the Excel file with all my median voter figures.

Off the top of my head, I remember people with incomes above $75,000 are exactly twice as like to vote as individuals earning below $10,000 -- 76% to 38%.

I want to say the median voter income is mid-to-high 30s, something like $38,000, but don't hold me to that. I know I have separate files with median voter income, and family income for the median voter, and I might be confusing the two.

[featherman]

15. median family income 2002

http://www.census.gov/statab/www/part3.html

In constant (2002) dollars (dol.)

39,949 39,931 43,848 42,900 42,409

(1990 1995 2000 2001 2002)

As you can see it has dropped under Bush in terms of constant dollars and barely above what it was in 1990.

[TankLV]

18. But what about all those poor persons struggling on a mere $200,000?

After all, those are the folks that bush considers "middle class"!!

And quite frankly, Kerry blew it BIG TIME when he included these rich a-holes in his tax breaks!

I make just over 75K and consider myself not so bad off - even upper middle class!

God I'd LOVE to make over 100K!

I know many professionals who make 36-50K!