Monday :: Oct 25, 2004

Today's Gallup Poll Based On Even Larger GOP Bias Amongst Likely Voters

by Steve

With no concern apparently for its credibility, Gallup has just trotted out its latest CNN/USAT/Gallup poll of likely (LV) and registered voters (RV). All I can say is that I give them credit for consistency and not being concerned with looking stupid should their LV assumptions get blown apart next week.

CNN just reported that Gallup’s latest poll for them and USAT showed Bush with a 5% lead over Kerry amongst LVs, 51%-46%. This was down from the 8% lead amongst LVs that Gallup claimed that Bush had last week, 52%-44%. But, as always with Gallup, there was a reason why their LV and RV results showed leads for Bush.

They continue to assume questionable advantages for the GOP over the Democrats amongst both LVs and RVs, and have actually increased those questionable advantages this week over last week.

Likely Voters

Poll of October 14-16
Reflected Bush Winning by 52%-44%

Total Sample = 788
GOP: 296 (38%)
Dem: 278 (35%)
Ind: 211 (27%)

Poll of October 22-24, 2004
Reflected Bush Winning by 52%-46%

Total Sample = 1195
GOP: 466 (39%)
Dem: 406 (34%)
Ind: 321 (27%)

Note that Gallup has used a sample today that turns a 3% GOP advantage in their LV sample last week into a 5% advantage. Yet Bush's margin went down.

Registered Voters

Poll of October 14-16
Reflected Bush Winning by 50%-46%

Total Sample: 942
GOP: 348 (37%)
Dem: 327 (35%)
Ind: 259 (27%)

Poll of October 22-24
Reflected Bush Winning by 49%-48%

Total Sample: 1461
GOP: 542 (37%)
Dem: 500 (34%)
Ind: 411 (28%)

Here again, Gallup uses a RV sample this week that actually enlarges upon the GOP advantage they used for last week’s poll. I guess Gallup is aware of some flood of GOP registrations and no offsetting surge in Democratic registrations that the rest of us have missed. But note that even with this larger GOP bias in their RV sample, the 4% gap in Bush’s favor has shrunk to 1% today.

So with next Monday being the last Gallup poll before Tuesday’s election, and with CNN’s and USAT’s credibility also riding on that poll, do you think Gallup will stick with this LV methodology and risk the damage to the credibility of all three, or will next week’s Gallup LV results suddenly show a tightening race or even a Kerry lead?

And what do both of these results say about the race if Gallup used samples with even larger GOP biases this week, only to see Bush's leads actually go down in both?

Update: Ruy Teixeira of The Emerging Democratic Majority Website has noticed a flaw in the RV sample party ID information given to me by Gallup today that apparently becomes clear when you use the party support percentages that Gallup shows for this poll on its website. I am checking with Gallup now to see if they gave me information that needs to be corrected.

Steve :: 1:10 PM :: Comments (31) :: Digg It!