Leftcoaster reports gross Gallup GOP LV overweighting!

[flpoljunkie]

http://www.theleftcoaster.com/

Monday :: Nov 1, 2004

Gallup And Fox (Yes, Fox) Have Bad News For Bush

As we know by now, Gallup released its last national poll last night, and as we expected, in order to preserve its credibility Gallup’s last poll amongst LVs shows last week’s 5% lead for Bush (51%-46%) now shrinking to a 49%-47% lead for Bush over Kerry. Amongst registered voters, Kerry has a 48%-46% lead.

But toss out those results by Gallup. Why? Because take a look at the composition of the LV and RV samples by party ID that Gallup used last night in their poll. I just got the internals from Gallup, and the LV sample they used had a 5% GOP advantage over the Democrats, and Kerry still almost tied Bush.

1573 Likely Voters
1559 In Sample
Bush Leads Kerry 49%-47%

GOP: 624 (40%)
Ind: 389 (25%)
Dem: 546 (35%)

When compared to the 2000 exit poll results, Gallup under sampled Democrats by 4% and over sampled Republicans by 5%, a nifty 9% swing. If the sample was reweighted to more closely follow the 2000 exit poll party ID breakdown of 39% Dem/35% GOP/26% Ind, can you imagine what kind of lead Kerry would have?

more...

Update: Fox's last poll shows the same thing. Call in the vans, Laura.

[deadparrot]

1. Good, it sound like we're winning the indys, then.

[Confident]

2. Can you re-weight vs. 2002 turnout (last national election)?

[Pithy Cherub]

3. The likely voter scenario

is misnamed. This year registered voters are MORE LIKELY to turn out so there should not be a scenario based on previous empirical evidence. This will be part of the 'lessons learned' media/pollster mea culpas that should be used to further impugn their integrity! Save us from "Gallup" and "the most trusted name in news".