New Rasmussen Weighting Targets: 38.7% Dem 33.6% GOP, 27.7 Indy

[The Vorlon]

New Rasmussen Weighting Targets: 38.7% Dem 33.6% GOP, 27.7 Indy

Rasmussen is modestly reworking his dynamic party ID weighting method.

He is adjusting targets more frequently (weekly versus monthly)

He is also shortening the timebase, using interviews over the last 6 weeks (+/- 22,500 interviews) versus the last 3 months (+/- 45,000 interviews)

http://www.rasmussenreports.com/public_content/politics/election_20082/2008_presidential_election/new_rasmussen_reports_party_weighting_targets_38_7_democrat_33_6_republican

As a matter of interest..

In August Rasmussen reported the following based on 15,500 interviews...

http://www.rasmussenreports.com/public_content/politics/mood_of_america/party_affiliation/party_affiliation/summary_of_party_affiliation

GOP 33.2% x 15,500 => 5146 republicans +/-
Dems 38.9% x 15,500 => 6030 Democrats +/-
Indy 28.0% x 15,500 => 4340 Independants +/-

Rasmussen data over the last 22,500 interviews, including all of August, plus two weeks in september, was as follows...

38.7% Dem x 22,500 => 8708 Democrats +/-
33.6% GOP x 22,500 => 7560 Republicans +/-
27.7 Indy x 22,500 => 6232 Independants +/-

Substracting one set of numbers from the other will thus yield the partisan ID for the first two weeks of September...

Dems = 8708 - 6030 => 2678
GOP  =  7560 - 5146 => 2414
Indy =  6232 - 4340 => 1892

(The numbers total 6984, not 7000 due to a bit of rounding error)

Dems => 2678/6984 => 38.3%
GOP => 2414/6984 => 34.6%
Indys => 1892/6984 => 27.1%

The gap of 3.7% between GOP and Dems is ~~roughly~~ what it was in the 1992, 1996, and 2000 elections.

2004 was unusual in that GOP and Dems were essentially tied in party ID among those who actually voted.

Just for fun..

This is what todays Rasmussen data looks like reweighted under the three ratios discussed above...

[muon2]

This makes a lot more sense in the final weeks of the campaign. I'd be happy with a correction every two weeks, but if they've got the data for weekly adjustments, that should be for the better.

[Ronnie]

I wonder if the weight was applied in today's tracking poll.

[The Vorlon]

This makes a lot more sense in the final weeks of the campaign. I'd be happy with a correction every two weeks, but if they've got the data for weekly adjustments, that should be for the better.

I think every two weeks is about as fast as Rasmussen can actually go..

he collects the partisan ID information in an entirely seperate survey he runs (500 a night) that is part of his never ending data collection on economic indexes.

It is correct that he collects the data seperately, because if you ask as part of the horserace questions the interaction of the two surveys tends to taint things a bit...

Every two weeks is 7000 interviews so the noise level is just a shade over 1% 19/20 which should be good enough.

[The Vorlon]

I wonder if the weight was applied in today's tracking poll.

"For polling data released during the week of September 14-20, 2008, the new targets are 38.7% Democratic, 33.6% Republican, and 27.7% unaffiliated"

I assume that includes today then..

This gap of +5.1 to the D side is still actually 1.4% more democratic than his data for the first 2 weeks of September, so IF (huge if) the modest shift in party ID holds, his current weighting method will stillunderestimate the GOP by about 1% or so.. not a huge matter... It will be interesting if the modest shift back to the GOP is temporary or fades....