2002 Poll Biases Suggest Coors Victory
The final polls are out in the Senate race, and we again have great disparity among the different surveys, as we did two years ago for the Wayne Allard-Tom Strickland showdown. In the final election weekend polling of 2002, the Post had Strickland up by 1, the Rocky Mtn News had Strickland up by 5, and Zogby had Strickland up by 9. The final result? Allard 50, Strickland 45. That means the Post undercounted the GOP candidate by 6 points, the Rocky by 10 points, and Zogby by an atrocious 14 (but Zogby's reputation for state polling is notoriously bad).
Today, looking at a comparable snapshot of the race between Ken Salazar and Pete Coors, Zogby has Salazar leading by 9, the Rocky (Public Opinion Strategies) has Salazar leading by 6, and the Post (Mason-Dixon) shows the candidates knotted at 46. Factoring in the same biases that affected Colorado pollsters in 2002, the results of the election could be forecast as follows:
Zogby: COORS +5 (Salazar +9, net GOP gain of 14)
Rocky: COORS +4 (Salazar +6, net GOP gain of 10)
Post: COORS +6 (TIE, net GOP gain of 6)
Average: COORS +5
If this model holds true, Coors will win by the same margin that Allard did two years ago. But it will only happen if the Republican GOTV effort does its job. And believe me, it's looking strong.
The key difference between the two years, of course, is who is at the top of the ticket. In 2002 it was the governor's race, and Bill Owens won convincingly. The Dems fielded a weak candidate. Colorado is going for President Bush this year, but it figures that some more Dems will come out to vote against him. So I think it's safe to add that handicap to the average posted above. My prediction? Coors wins by 2 to 3 percentage points.
One final note I can't repeat often enough: predictions don't come true if you don't vote, if you don't volunteer for the 96 Hour Victory Team to ensure that every possible supporter of our side gets out and votes.
The only one of the polls with any internals available is the Denver Post. It looks like their balance of party sample (using a model of 37% GOP, 33% Independent, 30% Democrat) is very close to the actual. The Rocky Mountain News wouldn't release their internals because of "proprietary" issues.
Cross posted at Mount Virtus.
Today, looking at a comparable snapshot of the race between Ken Salazar and Pete Coors, Zogby has Salazar leading by 9, the Rocky (Public Opinion Strategies) has Salazar leading by 6, and the Post (Mason-Dixon) shows the candidates knotted at 46. Factoring in the same biases that affected Colorado pollsters in 2002, the results of the election could be forecast as follows:
Zogby: COORS +5 (Salazar +9, net GOP gain of 14)
Rocky: COORS +4 (Salazar +6, net GOP gain of 10)
Post: COORS +6 (TIE, net GOP gain of 6)
Average: COORS +5
If this model holds true, Coors will win by the same margin that Allard did two years ago. But it will only happen if the Republican GOTV effort does its job. And believe me, it's looking strong.
The key difference between the two years, of course, is who is at the top of the ticket. In 2002 it was the governor's race, and Bill Owens won convincingly. The Dems fielded a weak candidate. Colorado is going for President Bush this year, but it figures that some more Dems will come out to vote against him. So I think it's safe to add that handicap to the average posted above. My prediction? Coors wins by 2 to 3 percentage points.
One final note I can't repeat often enough: predictions don't come true if you don't vote, if you don't volunteer for the 96 Hour Victory Team to ensure that every possible supporter of our side gets out and votes.
The only one of the polls with any internals available is the Denver Post. It looks like their balance of party sample (using a model of 37% GOP, 33% Independent, 30% Democrat) is very close to the actual. The Rocky Mountain News wouldn't release their internals because of "proprietary" issues.
Cross posted at Mount Virtus.
posted by Ben | 9:43 AM
<< Home