Two kinds of polling in California
By Neil Stevens on October 25, 2010
As always I give the note that any analysis I do of the California Senate race carries an unusual risk of bias because I live here and I have a strong emotional attachment to the outcome.
That said, I’m beginning to notice a pattern in the polling between Democrat Barbara Boxer and Republican Carly Fiorina that suggests serious, late-breaking movement in favor of the Republican. I see it in the way the polls are moving with their methodologies.
To demonstrate what I mean, I will take the most recent polls from Real Clear Politics and sort them by the length of time they were in the field:
Poll | Duration | Lead |
---|---|---|
Fox/POR | 1 day | D+4 |
Rasmussen | 1 day | D+2 |
Wilson | 2 days | R+3 |
Reuters/Ipsos | 3 days | D+1 |
SurveyUSA | 4 days | D+2 |
LAT/USC | 8 days | D+8 |
PPIC | 8 days | D+8 |
As we can see, the two pollsters that take their sweet times, running around over a week asking their questions, are the ones showing the large Boxer leads. Everyone else shows it a tight, competitive race.
And that, to me, suggests that polls starting earlier are missing some of the impact of the barrage of new ads from Carly Fiorina, which appear to be making the race so close.