Rasmussen Reports Polls Were Biased

[Lunar]

http://fivethirtyeight.blogs.nytimes.com/2010/11/02/live-blogging-election-night/#rasmussen-reports-polls-were-biased

SHOCKING

[Prez Duke]

I think most polls were biased towards the Republicans in most of the Senate races this year. In particular, Nevada and California.

[J. J.]

I don't know.  Didn't PPP have some of the same results?

[kevinatcausa]

He's put out a more detailed analysis.  Note that Rasmussen's numbers would be about 0.4 better in each column if it weren't for that one Hawaii poll. 

[You kip if you want to...]

He's put out a more detailed analysis.  Note that Rasmussen's numbers would be about 0.4 better in each column if it weren't for that one Hawaii poll. 

Let's be fair though, how could any "professional" polling company have gotten that Hawaii poll so wrong? It'd be like releasing a poll tomorrow showing Obama losing in D.C. to Sarah Palin.

[crypto-fascist superhero]

If we're on that subject, we could bring up PPP's infamous NY-23 poll, or that SurveyUSA this year which showed Gillibrand and DioGuardi tied.  Every prolific pollster has one poll in a cycle that way misses the mark.

[feeblepizza]

If we're on that subject, we could bring up PPP's infamous NY-23 poll, or that SurveyUSA this year which showed Gillibrand and DioGuardi tied.  Every prolific pollster has one poll in a cycle that way misses the mark.

Exactly. No need to get angry because Rasmussen made a simple error that all pollsters make once in a while.

[Senator Sbane]

I think it's consistent error and bias this year which is the issue, not the Hawaii poll.

[Lunar]

Weasel Silver is a delusional liar and a hack. He's been caught in so many lies that nothing he says is credible. He is purposefully manipulating data in order to try to discredit Rasmussen. The fact is, Rasmussen correctly picked the results in several States while others like PPP got the winner wrong.

Rasmussen also incorrectly picked the results in several states, factually.  Rasmussen also factually did not show poll results closer to the actual results than most other mainstream pollsters, on average.  

What evidence are you looking at?

Does anyone think that single-day sample robocalls with zero callbacks, inter-household randomization or cell phone numbers is going to fly as acceptable polling techniques among junkies in 10 years?

[Joe Republic]

Says it all, really.

[-]

Rasmussen. BIASED?

Not suprised. Just confrimed what I knew all along. Cheap, low cost polls have very little accuracy.

[Meeker]

Washington Senate - Rossi vs. Murray   FOX News/POR-Rasmussen   Murray 49, Rossi 47   Murray +2
Washington Senate - Rossi vs. Murray   PPP (D)   Murray 48, Rossi 50   Rossi +2

Alaska Senate - Miller vs. McAdams vs. Murkowski   PPP (D)   Miller 37, Murkowski 30, McAdams 30   Miller +7

California Senate - Boxer vs. Fiorina   PPP (D)   Boxer 50, Fiorina 46   Boxer +4
California Senate - Boxer vs. Fiorina   SurveyUSA   Boxer 46, Fiorina 38   Boxer +8

Yes, some of their polls were more accurate than other pollsters. But on average, they were more inaccurate than other pollsters. Taking all polls into account - not just ones you cherry pick - reveals this. What is so difficult to understand about that?

[I left.]

Rasmussen. BIASED?

Not suprised. Just confrimed what I knew all along. Cheap, low cost polls have very little accuracy.

lol

[Fmr. Emperor PiT]

Rasmussen. BIASED?

Not suprised. Just confrimed what I knew all along. Cheap, low cost polls have very little accuracy.

lol

     Rasmussen messed it up pretty badly, though the general consistency in the error of their polls suggests that they would have performed very well if they had just weighted the polls differently.

[Vasall des Midas]

Should point out that Rasmussen managed to be this bad on average  despite releasing the most polls of anybody. It's really a performance not explainable except by systemic bias (intended or not).