Told you Rasmussen was an outlier

[antiimperialist]

There was an article last week on Rasmussenreports.com, based on their own poll, where it was said that Hillary Clinton (mysteriously to me) fell behind Rudy Giuliani in a hypothetical election matchup, going from trailing him by 1% to trailing by 7% the next month.

Well, turns out that this one was the only one out of the last 5 similar polls (Fox, NBC/WSJ, Quinnipiac and now Gallup) to find such an outcome. Rasmussen has proven very accurate come election day, but its results have never been so much at odds with the rest of pollsters.
Gallup says today, "All three Democrats garner 50% of support among registered voters when pitted against Giuliani, with Giuliani getting 46% against Clinton and 45% against both Edwards and Obama."

[MADem]

1. Well, they've been "accurate" on election day after all the cheating, vote caging, and

funny ballots are sorta counted....what's THAT say about them? :shrug:

[paulk]

2. Rasmussen has only been accurate on election day

only if you accept the last two elections as legitimate.

[antiimperialist]

3. Gallup made an error. They say Clinton has 50%, but the table shows Giuliani with 50%

Something's wrong.
Gallup says all three Democrats "garner 50% of support among registered voters when pitted against Giuliani, with Giuliani getting 46% against Clinton and 45% against both Edwards and Obama." (11th paragraph top to bottom.)
http://www.galluppoll.com/content/?ci=17785

But the table in the same page reads: Giuliani 50%, Clinton 46%.

[BootinUp]

4. I generally find it hard to

dismiss poll results without some reasonable analysis as to why it failed. In general, I think we expect more accuracy from polls than is realistic, considering how many people refuse to participate or are not available to participate. So the margin of error can only be accurate in the population that participates. In the population that does not it is even a wider spread.

[Windy]

5. and with Caller ID, most don't answer their phone unless they

recognize the number. That MUST effect the sampling!