How much do the "cell-phone onlys" skew the polls?

[tommieharris91]

I found this report on rcp.com earlier. I've mentioned this could cause some inaccuracies in this election, but the writer of this report goes further to define exactly how much of the electorate is missing when doing a landline only poll.

http://www.pollster.com/blogs/pew_research_missing_cellphone.php

[Frozen Sooner]

***Disclaimer***

I am arguing no political point here.

Gallup does survey mobile-only respondents according to their published methodology.

[tommieharris91]

***Disclaimer***

I am arguing no political point here.

Gallup does survey mobile-only respondents according to their published methodology.

I think that's kinda strange because I think the Gallup polls have had the most variance this cycle. I mean, it seems like one week they have McCain +10, then next week they see Obama +6.

[Frozen Sooner]

I think that's kinda strange because I think the Gallup polls have had the most variance this cycle. I mean, it seems like one week they have McCain +10, then next week they see Obama +6.

Oh, man, Gallup's FAR from the worst this cycle, though they do swing a bit more than a couple of others. Hotline, for example, swings around like crazy. The CBS/WaPo poll that should get published tomorrow shows like an 18-point Obama swing over the last two weeks.

Rassmussen tends to be a bit more conservative one way or the other, inasmuch as their peaks tend to be lower. Their methodology encourages that, as they make sure to "balance" their results by party affiliation to reflect party affiliation in the general population. I'm not sure really how statistically valid that is. I dunno-Ron's the pollster, he can probably answer that better than I can. Maybe everyone does this.