By: Matthew Dickinson

Matthew Dickinson — Wed, 20 Oct 2010 23:35:00 +0000

Orion – Great question. The short answer is that yes, there’s a real chance the cell phone only users can skew the polling data. Pew put out a study recently showing that there is a definite political bias among cell phone users – they tend to be younger, better educated and more likely, as you indicate, to vote Democrat. Note that automated pollsters, like Rasmussen, don’t sample cell phone users at all. Gallup, in contrast, does include cell phone users in their surveys. Whether they are getting the correct weighting is hard to know. Interestingly, however, it is possible to get a reliable survey sample even without including cell phone users. How, you ask? Since this question comes up a lot, I should probably devote a separate post to it.

By: Orion Lewis

Orion Lewis — Wed, 20 Oct 2010 20:46:10 +0000

I agree that polling has proved to be quite accurate in past election cycles (your pre-election analysis of Scott Brown’s improbable victory comes to mind). However, Jess and I have long been interested in the question of whether pollsters miss young people who do not own landlines? The Economist just ran a piece to this effect, claiming that pollsters were nervous about their predictive accuracy due to the fact that “cell phone onlys” (CPOs) now make up 25% of the electorate. As we know, this younger demographic leans heavily Democratic. Do you think there is any chance that this skews the data? Would this sampling bias also carry over to impact the inclusion of minorities?

Comments on: Can We Trust the Gallup Generic Ballot Results?

By: Matthew Dickinson

By: Orion Lewis