Welcome to European Tribune. It's gone a bit quiet around here these days, but it's still going.
Display:
"For the first time, we're actually seeing a difference between cell-only voters and land line voters when you take into account age."

Did they bother to look for a difference before?

AFAIK, pollsters, up to now, have been proudly proclaiming there wasn't a difference.  They knew there wasn't a difference because they were Super-Pollsters! -- able to leap over demographic/statistical variance in a single bound.

Or something like that.

She believed in nothing; only her skepticism kept her from being an atheist. -- Jean-Paul Sartre

by ATinNM on Wed Oct 15th, 2008 at 12:13:49 PM EST
[ Parent ]
The problem is that the difference is not statistically significant in each individual poll, but taking them all together, as Nate did, it is.  So pollsters generally don't do it.  Think of it as adding an extra 1 or 1.5 to Obama current total (since a few of the polls do use cells).  Perhaps it's a 9- or 9.5-point lead rather than 8.

The good news is that this would be very relevant to some places Obama and some key Senate candidates are competing in.

Four years ago, the number of cell-only voters just wasn't big enough to justify inclusion.  I still think it's iffy now but probably worth doing.

Be nice to America. Or we'll bring democracy to your country.

by Drew J Jones (pedobear@pennstatefootball.com) on Wed Oct 15th, 2008 at 12:36:38 PM EST
[ Parent ]
As you say - a 1- 1.5% bias is not significant within a poll with 3-4% error margins - but is significant in the context of the aggregation of all of those polls - especially if you add in further possible sources of bias due to differential ground games, interviewer bias, changes in "likely" voting patterns, and the growth of differential early voting which turns "likely" into "definite"

Vote McCain for war without gain
by Frank Schnittger (mail Frankschnittger at hot male dotty communists) on Wed Oct 15th, 2008 at 12:54:05 PM EST
[ Parent ]
They used age or party ID to adjust land line based polls to account for any under-representation of cell-phone only population.  However what the Pew and other research seems to be saying is that the cell-=phone only population is qualitatively different   even after you adjust for age and part ID variables.  e.g. it is more liberal, more ethnically diverse, or more anti-incumbent.

Vote McCain for war without gain
by Frank Schnittger (mail Frankschnittger at hot male dotty communists) on Wed Oct 15th, 2008 at 12:44:45 PM EST
[ Parent ]
More liberal, more ethnically diverse, and far more urban, I'd guess.

Be nice to America. Or we'll bring democracy to your country.
by Drew J Jones (pedobear@pennstatefootball.com) on Wed Oct 15th, 2008 at 12:52:55 PM EST
[ Parent ]

Display:

Occasional Series