I am not convinced thousands and thousands of brand new voters added to the roles are going to show up to vote in communities across the nation.

I am willing to bet turn outs in most places will be good and that the national popular vote will be very close-hard to say with the electoral vote.

Either way I still think Obama has the advantage, but it isn't in the bag.

Why?

Hmmm, upon further reflection, you're right, weighting should be determined by census data, not sample size, so age weighting is out.

My second guess, then, is that there is an assumption that those without land-lines, technology early-adopters, are not being proportionately heard from.

The problem is that the weighting would be inaccurate

Why?

If the poll weights for age shouldn't it largely address it? Or is there some suggestion that even among young people, cell-phone only would vote differently from young landline users?

The problem is that the weighting would be inaccurate for that age group, which is an age group that tends to favor Obama.

Seems like Caller ID would have similar impact, with wider demographic. If I see a number I don't know, I let voice mail get it.

Not in and of itself; you could have ported a land line number to the VoIP line, and in any case a random dialer could hit your number if it is connected to the general exchange (i.e., if it's not, say, a free Skype account).

In my specific case, it was originally a Vonage number that was ported to my cable operator.

I'm not sure if there's a listing in the database mentioned above for VoIP like there is for cell/land lines, but since there's presumably no additional charge for a VoIP line to receive a call my guess would be that it's closer to a landline than to a cell.

I wasn't so much thinking about that, but rather the fact that my VoIP line isn't connected to any geographic region except in regards to E911 routing. Especially with Vonage (not so much with cable), you could plug your VoIP modem into any internet connection anywhere in the world, and your calls would be routed to you. I could have signed up for the number in Maine, but be living permanently in California, so it becomes problematic for geography sensitive polls.

I have a VoIP line, I wonder if that would prevent polls from contacting me. Anybody know?

Not in and of itself; you could have ported a land line number to the VoIP line, and in any case a random dialer could hit your number if it is connected to the general exchange (i.e., if it's not, say, a free Skype account). I'm not sure if there's a listing in the database mentioned above for VoIP like there is for cell/land lines, but since there's presumably no additional charge for a VoIP line to receive a call my guess would be that it's closer to a landline than to a cell.

Sampling -- The process of drawing of random sample of cell phone numbers is relatively easy and essentially the same as the "random digit dial" (RDD) technique for sampling landline telephone numbers. Here's the oversimplified version of how it works: Companies like Survey Sampling and Marketing Systems Group obtain listings of the special telephone exchanges (the first three digits of a seven digit telephone number) assigned for cell phone usage and draw a sample from all possible numbers that could occur within those exchanges. They randomly generate numbers from the pool of all possible numbers in those exchanges.

I'm not all that young myself, but haven't had a land line in years. I do not receive telemarketing calls of any kind except for robo-calls, and even they are exceedingly rare. The last time I was called or polled in an election, I had a land line.

I've seen a number of Obama folks maintain that polls as presently administered are inaccurate because cell-phone-only users are not contacted (and many young people do not have land lines).

Pew aside, I have but two words: Number portability. How would they know what kind of phone they're calling?

Another question with these polls is how they weight the RV and LV populations. In some states voter registration is radically increasing for this election. If the pollsters don't track it, their weights can come out wrong.

I agree, I think this year the likely voter screen may actually make the polls less accurate if they're relying on past voting behavior, given how many first-time voters we had during the primaries.

Does anybody know if any of these polls took into consideration a combination of not voting in the general 2 and 4 years ago, but having voted in this year's primary? And do they classify those people as likely voters or not?

Another question with these polls is how they weight the RV and LV populations. In some states voter registration is radically increasing for this election. If the pollsters don't track it, their weights can come out wrong.