Cell phones and polling: an excellent analysis here:

[chiburb]

Snip:

For the last 20 or so years, telephone surveys have excluded the roughly 5% of U.S. households without any form of home phone service. Those who lack phone service are disproportionately younger, non-white, and lower income, but their numbers are small, they vote at much lower rates than other adults, and pollsters typically weight by age, race and income, so the impact on political polling has been negligible.

However, the willingness of some to disconnect their landlines in favor of mobile phones threatens to increase the size of those missed by telephone samples significantly. So let's try to answer three questions:

1) What percentage of households are wireless only?
2) How are wireless households different?
3) Given the answers to the first two questions, how likely are these differences to affect political polls this year?

Snip:

If we were able to include the wireless only adults, it would change the overall preference by only one point - Kerry would lead 48.5% to 47.5%.

http://www.mysterypollster.com/main/2004/10/arianna_huffing.html

[sendero]

1. I don't consider this "analysis"...

... to be remotely credible, because it acts as though "wireless only" households are the big problem. They are not, caller id is the big problem.

Many people just don't answer any call unless they know who is calling. How many? Hard to say, but it is certainly more than 10% or 15%.

My personal opinion is that the polling companies are in a bind that they cannot fix, so now they are going to bleat and bleat that they are not really in a bind.

[chiburb]

2. Don't disagree with you

In fact, that point is made in the comments after the analysis. Did you read them?

[sendero]

3. No, sorry...

.... didn't read every word of the "analysis" :)

It's just getting to where every article you read is a puff piece trying to sell you on a fact that is anything but.

It's getting tiring.:mad:

[yo-yo-ma]

4. on the topic of caller id etc

Phone based medical research (surveys, case-control studies) had about a 90% participation rate 20 years ago. Now many researchers find phone based studies very problematic - for the same reason that phone based political surveys are flawed. The rate of participation has fallen to (hard to say, varies) 50+%. This is due to the technologies of caller id, answering machines etc, but also due to the fact that the last several decades has seen the emergence of telephone based marketing - and many people no longer want to participate in telephone anything at all.
This is a good article that highlight the complexities of survey research - and how biased the results can be.
For a study to be valid the sample must be generalizable - that is, must represent the population - hence the attempt to "control" for difference by weighting, say, white voters over nonwhite. However, a much better technique is to start with the most representative sample possible.
There are some other good threads that address this issue.