Do they vote at the same rate as the more landlocked population? I'm betting that they tend to be younger and more mobile, not the most likely population to vote. Until that demographic starts voting at the same rates as the general population, they won't be of much interest to pollsters. Besides, I remember a lot of people arguing that Kerry was going to win by several points because cell phone users were underrepresented in the polls. It didn't really work out then, and there is no reason to think it will now.

The term "outlier" applies to quantitative variables (ie 1.5*IQR)...not categorical variables (like being for/against a candidate).

So it's not obvious.

I had been wondering about the same point. The story could benefit from some more research, however.

First, pretty much the same factors should have been working in the primaries, and if significant should produce gaps between pre-election and exit polling. However, my memory is that the gap in several states were in the wrong direction--Obama was stronger pre-election (for which the usual explanation was reluctance to sound prejudiced).

Second, why not ask the people who do the National Election Study? They certainly pay close attention to sampling, and they may have more incentives to get it right than journalistic or political pollsters.

wasn't this idea floating around in 04?

You may think they are generally Obamatrons too but so many more of them are older people, often people just looking to cut costs. And, unless you ported a number across from POTS then you generally are not in the pool of robo numbers either.

I don't talk on the phone all that much, so I have a prepaid cell phone. (I'm 38, so there is at least one middle aged person who uses them.) One thing I love about it is that I don't get annoying political calls all the time. A few months ago, my cable company offered me a cable phone for a cheap rate, but that meant I started getting all those charity calls and political calls. I'd forgotten how annoying those calls were. During the PA primary, I got as many as 8-10 calls EVERY DAY, asking me who I was voting for, whether I supported issue X or Y, reminding me to vote, and so forth. I finally unplugged my phone, just so I could stop the phone from ringing all the time.

If a pollster calls me on my cell phone, where I pay for every minute of air time, I'll be even more annoyed. It may be my patriotic duty to vote, but it's not my patriotic duty to be polled. Please, don't call. If you run up my cell phone bill, I'll be less inclined to vote for your candidate.

The country doesn't really benefit from accurate polls on the McCain/Obama race, especially those conducted months before the election. If anything, the constant polling of this race fuels horse race coverage and distracts from real issues. So, I don't really care if these polls are off by a few percentage points.

However, Brown and Maslin's point can be applied to surveys of public opinion on policy issues -- surveys that arguably do have an important place in our democracy because they tell our leaders what the public wants. If these surveys do not include cell phone users (and many don't), researchers will underestimate the number of Americans holding liberal policy positions.

So, I don't mind if horse race polls are biased: We'll know the real numbers on election day. But researchers who are measuring public opinion on actual issues (the economy, health care, the war) have a responsibility to strive for a representative sample -- one that includes cell phone users.

Pollsters already operate with the shaky assumption that survey respondents are representative of the people who refuse to answer surveys, when this is almost certainly not the case. Survey researchers cannot make anyone answer the phone -- but they should at least take the time to make sure that all types of Americans get the phone call.

It's disappointing that Salon has chosen to feature this article, which is no more than a rerun of Democrats' attempt to view Kerry's ratings through rose-colored glasses in 2004. Unfortunately, the article adds nothing new. It even fails to address the most obvious question: why the cell phone discrepancy didn't seem to produce a pro-Clinton bias in polling for the primary race.

Not getting robo-calls is one of the definite advantages to having a cellphone only. For this reason alone, I'm surprised at the number of people who still have land lines.

It's true that we have heard this idea in earlier election. However, most young people are going cellphone only and don't see the point of getting a land line as they age. That gives you a steadily enlarging and aging group that isn't counted in these polls. (I, for example, am 30 and have only paid for a land line once, briefly, for dial-up internet.)

I haven't read the piece yet, but I just had to respond to the headline: they said the same thing about Howard Dean in Iowa.

Millions of stealthy cellphone kids were to come out of the woodwork for Dean.

Just saying.

In the primaries; they were routinely over-counting Sen. Obama's vote. How do you reconcile with that?

Barack Obama's campaign generated record numbers of new voters African American and young voters) for the primaries. Its very possible that there is a dwindling supply of new voters with those demographics that the Obama campaign can register. The automatic assumption that all those cellphone only voters are African American, young people and loyal Democrats is false. Many of these people are working class, lower income whites and Hispanics who have given up landlines and who have not signed on to Camp Obama.

You guys all said that last time during Kerry v. Bush...and look how THAT turned out.

For those who cite the 2004 election:

The percentage of people using cellphones has increased significantly since the last election, especially those (like me) who are cellular-only. While I'm reluctant to wholly agree with the writers' assumption, I don't think that comparing 2004 and 2008 is entirely valid in this case.