posted on July 31, 2008 14:01
A very smart (by definition) fan of this blog asked a question I could not answer: “How can telephone polls be accurate when so many people – especially young people – have cell phones and no landlines?”
So I posed the question to three smart pollsters (Democratic and Republican). Here are their answers (with only slight editing):
Pollster Number One:
“The issue is whether young people with cell phones have different attitudes than those with landlines. So far, appears not to be a big difference -- but we are constantly checking by adding online components to surveys when it's doable.”
Another expanded on that point – and raised the Obama impact:
“Pew and other organizations have done pretty extensive research and found that the political attitudes of non-land line voters are not discernibly different from the political attitudes of landline voters. As long as pollsters weight to get the correct age distribution, they're OK.
“That said, I think the situation gets more and more volatile by the year and if it turned out that all the polls were underestimating Obama by a point or two this year I would not be surprised. We are really going to have to figure out a viable way to move polling online in the next few years.”
And from a third:
“(It’s) another opportunity for error in an error-prone business. You have to actually take more care to make sure it's right. One in six Americans have no landline at all. Only cell. Tend to be younger. Not necessarily registered. And in some parts of the country recent immigrants. You have to know the population and make sure your sample reflects the population. For voter polls we generally ask if the interview is being conducted on a cell phone or landline. It usually comes out 5-10% cells. We call off registered voter lists. Many voters are now listing their cell as part of their voter record and sample companies are matching cell numbers to voter records. If you can get cell numbers, the potential for text mailing election messages is huge.”
Click Here to discuss and comment on this and other articles.