Lessons: Mobile Phones

Legacy blog posts Mobile phones Sampling Issues

In today’s Washington Post, Richard Morin provides an important epilogue on the issue of mobile-phone-only voters. As I wrote before the election, some said that polls were missing a hidden Kerry vote among young people who had switched off their home phone service in favor of cell phones. While we had data on the percentage of adults with mobile phones only, we could only guess as to what share they would be of voters and what their preferences would be in the Presidential race. Now, Morin shares some important exit poll data that went unreported on election night:

Buried in the [exit poll] data is the answer to a critical question raised during the campaign about traditional telephone surveys…The exit pollsters cast new light on the issue by asking people leaving the voting booths about their phone service and use.

The good news for pollsters is that only 7 percent of all voters in 2004 were using cell phones as their sole service. The bad news is that this figure swelled to nearly 20 percent among voters between the ages of 18 and 29 years old.

Uh-oh. Not only are there lots of young people without household phone service, but these cell-phone-only voters voted 56 percent to 41 percent for Kerry, meaning missing them in telephone polls could produce polls that underestimated the Kerry vote [emphasis added].

Morin concluded with a point that some may find confusing:

But, happily, one other fact may have saved pollsters, at least during this campaign. Young people with cells were not much more likely to back Kerry than those in homes with traditional phone service only or those who had both cell and traditional service. So missing them wouldn’t dramatically skew the results.

Why is that? Most national polls adjust or “weight” their data by age so that the percentage of each age group among all adults on the survey matches what the U.S. Census says it should be nationwide. Surveys will typically under-represent younger people, partly because they are more likely to lack home phone service.  So the weighting procedure essentially replaces one type of 18-to-29 year old (those without home phone service) with another (those who do have home phone service). Morin’s point is that both types of younger voters preferred Kerry by roughly the same margins, so weighting the data by age helped reduce the “skew” in the results.

But not so fast. What about all the other questions on the exit poll? Were cell-phone-only 18-to-29 year olds similar in other respects as well?

More important, what about the cell-phone-only voters over the age of 29? If 20% of 18-to-29 year olds had only cell phones, and 17% of all voters on the exit polls were age 18-to-29, then roughly half of the 7% without cell phones were under 30 (20% * 17% = 3.4%). What about the rest? Were cell-phone-only voters over 30 just like all other voters over 30?

And what happens when the 7% grows to 20%? Morin is right that cell-phone-only voters failed to “dramatically skew” surveys this year but, as I wrote a month ago, “things could be very different next time.”

Mark Blumenthal

Mark Blumenthal is political pollster with deep and varied experience across survey research, campaigns, and media. The original "Mystery Pollster" and co-creator of Pollster.com, he explains complex concepts to a multitude of audiences and how data informs politics and decision-making. A researcher and consultant who crafts effective questions and identifies innovative solutions to deliver results. An award winning political journalist who brings insights and crafts compelling narratives from chaotic data.