Pollsters’ cellphonobia

Two-thirds of Americans are talking on them, while every careful pollster is talking about them. The question is simple: Will the rise in cell phone-only households render traditional telephone polls inaccurate? The answer is far from simple, but the bottom line seems to be no, for now.

Mucking up poll results requires cell-only voters to be both substantial in number, and to have different views from those still clinging to landlines.

The number of Americans cutting their phone cord, going wireless only, is exploding. In December 2004, the National Health Interview Survey, using in-person interviews, found 5.4 percent of adults living in households with only a cell phone. By the end of 2006, two years later, that proportion more than doubled to 11.8 percent.

If growth continues at this rate, over a quarter of Americans will only have cellular service, and no landline, by Election Day 2008. Of course, change does not always continue “at the same rate.” ABC pollster Gary Langer cites his baby. Observing her growth over her first three months, he reports, “if present trends continue,” she will be 18 feet tall and weigh 3 tons by the time she reaches age 17!

While genetics will slow the growth of baby Langer, demographics may slow the cell-only trend. To date, cutting off one’s landline seems to be a lifecycle phenomenon. Those who are young, single, childless, living with unrelated roommates and mortgageless are much more likely to be cell-only, while older, married parents who own homes overwhelmingly retain landlines. Even among the young, those who are married, or have kids or own homes, are more likely to be available on a landline. Thus, the size of its niche may limit the growth of the cell-only trend.

Nonetheless, even at 12, 15 or 20 percent, the cell phone-only constituency is substantial. However, its size matters to poll accuracy only if this segment behaves differently from others. The bad news is these folks are different.
Just 6 percent of cell-only households have dial-up Internet connections, compared to 33 percent of those with landlines. Wireless-only households are twice as likely as landline homes to contain binge drinkers, and half as likely to house diabetics and flu shot recipients.

These differences extend to politics as well. In 2004, those with cell phones only voted for John Kerry by a nine-point margin, while the electorate as a whole supported Bush by three points. Similarly 17 percent of cell-only voters said terrorism was the most important issue determining their vote, compared to 21 percent overall.

Those differences did not make much impact on phone polls in 2004, when just 5 to 7 percent of voters were cell-only, but as that number doubles or triples the problem seems to become insurmountable for traditional phone polls.

And it would be, except for one saving grace — so far, while cell-only voters are different from the electorate as a whole, they are quite similar to other voters their age. Overall, voters under 30 gave Kerry a 14-point margin, close to his 17-point margin among cell-only, under-30s. Terrorism was most important to 16 percent of those under 30 and to 17 percent of those who were wireless-only in the same age cohort.

As a result, ensuring a poll has the correct percentage of younger people now yields accurate results, overall.

A better solution, of course, would be to sample cell phone users. However, such interviews are extremely expensive (five to 15 times the usual cost) and firms wedded to predictive dialers cannot legally call cell phones. (Neither can automated polls like SurveyUSA and Rasmussen.)

For now, wireless households should be giving pollsters pause, but in the near future, they could cause real heartburn.

Mellman is president of The Mellman Group and has worked for Democratic candidates and causes since 1982. Current clients include the majority leaders of both the House and Senate.