The collapse of landline telephones in the U.S. plagues political pollsters and their consumers. Failing to call cell phones, as Interactive Voice Response (IVR) and some live-caller campaign pollsters do, risks missing between a third and half the U.S. population. What's more, those Americans who eschew land lines are demographically different than those who stick to the plug-in version, which impacts a pollster's sample.
Nearly 60 percent of Americans aged 25-29 don't own landline phones. More than half of those living in poverty also don't have landline phones. Failing to call cell phones risks underrepresenting men, Hispanics and African-Americans. Residents of the Northeast are likely to be overrepresented in landline-only surveys.
The number of pollsters who call cell phones is on the rise, despite the attending costs. Many live-interviewer telephones polls are conducted with computer assistance: an automatic dialer generates and places a call to a random phone number. But federal law prohibits automatic dialers from calling cell phones, making the process of reaching these households significantly more expensive. What's more, cell-phone respondents are less likely to respond to survey calls, again adding to the cost of these interviews.
Some campaign pollsters -- political consultants working for candidates, committees or outside groups -- do call cell phones, but many still don't. In conversations with a handful of campaign pollsters who do incorporate cell phones in their sampling, some said that their campaign clients were the ones deciding to forego the extra cost.
Some pollsters who use automated IVR systems supplement their landline samples with live-interviewer calls to cell phones. Others add in more controversial, less reliable online sampling of cell-phone-only households. But many do not, and they risk missing huge chunks of the population. For example, nearly half the sample of a recent IVR poll of the presidential race in Michigan was composed of seniors, despite the fact that those over 65 made up only 16 percent of the electorate in 2008, according to exit polls. Seniors are the least likely of all the age groups to live in cell-only households.
Pollsters who don't call cell phones say they can weight the respondents they do reach to achieve a representative sample. But that puts additional strain on existing weighting procedures and increases the risk of error.
Thursday's report is an important reminder that, as the ways in which Americans communicate with each other change, so must methods of survey research. Landline-only phone polling is rapidly declining as a sufficient method of political research as the percentage of landline-only households decreases.