Pollsters' Cell Phone Problem Getting Worse
The percentage of adults without a landline telephone rose again during the first half of 2012, according to a new report released last week, continuing to present pollsters with unprecedented challenges.
The report, compiled by the Center for Disease Control, shows that fully 34 percent of adult Americans lived in households that only had wireless phones between January and June of 2012. That is up from 32.3 percent in the second half of 2011. While the 1.8-percentage-point increase seems high, it is the smallest six-month increase reported since 2008.
For decades, pollsters have relied on randomized telephone calls to landline phones to complete their surveys. These calls may have been randomly selected by a computer, or voter files may have been randomly selected, with the file then being matched to a phone listing to make the call. But this new report underscores that pollsters who only call landlines are missing more than a third of the potential electorate. And that third of the electorate differs from the roughly 62 percent who have landline phones.
The greatest danger for pollsters lies in the demographics of those who have abandoned their landline phones. There are four demographic groups in which the majority of adults are cell-phone-only: adults aged 25–34 (60.1 percent), adults living only with unrelated adult roommates (75.9 percent), adults renting their home (58.2 percent), and adults living in poverty (51.8 percent). Other groups are approaching that majority are Hispanics (46.5 percent) and adults aged 18-24 (49.5 percent).
Pollsters who call only landline phones -- or for whom landlines make up the vast majority of their interviews -- would likely be undersampling members of these demographic groups. These demographic groups were more likely to vote for President Obama in 2012 than the overall population, according to exit polls (these exit polls did not ask about living situation). That may explain why Obama outperformed his margin in the final Real Clear Politics average by 3 percentage points, and polls that don't call cell phones were among the most biased towards Republican candidates.
Some pollsters are confronting the cell-phone problem head-on. The NBC News/Wall Street Journal poll, conducted by Democrat Peter Hart and Republican Bill McInturff -- increased their percentage of cell-phone respondents from 25 percent of their overall sample to 30 percent. Republican pollster Glen Bolger, a colleague of McInturff's at the Alexandria, Va., firm Public Opinion Strategies, wrote after Election Day that his GOP colleagues must call more cell phones in order to produce accurate surveys. But measures like that are not always successful: The Gallup Organization, which increased their cell-phone interviews to a whopping 50 percent in the final weeks of the 2012 campaign, were among the least accurate pollsters.
And while many public pollsters have embraced calling more cell phones, there are exceptions. Automated-phone pollsters, which are prohibited by federal law from calling cell phones, continue to sprout up. The Raleigh, N.C.-based Democratic firm Public Policy Polling and their sponsors have taken a victory lap over their apparent accuracy in 2012, and last week, a GOP counterpart, Harper Polling, began conducting automated surveys, vying for attention from media and consumers hungry for poll results on a multitude of issues, often regardless of quality.
Correction: The original version of this post misstated the report's findings. The number of landline-less households increased in the first half of 2012.