Chet Bowie just sent me a publication from the Committee on National Statistics titled, "How Can We Conduct Telephone Surveys in a Cell Phone Age?" This is a report from a two day meeting of experts in Washington in October.
At this point it’s a somewhat reassuring story, although there no doubt is trouble ahead. Despite what you may hear, the current best estimate of the percent of the adult population in cell phone only households (and not reachable by our standard telephone survey practices) is about seven percent. And it’s a very biased seven percent heavily over represented by the young and well educated in small, childless households. In other words, just what we suspect and the same people we have had a tough time getting even before cell phone only households became an issue. People who have attempted to survey these folks by cell phone have experienced higher refusals and lower response rates than landline studies.
So the conclusion here is that this seven percent is not yet a problem and the fact that we miss them in our standard telephone study is not a major bias issue unless you are especially interested in this subpopulation. Exhibit A at this conference was a paper from the Pew Center pointing out the accuracy of 2004 election polling in which cell phone only households were ignored. If you are interested in this subpopulation and we need to survey cell phone users, be prepared for a tough and expensive study.
Everyone expects that seven percent to grow over time and in that connection it’s worth keeping an eye on what the Europeans are doing to cope with it. A number of countries there, most notably Finland, have higher cell phone penetration than here in the US and have been thinking about this problem for years.