Cutting the cord: households dropping land lines for cell phones.(Verizon Communications Inc.)

BBER survey researchers and others in the industry and in academia are exploring the possibility of supplementing traditional land-line telephone survey samples with cell phone numbers to get a more accurate look at the population. Survey researchers also are examining statistical methods to reduce the bias caused by leaving cell-phone-only households out of our studies. In fact, the Bureau just completed its first telephone survey that included cell phone numbers.

Results from the Bureau survey indicate that the cell-phone data quality was quite good, but challenges remain. More respondents refused to participate in a cell phone interview than did those called on a land line. In addition, the blocks of cell phone numbers BBER sampled yielded more non-working numbers than did comparable blocks of landline numbers. Both of these factors increase the cost of obtaining an adequate number of survey interviews. In spite of these challenges, the Bureau views adding cell phone samples to telephone surveys as highly important.

However, adding cell phone numbers to samples is far from risk-free. Public reaction to being called on cell phones for legitimate survey research may be negative. The public may come to view cell phone survey research in the same light as land-line telemarketing. If this happens, will the public seek a legal ban on cell phone survey research or clamor for a "do not call" list? Either of these two possibilities could cause telephone survey research as a method to obtain general population estimates of vital public opinions and behaviors to lose viability because of the bias in the results of land-line-only surveys.

Mail surveys are not an acceptable alternative because up to 40 percent of households have unlisted addresses. Old fashioned face-to-face interviews will remain an effective way to conduct general population surveys but they, too, have problems. The cost of doing face-to-face survey research often exceeds $300 per interview, whereas telephone survey research costs about $30 per interview. The cost of obtaining the vital information gathered for governments and businesses by survey research may increase tremendously if telephone survey research loses viability. Smaller organizations and businesses may either be required to accept biased information for a lower cost, or general population survey research will be available only to organizations that can afford it--namely the largest corporations, foundations, and the federal government.

The upcoming national election season points to yet another concern caused by the rise of cell phones. Election polls are vulnerable to bias because a large majority of election polling is conducted using traditional telephone surveys, leaving cell-phone-only households out of the picture. Much of the survey methods research done by reputable organizations like the Pew Center for the People and the Press has not yet found a significant bias in election poll results caused by cell-phone-only households. This is largely due to the fact that voter turnout among young adults, who make up the largest portion of cell-phone-only households, is much lower than among older adults. If the voting pattern changes because young people are more enthusiastic about current candidates, then many election polls may obtain biased results. For example, is it possible that Barack Obama may attract more young voters than previous presidential candidates? Given the growth in cell phone use described elsewhere in this article (about 2 additional percentage points every six months), it seems very likely that an increased share of voters will use only cell phones by November 2008. Major survey research firms are now trying to decide whether to include cell phones in election polls.

Jenny Donohue worked as the Bureau's publications assistant. She is currently the field organizer for the Max Baucus campaign. John Baldridge is the Bureau's survey development director.