ALL THINGS JURY: Do telephone surveys still work?

By R. Robert Samples | Jan 31, 2008

One of the concerns we hear frequently is whether traditional public opinion polling still works in today's cell phone and Internet world.

The speculation is that landline telephones are becoming obsolete given widespread use of the Internet, cell phones, and the "hangover" of the telemarketing age. On the surface, these concerns seem to be warranted. According to the CDC, more than seven in 10 American households had a cell phone in 2006. And the U.S. Census Bureau reports that the percentage of households with a computer has increased from 8 percent in 1984 to 62 percent in 2003.

So, does it make sense to continue conducting quantitative public opinion surveys using the traditional telephone interview methodology?

Typically, large scale quantitative surveys have been conducted through the use of telephone surveys since the advent of mass utilization of the telephone. While it is certainly true that the American public has become more and more reliant on the use of cell phones and home computers (with Internet access), a closer look at the numbers reveals some interesting insights.

Consider the use of Internet surveys. Over six in ten may have a home computer (and access to the Internet), but only a very, very small percentage of these computer users are signed up for a research panel. When conducting quantitative projectionable research surveys, everyone in the population must have an equal chance of being selected for participation in the survey. Clearly, that is not the case with Internet surveys (unless you are conducting a survey of, say, readers of an online publication).

With respect to cell phone usage, seven in ten households now have a cell phone, but only 12 percent are "cell phone only" households. The use of cell phones has not made landline telephones obsolete. In fact, the 2006 CDC Survey shows that nearly nine in ten households still maintain a landline. And the small segment of "cell phone only" respondents can be included in the sample if researchers manually dial these numbers (the FCC prohibits market researchers to use automated dialing to reach cell phone numbers).

Our profession also employs other safeguards to ensure that our telephone sample is representative of the population that is being studied. Multiple callbacks are made to ensure that hard-to-reach respondents have an equal chance of being included in the surveyed sample. During data collection, target demographics for the population base are continuously compared to the demographics of the surveyed sample to make sure that the collected sample is representative.

Yes, cooperation rates (the percentage of people who agree to be interviewed) have fallen due to abuses by the telemarketing industry over recent years. However, enough people still complete legitimate telephone surveys to keep this methodology very viable. Cooperation rates have generally fallen from as much as 75 percent or so to less than 50 percent. Cooperation rates are very dependent on the subject matter of the research. Jury research is very interesting to many respondents and generally enjoys a high level of cooperation. The bottom line in terms of cooperation rates is an even greater need for jury consulting firms to enforce and follow proper data collection protocols.

There's no doubt that it is becoming more challenging to conduct large-scale quantitative telephone surveys. But keep in mind that the vast majority of households still have landline telephones and if thorough data collection procedures are followed, there is every reason to expect that survey results will accurately represent the population you are studying.

Samples is president of RMS Strategies, a communications and opinion research agency headquartered in Charleston. RMS Strategies has extensive crises communications, counseling and litigation research experience and has worked for clients throughout the nation during the last 25 years. They can be contacted at 304.343.7655 or

More News

The Record Network