Equally important were Mitofsky’s contributions to RDD, for this technological shift in the way surveys are done provides useful purchase on the ways the field is changing today. Prior to the advent of RDD in the 1970s, pollsters relied largely on professional interviewers who interviewed respondents in the respondents’ homes. If they did telephone interviewing, they used mostly local telephone directories for selecting respondents. Each method presented problems.

Using trained interviewers to conduct in-person interviews was expensive, time-consuming, and always a potential source of error. Although some thought that in-person interviews provided higher-quality data than telephone interviews, others cast doubts on that notion. Nevertheless, early methods of telephone interviews were problematic not only because they missed potential respondents with unlisted telephone numbers, but also because directories rather quickly got out of date.

Random digit dialing initially emerged in the 1960s as a means to overcome what James Frey calls “undercoverage in telephone directories.” By the 1970s, RDD had been adopted by several academic survey organizations, and was rapidly becoming pollsters’ “method of choice” for contacting respondents. Once Mitofsky and Waksberg improved the technique by developing a two-stage, cluster design to further reduce “unproductive dialings,” RDD became even more popular among pollsters and survey researchers. Combined with a CATI (Computer Assisted Telephone Interviewing) system, RDD made polling relatively inexpensive and very fast, and, assuming interviewers were well-trained and supervised, probably less susceptible to biases introduced by interviewer effects.

Public opinion research continues to evolve. New technologies—such as “caller ID” and, especially, cell phones—will very likely entail significant changes in telephone-based interviewing. Survey researchers are increasingly vexed by declining response rates; it is estimated that polls miss between 25 and 30 percent of the public, and, if these people are systematically different from those who agree to be interviewed, the resulting poll, no matter how otherwise technically sophisticated, is biased.

Even when not coping with technological and social changes, surveys may confront serious problems. Vastly uneven distributions of political information across the public, for example, may severely crimp their capacity to shed light on, as Scott Althaus puts it, “what the people really want.” If Robert Weissberg is right, the way survey organizations pose some queries introduces systematic, and serious, bias into their results.

Students of elections, public opinion, and survey methodology will have to cope with these—and a myriad of other problems—without Warren Mitofsky. It may be that survey researchers will “muddle through,” just as they have done after other pioneers of the field, such as George Gallup Sr., Archibald Crossley, and Elmo Roper, to name just three, have passed from the scene.

As we who rely on the polls face the future, something that Mitofsky wrote in 1999 ought to be our touchstone:

I do not want to abandon tested and proven survey methods that are based on a solid foundation of probability theory until we have a new theory-based methodology for doing so… A growing number of survey researchers are unfortunately being led to the rocks like Ulysses’ sailors following the Siren call of cheap, but worthless, data.

Mitofsky knew the field was facing changes, but he opposed change for change’s sake.