Random-Digit Dialing (RDD)

Random-digit dialing (RDD) refers to a set of techniques for drawing a sample of households from the frame or set of telephone numbers. The telephone number is the sampling unit that is the link to the household and its members. While the specific sampling techniques employed to draw RDD samples have changed over time, the three most influential RDD techniques are briefly described in this entry. RDD is distinguished from other telephone sampling methods because RDD selects the sample from the frame of telephone numbers, whereas the other methods select from lists of numbers in directories or commercial lists. The ability to sample all telephone households, not just those households on a list, is one reason for the popularity of RDD sampling. This entry ends by discussing two issues that often arise in RDD sampling: selecting RDD samples for local areas and selecting persons within the household to interview.

The basic RDD approach is simple. Using information on the structure of the telephone numbering scheme used in North America, the set of all numbers that could be assigned to households is identified. These numbers are randomly sampled, often with equal probability. The techniques for randomly selecting the numbers are defined by the specific RDD sampling scheme. The sampled numbers are dialed, and those that are residential are the sampled households. In many RDD surveys only members in a certain age range or of a specific sex are eligible to be interviewed. The data collected from the sample are used to make estimates or inferences. The estimates usually refer to all households, even though households without telephones are not covered.

RDD surveys became popular by the late 1970s, and the technique soon became the predominant method of sampling households in the United States. The popularity of RDD coincided with evidence that a large proportion of households lived in telephone [Page 676]households and could be sampled or covered in RDD surveys. By 1980, over 90% of adults lived in households with landline telephones and the percentage continued to grow slightly, to around 97% by 2000.

An important advantage of RDD sampling that helped to spur its acceptance is the relatively low cost of sampling and conducting surveys by telephone as compared to face-to-face surveys. An RDD sample is likely to cost less than 20% of the costs of an area probability sample that has the same precision. These great cost advantages fueled the early acceptance of RDD in commercial surveys. It also is the feature that makes it possible to conduct surveys of rare population groups with probability samples. For example, the Centers for Disease Control and Prevention conduct a survey to monitor childhood vaccinations in each state and large metropolitan area, even though they sample only households with children between 19 and 35 months old, that is, approximately 1 in 50 U.S. households. The cost of this type of survey by face-to-face methods would be astronomical.

Despite these advantages, the utility of RDD surveys has been questioned in recent years. One concern is the decrease in response rates in telephone surveys. Response rates in RDD surveys have always been lower than in face-to-face surveys but higher than in mail surveys conducted without extensive follow-up mailings. However, as response rates have continued to decline in RDD surveys, the potential for nonre-sponse bias in estimates has increased. An important reason for the lower response rates is that technological advances have made it easier for households to identify and avoid telephone calls from unrecognized telephone numbers.

...