Random-Digit Dialing

Random-digit dialing (RDD) is a method used to select participants for telephone surveys and for related purposes such as selecting control group subjects in casecontrol studies. The basis of RDD is the random generation of telephone numbers that are used to contact potential survey respondents or study participants. Several major U.S. Federal Government public health surveillance projects use RDD, including the Behavioral Risk Factor Surveillance System (BRFSS) and the National Immunization Survey (NIS). RDD does not require the use of telephone directories and has the advantage of including as potential respondents households with unlisted numbers or who have recently moved or changed phone service; failure to include these types of households can seriously bias the sample. However, RDD has the disadvantage that many of the numbers generated may not be in use or may be nonresidential leading to wasted time and effort.

RDD can be a cost-effective method of selecting subjects in an area where telephone ownership is nearly universal. However, it shares with all telephone-based survey methods the disadvantage that households that do not have telephones generally differ systematically from those who do (e.g., in terms of income, education, and other measures of social capital) and these differences can introduce bias into a study. This can be a major concern in some geographical areas; for instance, in parts of the rural Southern United States, as many as 40% of renter households do not have a telephone. In addition, calculating response rates may be more difficult in RDD surveys than in surveys that used a published telephone directory as a sampling frame.

List-assisted RDD can increase the efficiency of the sampling process. The basis of list-assisted RDD is limiting the randomly generated numbers to groups of numbers, known as 100-blocks, which are known to be in use and contain a high proportion of residential numbers. Each telephone number in the United States is made up of 10 digits—the area code (first three digits), the prefix (the next three digits), and the suffix (the last four digits). The first eight digits are sometimes collectively called 100-blocks because they define sets of 100 telephone numbers with the same first eight digits. Lists of these 100-blocks for the geographical area to be sampled, as well as lists of working phone numbers, may be purchased by firms that specialize in providing this information. Comparing the randomly generated numbers to a list of known business numbers and eliminating those that do not also have a residential listing can further improve efficiency, as can use of a machine to detect the dial tone that precedes the ‘number not in service message’ and eliminating these numbers from the sample.

The increasing popularity of cell phones, in particular the increase in households that do not also have a ‘land line’ (traditional phone) has introduced several other issues. Because cell phone numbers have not traditionally been included in telephone surveys, households with only a cell phone (about 7% of U.S. households in 2005) are excluded from the possibility of participation. In addition, concerns such as safety (a person could answer his or her cell phone while driving, which could lead to an accident), cost to the respondent (because cell phones contacts often include a charge for receiving incoming calls), and low yield (because cell phones are disproportionately owned by children and adolescents) are issues that must be dealt with.

...