Number Changed

When a telephone survey is conducted, some dialings will result in a message stating that the number they have dialed has been changed. The message often includes the new number. “Number changed” dispositions in RDD samples are normally classified as ineligible, but there are some circumstances for which the researcher might want to considered such numbers eligible and call the new number.

When a household changes its landline phone number, it is usually because that household has moved to a new location. If the move was to a location outside the exchange boundaries of the old telephone number, a U.S. household traditionally had to obtain a new telephone number in an exchange serving their new address. However, number portability has changed this by allowing people to keep their telephone number when they move. A household also might change its number without relocating. For example, they might want to replace their directory-listed number with a new, unlisted number. Or a household might elect to change their service provider to a competitive local exchange carrier or to a provider of Voice over Internet Protocol (VoIP) services. In areas where number portability is not available, the household would be required to get a new telephone number in order to change service providers.

Usually numbers that result in a “Number changed” message are removed from a random-digit dialed (RDD) sample because the new geographic location makes them ineligible for the survey. A more important reason for their exclusion is that an RDD sample could have allowed both numbers—the old number and the new number—to be eligible for selection. If researchers elect to call the new number, they must also apply a weight to correct for the multiple probabilities of selection of that household.

However, for telephone surveys that sample specifically named persons—for example, when working from a client-supplied list of customers—it often will be appropriate to call the new number. Other instances in which it would be appropriate to call a changed number would be (a) recontacting a respondent for a follow-up interview and (b) conducting longitudinal surveys of the same respondents. In such cases, multiple probabilities of selection usually would not be a concern.