Handling Frame Problems When Address-Based Sampling Is Used for In-Person Household Surveys

In-person household surveys usually employ multistage sample designs in which geographic segments are sampled within selected primary sampling units. Within the sampled segments, housing units are sampled from a list of such units. These lists used to be compiled by the survey field staff, but field listing is now often being replaced by lists of residential postal addresses. This paper reviews the frame problems that arise with the use of U.S. Postal Service lists as a sampling frame in the context of in-person household surveys and the methods used to address these problems. The problem of greatest concern is noncoverage, for which the possible remedies are a linking procedure, such as the half-open interval or the address coverage enhancement procedure, or the use of a supplemental frame. Frame problems of clusters of households, blanks and foreign elements, and duplicate listings are also covered.