서베이앤헬스폴리시리서치센터

Lepkowski, J. M., Kim, S. W., and Steeh, C. (2005). “Dual-frame landline/cellular telephone survey design,” presented in the invited session of the Survey Research Methods Section, American Statistical Association, Minneapolis, Minnesota: Joint Statistical Meetings.

Abstract

Selecting telephone numbers both from landline frames (that is, list-assisted RDD frames) and from prefixes reserved for cellular telephone numbers poses significant challenges for subsequent estimation.  It has been assumed in RDD household surveys that telephone numbers are attached to a household, and based on a respondent report on the number of landline telephone numbers in the household, a simple weighting adjustment can account for multiple chances of selection of the household.  Number portability has introduced a small proportion of cellular telephone numbers into the landline telephone number frame, but these can be removed relatively easily through a screening question in the introduction to the interview.  However, the growing number of mobile only households, those with only cellular telephones and no landline telephones, is decreasing the coverage of the landline frame. Dual frame sample designs are being tested which draw numbers from the landline and the cellular number frame. These designs introduce additional chances of selection for individual households with both landline and cellular telephone numbers. Cellular telephones may be used as personal or household devices.  Probabilities of selection in dual frame designs accounting for multiple chances of selection for households and persons are examined.  Weights are proposed, and computed for a Georgia State University survey which selected both landline and cellular telephone numbers. The properties of the weights investigated for such dual frame telephone survey designs.