An Enhanced Approach for Solving the Overlap Problem in Dual-Frame RDD Surveys

• Activity Number: 514
• Type: Contributed
• Date/Time: Wednesday, August 3, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Section on Survey Research Methods
• Author(s): Sohyung Park*+ and Sun-Woong Kim and Michael Traugott and Sang-Kyung Lee
• Companies: Dongguk University and Dongguk University and University of Michigan and Hyundae Research Institute
• Keywords: household number ; shared number ; personal number ; person-level weighting

Abstract

Dual frame RDD designs of landline and cell phone numbers to eliminate the coverage bias due to cell-only populations have been popular in many countries, but they are subject to the overlap and overrepresentation problem. One of the solutions to the problem is to use a compensatory weight for unequal probabilities of selection, but weighting is often inappropriate because of lack of information on actual phone usage. We consider a two-stage procedure, illustrated by Lepkowski and Kim (2005), different from a conventional within-household selection. First, with the assumption that both landline and cell numbers are 1) for the household, 2) shared, or 3) personal numbers, a few questions are used to identify such status from the informant for each number selected from the frame. Second, if the phone number is for the household or a shared number, one eligible person using the number is randomly chosen and asked to provide the information on other phone numbers that could have been reached to the person. This approach was successfully applied to a national survey in Korea. Several person-level weighting strategies based on phone ownership data from the survey are examined.