Improving Random Respondent Selection in Telephone Surveys

Abstract

A split-ballot respondent selection experiment was conducted on a probability sample of Illinois residents surveyed by telephone. Respondent selection was conducted in two ways: in a random half of the 2252 households, all adults were enumerated and one was selected randomly; in the other half, household informants were asked to identify the adult in the household whose birthday was the most recent. Results indicate that the birthday method is valid and effective.