Solving the Nonresponse Problem With Sample Matching?

Abstract

Over the years, it has become clear that probability sampling is a scientifically sound way to select a sample for a survey. With decreasing response rates, one may wonder, however, to what extent this paradigm still holds. Indeed, due to nonresponse, probability sampling surveys more and more resemble self-selection surveys. Rivers has proposed a different way of data collection based on sample matching. The idea is to select a probability sample from a sampling frame. Selected persons are not asked to complete a questionnaire form, as this would lead to high nonresponse rates. Instead, resembling persons are located in a large (possibly not representative) web panel. These panel members are invited to complete the questionnaire. Their response rate will be high as they agreed to complete survey questionnaires regularly. This article investigates the properties of sample matching in some more detail. Using simulated data, it is explored under which conditions sample matching may work.