Leveraging Auxiliary Data, Differential Incentives, and Survey Mode to Target Hard-to-Reach Groups in an Address-Based Sample Design

Abstract

Identifying efficient and effective ways of tailoring survey designs is critical in the face of decreasing response rates, increasing differential nonresponse, and rising survey costs. Utilizing auxiliary information appended to an initial sample frame allows researchers both to drive differential or tailored treatments across the sample and to enrich back-end analyses, including a simple and efficient form of nonresponse assessment. In this study, we leverage variables appended to an address-based sample (ABS) design to (1) target different levels of incentives to some of the traditionally hardest-to-reach groups (e.g., younger adults, Blacks, and Hispanics) and (2) examine the potential for bias in the survey results by comparing respondents and nonrespondents across an array of key sample frame variables. Using data from a national ABS sample of 4,000 addresses, the study demonstrates that differential nonresponse is a problem with particular sets of addresses on the ABS frame (those with no telephone match), and although use of a mail survey combined with targeted incentives seems to reduce this effect, there is no single incentive structure-survey mode combination that works well across the three hardest-to-reach groups. The approach also allows researchers to assess the potential for nonresponse bias before determining if more extensive (and expensive) nonresponse follow-up methodologies need to be employed.