44 – Address Based, RDD and Cell Phone, and Respondent-Driven Sampling
Quasi Address-Based Sampling: A New Cost-Efficient Rejective Sampling Design for Reducing Undercoverage, Nonresponse Rate, and Nonrespondent Substitution Bias
Ned English
NORC
Kyle Fennell
NORC
Jizhou Fu
NORC
Avinash Singh
NORC
We propose a cost-efficient rejective sampling alternative to area sampling (for implementing two-phase multistage stratified cluster sampling designs) to satisfy three goals for reducing (1) nonresponse without field interviewing by making in-person contact for a leave-behind screener to selected HUs for eligibility and contact information for the telephone main interview;(2) undercoverage without counting and listing and without using half-open interval by adjusting selected HUs from the address frame; and(3) nonrespondent substitution bias without releasing additional replicates but by selecting alternate HUs at random in neighborhoods within each segment. The basic premise in rejective sampling for the proposed design is that it would be infeasible to sample directly from the domain subpopulation (such as all valid, eligible and responding HUs in a segment).