613 – Nonresponse Adjustment and Nonresponse Bias Reduction Methods
Randomly Split Zones for Samples of Size One as Reserve Replicates and Random Replacements for Nonrespondents
A.C. Singh
American Institutes for Research
Cong Ye
American Institutes for Research
Low response may render a probability sample behave like a nonprobability sample. High WRR using an NRFUS may be misleading due to instability in the resulting estimator. Release of many reserve replicate samples helps in reaching the target sample size but puts a lot of burden on the nonresponse model-based adjustment. Use of ad hoc substitution by similar units to offset nonresponse is subject to selection bias. As an alternative, a random replacement strategy for unbiased estimation is proposed based on the idea of reserve samples of size one which can be viewed as follow-ups for nonresponding units. It is a take-off from the random group method of Rao-Hartley-Cochran where each stratum is randomly split into groups, and then a single unit is drawn within each group. In the proposed method, each stratum is sorted via implicit stratification before forming zones. The number of zones is about half the allocated sample size. Each zone is randomly split into groups within which replicate samples of size one are selected. Weighted estimates from all responding groups are combined after adjustments for nonresponding groups. Here group RR is meaningful and, in fact, can be made high.