Use of address-based sampling (ABS) frames for face-to-face surveys can suffer from undercoverage. This undercoverage has typically been more prominent in rural areas but can also occur in urban areas where recent construction of households has taken place. Procedures have been developed to supplement ABS frames in order to address this undercoverage. A new article coauthored by Westat staff, Effects of Address Coverage Enhancement on Estimates from Address-Based Sampling Studies in the Journal of Survey Statistics and Methodology, investigates a procedure called address coverage enhancement (ACE), which supplements the ABS frame with addresses not found on the frame, and the resulting effects the addresses added to the sample through ACE have on estimates.
Michael Jones, M.S., J. Michael Brick, Ph.D., and Wendy Van de Kerckhove, M.A., use weighted estimates in 2 studies to analyze noncoverage bias. “Our findings show that the noncoverage bias was likely minimal across both studies for a range of estimates,” explains Mr. Jones. “The main reason is because the computerized delivery sequence (CDS) file coverage rate is high; when the coverage rate is high, only very large differences between the covered and not covered will result in meaningful bias.”