Consistency and Accuracy of Usps-provided Undeliverable Codes: Implications for Frame Construction, Data Collection Operational Decisions, and Response Rate Calculations
Kristine Wiant
RTI International
Joe McMichael
RTI International
Joe Murphy
RTI International
Katie Morton
RTI International
Megan Waggy
RTI International
When survey mailings are returned as "undeliverable" in the US, we rely on information codes provided by the U.S. Postal Service (USPS) to determine the eligibility of a sampled household. Although little is published about the consistency and accuracy of these codes, they are often accepted at face value and used to make operational decisions. In this paper, we review USPS codes from 3 mailings for the self-administered Residential Energy Consumption Survey (RECS) National Pilot sent within a 12-day period to all 9,650 sampled households with particular interest in codes that influence whether a case is disposed as an eligible occupied household or an ineligible, vacant or seasonal household. Next, we compare USPS codes to field-verified status collected from the 2015 RECS main, field study. Then, we assess a model to predict the occupancy status of a case using auxiliary data that enhances known characteristics from the sampling frame and we assess the impact of that model on occupancy, coverage, and bias. Finally, we discuss the implications for frame construction, data collection operational decisions, and response rate calculations.