Improving the efficiency of address-based frames with the USPS No-Stat file
Address-based sampling (ABS) frames are based on the Computerized Delivery Sequence (CDS) file, made available to qualified vendors by USPS. Because of low rural coverage for in-person surveys, researchers often resort to hybrid sampling frames that utilize ABS in areas where coverage is expected to be high and traditional field enumeration in areas where coverage is expected to be low. The more areas that are allocated to the ABS portion of the hybrid frame, the greater the cost savings. Since 2009, USPS has made available the No-Stat file, a supplement to the CDS file that contains approximately seven million predominately rural addresses not found on the CDS file. Although the overall coverage gains provided by the No-Stat file are modest, No-Stat addresses are clustered in relatively small geographic areas. In a hybrid frame design, clusters of No-Stat addresses could move some areas that otherwise would rely on field enumeration to the ABS portion of the frame, lowering field costs. This paper measures the efficiencies that are gained by including the No-Stat file for a specified hybrid frame design and coverage estimator. Efficiency gains vary widely across states, and are greater for higher coverage thresholds.