A Robust Procedure to Supplement the Coverage of Address-Based Sampling Frames for Household Surveys

Abstract

Address-based sampling (ABS) is an alternative to field enumeration that uses a frame of mailing addresses from the delivery sequence file (DSF). In-person household surveys using ABS are more cost effective and are implemented faster than what is possible with field enumeration. The primary drawback of a frame based on mailing addresses is the under-coverage of households with unlocatable mailing addresses (e.g., simplified addresses). This drawback can be addressed with a frame supplementation procedure. The half-open interval (HOI), a frame supplementation procedure develop for and used on field enumerated frames, does not translate to ABS surveys because of its reliance on a predetermined path of travel for evaluating the interval between the sampled and expected next address. Although a frame based on the DSF has some proximal sorting from the carrier route delivery sequence, this information may not be a sufficient substitute to the path of travel that is essential to implementing HOI. We have developed a new procedure that departs from the HOI and addresses many of the fundamental problems underlying HOI: instead of asking field staff to check an interval between the sampled address and the expected next address we instead ask the field staff to dynamically determine the next address on the ground. This robust, three component procedure called the Check for Housing Units Missed theoretically improves coverage on an ABS study to 100%. In this paper we discuss the implementation and testing of the procedure as well as future applications.