On-site enumeration is generally regarded as the most comprehensive method for developing sampling frames for area household surveys. However, the time and expense associated with on-site enumeration often precludes it from being a viable option for many household surveys. Residential mailing lists provide an alternative that enables in-person surveys to be done cheaper and faster than is possible with on-site enumeration. The primary drawback of mailing lists is that the completeness of the lists is unknown. In this article, we evaluate the coverage of mailing addresses that were used as a sampling frame for a probability-based survey of 15,000 households in Dallas County, TX. The addresses were obtained from the Delivery Sequence File (DSF) offered by the U.S. Postal Service (USPS) through a nonexclusive license agreement with private companies. The DSF is a computerized file that contains all delivery point addresses serviced by the USPS, with the exception of general delivery. To evaluate the coverage of the mailing addresses, we used Kish's Half-Open Interval (HOI) procedure to search for missed housing units in the interval between the selected address and the next address in delivery sequence order. A total of 46 missed addresses (1.9 percent) were found among the 2,380 HOIs randomly selected for examination. In addition, we discovered that the vast majority of persons who maintained a residential P.O. box also have mail delivered to their street address. Finally, the mailing addresses yielded a 90 percent occupancy rate, which is consistent with metropolitan household surveys that use on-site enumeration methods.
Evaluating the use of residential mailing addresses in a metropolitan household survey