Of critical consideration for the success of regional household travel surveys are socio-demographic and geographic representativeness of the resultant weighted data files. It has well been established that there is a growing resistance by householders to surveys in general and to telephone surveys in particular. This resistance, as well as the changing patterns of household telephone use and access, has resulted in an increased need for advanced and innovative sampling strategies capable of reaching populations representative of the survey universe. As such, we have seen a shift from traditional Random-Digit-Dial (RDD) and General Listed Household (LHH) sampling frames toward Address-Based-Sampling (ABS).

Traditional RDD telephone sample, randomly generated within specified area code and exchange combinations, provides the benefit of ensuring each household in the frame has an equal probability of selection. However, by definition, RDD samples also contain all non-working, unassigned, business, and other telephone numbers, resulting in low survey response rates and high survey administration costs. General Listed samples are pulled from commercial consumer databases and although they contain a wealth of household level socio-demographic information, telephone numbers are limited to those published in the white page directories resulting in the exclusion of cell-phone mostly and cell-phone only households.

The ABS frame presents the survey research industry with an interesting alternative to the RDD and LHH sampling frames. Utilizing the US Postal Service’s Computerized Delivery Sequence File (DSF), which contains over 135 million residential addresses and provides nearly 100% coverage of all households in the U.S., this sample frame can be defined by any level of geography and includes all households regardless of telephone ownership status.

This paper explores the benefits and constraints of the ABS frame as an alternative to other traditional sample frames for regional household travel surveys. Using a survey conducted in 2011-2012, the analysis provides an example of expected response rates and an estimated level of accuracy of residential address information. Finally, the paper presents an innovative method for targeting desired socio-demographic characteristics within the sampling frame, through the use of Census estimates.