AAPOR
The leading association
of public opinion and
survey research professionals
American Association for Public Opinion Research

Appendix A

 Table of Contents 

Address-Based Sampling (ABS) as an Alternative to Sampling Cell Phone Exchanges

 

Given the challenging issues related to cell phone sampling, some researchers have begun to explore other sampling options that altogether forego the use of telephones as a primary sampling unit (Link et al. 2006; Link et al. 2008; Steve et al. 2007). The growth in database technology has allowed the development and maintenance of large, computerized address databases, which may provide telephone survey researchers in the U.S. with a cost-effective address-based sampling (ABS) alternative to RDD for drawing representative household samples. 

To date, address databases compiled based on the Delivery Sequence File (DSF) of the U.S. Postal Service (USPS) have proven most promising. Since the USPS is prohibited from selling or leasing addresses, access to the USPS files is carefully controlled through licensing. The USPS offers a variety of products and services known as Address Management Services (AMS) that allow licensees to improve the quality of their lists. Most of the companies that do list compiling have a DSF license that only allows them to standardize and validate addresses on their list. AMS also provides full access to a weekly snapshot of their Computerized Delivery Sequence (CDS) file to qualified list owners specially licensed by the USPS. Two such vendors are Valassis (formerly ADVO) and Compact Information Systems (CIS). These vendors in turn may license the resulting lists to other list vendors for research and commercial applications. The CDS file contains all delivery point addresses serviced by the USPS, with the exception of general delivery (USPS, 2009).1 Each delivery point is a separate record that conforms to all USPS addressing standards, thereby facilitating the drawing of area probability samples of postal addresses from any geography within the U.S. using the same file structure.  

Benefits of ABS. From a sampling perspective, ABS provides a very high level of coverage, with some estimates placing frame coverage of U.S. residential postal households in the mid-90 percent range. As such, ABS provides an alternative way to sample and reach cell phone only and other cell phone households without having to sample them from cell phone exchanges. The frame also provides coverage of traditional landline households as well as providing access to households with no telephone and newly emerging VoIP-only based computer telephones – groups that heretofore have been underrepresented in traditional RDD telephone survey methods. Additionally, because addresses are in a fixed location, telephone portability is not an issue and sample selection can be conducted with much more geographic precision than can sampling from cell phone exchanges. In particular, geographic eligibility within the sampled target area in ABS typically does not require the onerous and unreliable screening that non-national cell phone surveys may require.

Another important benefit of using an ABS frame is the rich amount of auxiliary information that can be matched to an address, facilitating more complex sample designs and providing information for enhanced contacting and recruiting approaches. Perhaps most importantly, the last name of the “head of the household” can be retrieved for the vast majority of addresses, and in turn, a majority of such addresses can be matched to a landline telephone number via commercial databases, thereby facilitating multiple potential modes of contact with many of the sampled households. In addition to matching landline telephone numbers to addresses, survey sample vendors also can provide case level variables such as Spanish surname indicator for the household head, her/his likely age, as well as geocoding and attachment of census tract information such as the percentage of racial/ethnic groups within a particular local geography, median household income of the area, and in some cases even e-mail addresses. These variables can be used in a number of ways to enhance the survey design, such as through sample stratification on key variables, advance mailings to households, and tailoring of materials, contact scripts, or incentives based on household characteristics such as likely age, race, or ethnicity of the head of household. Moreover, such supplementary data elements can provide valuable enhancements for analytical applications by providing information beyond what a survey can secure. In particular, this will enable survey research to conduct nonresponse bias investigations as such data items will be available for both respondents and nonrespondents.

Implementation of ABS. In terms of survey operations, telephone researchers have tended to choose one of two main approaches to use ABS to reach cell phone only households: (a) sampling and attempting interviews with households from the entire sample or (b) screening to identify cell phone only (or sometimes including cell phone mostly) households. In this respect, the survey designs using ABS mirror those used when sampling directly from cell phone sample frames. With either approach the process used most often involves three steps:

  1. Drawing a random sample of addresses from the target area;
  2. Matching the addresses to directory listed and commercially available telephone numbers – with the assumption that these will be (nearly) exclusively landline numbers (given that list vendors are prohibited from knowingly providing matches to cell phone numbers); and
  3. Surveying just the “unmatched” portion of the sample by mail or in-person to contact cell phone only homes. 


Survey contact can be carried out in one or two stages. With a one-stage approach, the only cost-effective contact mode is via the mail. Researchers can either send a hardcopy questionnaire or direct respondents to self-initiate a Web survey, call-in survey, or telephone-audio computer-assisted (TACSI)/interactive voice response (IVR) survey. With a two-stage approach, an attempt is made to collect a contact telephone number from the sampled address. Again, initial contact is limited to mail for cost reasons, but the future contact information can be collected by various other modes as described previously. Households that return a valid telephone number, be it landline or cell phone, can then be contacted by telephone interviewers if a more traditional CATI survey is being used. Note that because the respondent is willingly providing their number as their preferred contact number (and the numbers are not obtained through database matching), the legal restrictions on the use of autodialing cell phone numbers in the U.S. do not apply. 

In terms of cost, a given number of sampled addresses for ABS are about twice as expensive as a comparably sized sample of telephone numbers. However this can vary broadly based on the sample vendor, total number of cases sampled, and amount of additional data appended to each sampled case. In terms of total survey cost, however, the cost of obtaining an address per sampled ABS unit is minimal. Additionally, because of the efficiency of the ABS frame (i.e., there are relatively few nonresidential addresses if prescreening is conducted by the sample vendor), far fewer addresses (than telephone numbers) are needed in the designated sample in order to reach the requisite final sample size of residential households. 

Drawbacks of ABS. Address-based approaches do, however, have some drawbacks. For example, the quality and completeness of the address information obtained from the commercial vendors varies widely depending on (a) how often the vendor updates the listings, (b) the degree to which the listings are augmented with information from other databases, and (c) whether the vendor purges the records of householders who request that their information not be released (Link et al. 2006). Vendors also differ in their experience with and ability to draw probability samples from the DSF list, which can be problematic if researchers do not wish to draw their own samples. The DSF contains post office (P.O.) box and multi-drop addresses (multiple persons and/or unit numbers associated with a single delivery point address), which may be problematic for in-person and telephone surveys where a street address including apartment number is required to locate the household or an associated telephone number. Such addresses may be less problematic for surveys that use mail as the recruitment mode (such as with mail or Web surveys). Households with multiple mailing addresses (e.g., a street address and a residential P.O. box) introduce selection multiplicities in mail surveys. In some areas, households with a P.O. box address do not receive home mail delivery. This circumstance may be more prevalent in rural areas where a P.O. box may be provided at no cost and no home mail delivery is made. Thus, including P.O. boxes may be necessary to ensure coverage of all households. 

From an operational perspective, ABS often limits the ability of a research organization to conduct quick turnaround studies. Although a majority (more than 60 percent) of the sampled addresses can be matched to a telephone number, the remaining 40 percent must be contacted/recruited first by mail (or in-person) regardless of the actual survey mode used for data collection. This process takes time (and can be quite expensive). As an alternative, an organization can conduct on-going prerecruitment efforts with these “unmatched” cases (i.e., those with no matched telephone number), obtaining telephone contact information from respondents and providing a ready bank of numbers from which to sample for this portion of addresses. This is, however, a relatively expensive and somewhat complex proposition.

If limited to mail only, many ABS surveys would also need to be adjusted in terms of complexity, as complex surveys (i.e., those with complex skip patterns and ones that use various randomizations of item wording and ordering) are not readily feasible with a paper-and-pencil approach. Use of a Web survey option and/or a call-in number to a CATI interviewer can alleviate this problem. However, only households with Web access would be able to use the former approach and relatively few respondents are likely to call in to complete a survey with the latter design.

 

Appendix A References

Link, M.W., Battaglia, M.P., Frankel, M.R., Osborn, L. and Mokdad., A.H. 2008. Comparison of address based sampling (ABS) versus random-digit dialing (RDD) for general population surveys. Public Opinion Quarterly, 72(1): 6-27.

Link, M.W., Battaglia, M.P., Frankel, M.R., Osborn, L. and Mokdad., A.H. 2006. Address-Based versus Random-Digit Dialed Surveys: Comparison of Key Health and Risk Indicators. American Journal of Epidemiology, 164: 1019-1025.

Steve, K., Daily, G., Lavrakas, P.J., Bourquin, C., Yancey, T. and Kulp, D. 2007. “R&D studies to replace the RDD-frame with an ABS-frame.” Paper presented at the 62nd annual conference of the American Association for Public Opinion Research Conference, Anaheim, Calif. 

U.S. Postal Service (2009) Computerized Delivery Sequence. Available online at:
http://ribbs.usps.gov/cds/documents/tech_guides/CDS_USER_GUIDE.PDF (accessed October 20, 2009).

U.S. Postal Service (2009) Delivery Sequence File. Available online at:
http://ribbs.usps.gov/dsf2/documents/tech_guides/DSF2CERT.PDF (accessed October 20, 2009).

 Table of Contents

 


1 The USPS defines general delivery as “An alternate delivery service that allows customers with proper identification to pick up mail at post offices. Provided primarily at offices without letter carrier delivery or for transients and customers who do not have a permanent address or who prefer not to use post office boxes.”  (http://www.usps.com/cpim/ftp/pubs/pub32.pdf)