A Comparison of Mixed-Mode Address-Based Sampling (ABS) Versus Random Digit Dialing (RDD) for General Population Surveys (original) (raw)
Related papers
2006
As part of an on-going effort to expand coverage for and participation in the Behavioral Risk Factor Surveillance System (BRFSS), a pilot study was conducted to determine if a mail survey conducted with a random sample of adults selected from an address-only sampling frame (the USPS Delivery Sequence File) could rival the quality of the data collected using more traditional random-digit dial (RDD) methods. As one of the largest, RDD-based health surveys, the BRFSS is an ongoing surveillance system designed to collect uniform, state-specific data on preventive health practices and risk behaviors that are linked to chronic diseases, injuries, and preventable infectious diseases in the adult population. Moving from a telephoneonly approach to one which utilizes alternative survey modes and sampling frames has the potential to improve response rates and increase the validity and reliability of the estimates obtained. This paper reports the results of a pilot study conducted in six states. Item non response rates and estimates derived from mail and telephone versions of the BRFSS are compared across 8 key health conditions and risk behaviors, including diabetes, high blood pressure, obesity, alcohol use, and HIV testing. While a considerable amount of research and refinement will be necessary before DSF-based survey designs can be developed which truly rival current RDD designs, the data produced appear of reasonable completeness and the estimates produced are largely equivalent to those produced using current telephone survey methods.
Address-based versus Random-Digit-Dial Surveys: Comparison of Key Health and Risk Indicators
American Journal of Epidemiology, 2006
Use of random-digit dialing (RDD) for conducting health surveys is increasingly problematic because of declining participation rates and eroding frame coverage. Alternative survey modes and sampling frames may improve response rates and increase the validity of survey estimates. In a 2005 pilot study conducted in six states as part of the Behavioral Risk Factor Surveillance System, the authors administered a mail survey to selected household members sampled from addresses in a US Postal Service database. The authors compared estimates based on data from the completed mail surveys (n ¼ 3,010) with those from the Behavioral Risk Factor Surveillance System telephone surveys (n ¼ 18,780). The mail survey data appeared reasonably complete, and estimates based on data from the two survey modes were largely equivalent. Differences found, such as differences in the estimated prevalences of binge drinking (mail ¼ 20.3%, telephone ¼ 13.1%) or behaviors linked to human immunodeficiency virus transmission (mail ¼ 7.1%, telephone ¼ 4.2%), were consistent with previous research showing that, for questions about sensitive behaviors, self-administered surveys generally produce higher estimates than intervieweradministered surveys. The mail survey also provided access to cell-phone-only households and households without telephones, which cannot be reached by means of standard RDD surveys.
Using Addresses as Sampling Units in the 2007 Health Information National Trends Survey
The Health Information National Trends Survey (HINTS) is a recurring household survey that collects data about the sources and characteristics of health information 2 from a representative sample of all adults in the United States. Data for the 2007 HINTS were collected by telephone from an RDD sample of phone numbers and by mail from a sample of addresses. This paper focuses on the address sample, which was selected from a sampling frame based on the U.S. Postal Service's Delivery Sequence File and supplemented with non-USPS sources to increase coverage in rural areas. In this paper we discuss the address sample stratification, the household and person-level weights, rates of non-delivery and response by type of address, the handling of multiple ways a household can receive mail, and the characteristics of the supplemental list of addresses.
Coverage Gap : Out-of-State Phone Numbers for State Surveys
2018
As survey designers contemplate using a single frame cellphone design for telephone RDD surveys they consider the potential for coverage bias. From a coverage perspective, there has been discussion about the impact of excluding landline only households. There has not been as much consideration to the impact on coverage bias regarding residents of an area of interest who do not have a telephone number associated with that area. Unlike with landline numbers, a cellphone number is portable and therefore many residents of an area to be studied might not have a phone number associated with the area. Moreover, the latest data indicate that 50.5% of adults can only be reached by cellphone (Blumberg & Luke 2017). As more cellphone only persons move from where they purchased their phone, the undercoverage rate in the area to which they move increases and the potential impact on bias increases. Based on auxiliary data, the undercoverage rate averaged 3% in Ohio and was as high as 40% in some ...
Journal of Health Communication, 2010
Research on mode of administration of surveys increasingly appears in the literature. Little research includes comparisons by sample frame as well as by mode. This research examines differences in efficiency using two types of sample frames (address-based samples [ABS] or random digit dialing [RDD] samples) and multiple modes (web-based surveys, mailed questionnaire, and telephone interview) among adult respondents. Matching telephone numbers to addresses was conducted on both samples. A test of the effectiveness of making modifications to drop point locations in the ABS was also undertaken. A higher proportion of addresses were matched to telephone numbers in an ABS but with less accuracy than matching telephone numbers to addresses in an RDD sample. Costs per competed interview were lower using the RDD than when using the ABS. Efforts to specify apartment numbers in drop point locations in the ABS were not found to be cost effective. Overall, for both demographic and substantive question outcomes, survey frame has less of an impact than survey mode on measures of response rate and cost.
2009
The use of commercially-available address lists as a replacement for traditional listing has been a topic of interest in recent years among survey research organizations due to the potential of substantial cost savings. Of concern has been the possibility that the coverage of commercially-available listings is inferior when compared to the traditional "gold standard", at least in specific environments. NORC has employed the Valassis (formerly ADVO) commercial address list as the basis for area-probability studies in urban and suburban areas since 2003, while still conducting traditional listing in rural areas. The reason for the distinction has been the understanding that residents of rural areas tend to have PO BOX mail delivery, and thus do not have addresses that can be assigned directly to Census geographies through GIS (Geographic Information Systems) technology. NORC conducted a field examination of Valassis address list on a nationallyrepresentative sample of segments during the summer of 2008. The purpose of this address validation was to determine the coverage of the list generally, but with a specific interest in areas considered to be too rural for list sufficiency. Our paper first compares the results of our validation with what was expected using model that predicts list coverage based on a-priori information, such as urbanicity, population density, segment morphology, and other relevant attributes. Secondly, we present further refinements to our coverage prediction model, and demonstrate how it can be applied to any segment before listing begins. In so doing we argue that the suitability of a segment for commercial address lists or traditional listing contains a substantial random component. The third component of our research discusses the improvements that address list vendors have made in rural areas, and considers the implications for the ongoing need for traditional listing.
Nonresponse bias has become a big concern in all telephone surveys conducted in the U.S. Like other RDD telephone surveys, response rates in the Behavioral Risk Factor Surveillance System (BRFSS) have declined steadily since the mid-1990s. In order to improve state-level BRFSS response rates and to assess nonresponse bias, BRFSS is using a mixed-mode data collection design to achieve these goals. A mail follow-up survey to non-respondents of the landline telephone survey was implemented at 4 states in 2010. Characteristics of respondents at the mail follow-up survey (N=2,390) were compared to those from the landline survey (N=11,646). We developed logistic models for each of 14 health indicators to examine whether exclusion of adults from the mail follow-up survey affected estimates after adjusting for demographic characteristics. The extent of the potential for non-response bias in landline telephone surveys was estimated. Using the mail follow-up survey for non-respondents of the ...
Where to Start: An Evaluation of Primary Data-Collection Modes in an Address-Based Sampling Design
Public Opinion Quarterly, 2015
As address-based sampling becomes increasingly popular for multimode surveys, researchers continue to refine data-collection best practices. While much work has been conducted to improve efficiency within a given mode, additional research is needed on how multimode designs can be optimized across modes. Previous research has not evaluated the consequences of mode sequencing on multimode mail and phone surveys, nor has significant research been conducted to evaluate mode sequencing on a variety of indicators beyond response rates. We conducted an experiment within the Racial and Ethnic Approaches to Community Health across the U.S. Risk Factor Survey (REACH U.S.) to evaluate two multimode case-flow designs: (1) phone followed by mail (phone-first) and (2) mail followed by phone (mail-first). We compared response rates, cost, timeliness, and data quality to identify differences across case-flow design. Because surveys often differ on the rarity of the target population, we also examined whether changes in the eligibility rate altered the choice of optimal case flow. Our results suggested that, on most metrics, the mail-first design was superior to the phone-first design. Compared with phone-first, mail-first achieved a higher yield rate at a lower cost with equivalent data quality. While the phone-first design initially achieved more interviews compared to the mail-first design, over time the mailfirst design surpassed it and obtained the greatest number of interviews.
Incorporating A MultiMode Design Into A Random-Digit-Dialing Survey
Random-digit-dialing (RDD) surveys are typically conducted using only the telephone survey mode of data collection. However, because RDD survey response rates have been declining over the past ten years it is important to examine alternatives to the single-mode approach. This paper describes a test of one multi-mode alternative conducted as part of the Behavioral Risk Factor Surveillance System (BRFSS), a monthly state-based RDD survey that measures health conditions and risk factors. In six states, a list-assisted RDD sample was selected and matched with a database of residential telephone numbers and addresses. For the sampled telephone numbers with an address match, a mail survey with telephone survey follow-up of mail nonrespondents was employed. For sampled telephone numbers with no address match, the survey was conducted by telephone alone. After discussing the design and implementation of the six-state pilot survey, including procedures for the selection of the adult respondent(s) from the sample households, we focus on response rates and respondent characteristics by mode using the ongoing, single-mode BRFSS survey in each state as a comparison sample. The paper also discusses weighting and estimation issues for multi-mode RDD designs and examines potential mode effects.