Although people in the United States have historically been reasonably supportive of federal censuses and surveys, they are increasingly unavailable for or not willing to respond to interview requests from federal as well as private sources (National Research Council, 2013b). Moreover, even when people agree to respond to a survey, they increasingly decline to complete all questions, and both survey and item nonresponse are growing problems (National Research Council, 2013b).
In recent years, the American Community Survey (ACS)—the mandatory survey that replaced the census long form that was last used in 2000—has seen an increase in nonresponse, and it has been a target of criticism for invasion of privacy and excessive burden.1 Although it covers far fewer people than the census long form, it is large by any other measure, requesting responses from 295,000 households every month. The ACS is very visible in the public eye, and it generates a small but continuous stream of complaints to members of Congress, who have held several congressional hearings on the survey.
Four items on the survey have been identified by the Census Bureau as giving rise to the most complaints—income, disability, time of leaving for work, and plumbing facilities (U.S. Census Bureau, 2014). Some of these items are seen by many as intrusive, and the questions attempting to
1 In the decennial censuses from 1960 through 2000, a sample of households (one in six in 2000) received a long-form questionnaire that contained additional questions and provided more detailed socioeconomic information about the population, asking more detail than the shorter form that went to all respondents.
measure plumbing facilities have been a long-standing source of jokes and a major source of complaints.
There have also been complaints about the burden of housing-related questions. For example, many household respondents, particularly those who own their homes (about two-thirds of households nationwide) and those with a mortgage (more than two-thirds of homeowners nationwide), face a total set of about 30 housing questions. Other respondents complain that the time required to fill out the survey (estimated at 40 minutes) is too long.
The Census Bureau has responded to the concerns about ACS burden in a number of ways. In 2012, it asked the Committee on National Statistics (CNSTAT) to convene a workshop to consider the benefits and costs of the ACS for a wide variety of nonfederal users of the data products. That workshop considered both the burden of responding to the ACS questions and the importance of the ACS to the nation and the economy (National Research Council, 2013a).
Over the years, Census Bureau staff and outside organizations have carried out research on the costs and benefits of reducing the number of follow-up calls and visits (Zelenak and David, 2013). The results of that research led to a decision to implement some cutbacks. The bureau also established an ombudsman-type position (a “respondent advocate”) to handle congressional and respondent concerns.
Since the 2013 CNSTAT workshop, the tempo of congressional interest has increased, particularly with regard to the mandatory nature of the survey and its burden on respondents. In response, the Census Bureau in 2013-2014 completed a review of the ACS content in terms of the need for each item. Based on that review, it recommended to the Office of Management and Budget that two questions be dropped: one on business or medical office on the property and one on the availability of a flush toilet (U.S. Department of Commerce, 2015). However, it recommended keeping the questions on hot and cold running water and bathtub or shower.
Congressional criticism has continued since that review. In his opening remarks at the workshop, Census Bureau Director John Thompson noted that at the 2016 Senate hearings on the U.S. Department of Commerce budget, a senator expressed concerns regarding the length of the ACS and asked why the Census Bureau could not get the necessary information from the private sector. At the same time, the House of Representatives passed appropriations bills in 2014 and 2015 that would have turned the ACS into a voluntary, instead of mandatory, survey.
STRATEGIES FOR REDUCING RESPONDENT BURDEN
The Census Bureau has conducted an active research, development, and evaluation program to address ACS response burden issues. The components of this program were summarized in a 2015 paper, Agility in Action: A Snapshot of Enhancements to the American Community Survey (U.S. Census Bureau, 2015a). The paper outlined a comprehensive and ambitious program to minimize burden for ACS respondents while still allowing the survey to respond to emerging issues by updating content as needed and maintaining high-quality data.
In her opening remarks at the workshop, CNSTAT Director Constance Citro stressed the importance of maintaining high response rates and relevant content and quality. Thompson expanded on this notion, stressing that the ACS “is extremely valuable to the country. It is used to allocate $400 billion of federal funds a year, and it is the only source of consistent data for many population groups, such as veterans.” He also noted its importance to businesses: For example, at a White House event he attended, a number of technology companies reported on projects that combine ACS data with data for cities and other indicators. The companies had developed a rich array of applications, including one to create opportunities for disadvantaged individuals and identify schools that produced the right kind of skills and another to identify locations of affordable housing within reasonable distance of available jobs.
Steering Committee Cochair Joseph Salvo explained the committee decided to focus on four areas of investigation previously identified by the Census Bureau:
- Building respondent support for the ACS through a communication and education strategy that focuses on respondents and considers stakeholder materials and efforts at marketing or branding the ACS with the goal of increasing participation by increasing understanding of how the ACS data are used.
- Direct substitution of information from administrative records as a means of eliminating some questions.
- Matrix sampling and other statistical methods that could reduce the number of individuals to whom the various questions are posed.
- Changing the strategy for the collection of group quarters data.
This summary of the workshop is organized around a discussion of response burden and those four themes. Chapter 2 defines response burden and summarizes methods that have been employed and suggested to reduce the burden. Chapter 3 addresses means of improving response by building respondent support for the survey. The use of administrative records in
addition to or as a substitute for the questionnaire is discussed in Chapter 4. Using matrix sampling and other statistical methods for reducing the number of respondents or the complexity of the questionnaire is the topic of Chapter 5. Improvements to the collection of information from group quarters are addressed in Chapter 6. The workshop’s discussion of future directions for efforts to reduce burden is summarized in Chapter 7. The agenda for the 2-day Workshop on Respondent Burden in the American Community Survey is in Appendix A, and biographical sketches of the steering committee and presenters are in Appendix B.