Balancing Respondent Burden and Data Needs in Question Design: Reducing Respondent Burden in the ACS (303397)

Keywords: Questionnaire Design, Respondent Burden, American Community Survey

The content for the American Community Survey (ACS) is designed to meet the needs of federal government agencies and the ACS data are also a rich source of local area information for state and local governments, universities, and private business. Historically, changes to questions are developed to improve the quality of the data when problems are identified or when changes in society dictate modification to some aspect of the question. However, in order to be more responsive to concerns raised by respondents about the burden of participating in the ACS, the Census Bureau has begun efforts to reduce the burden on the survey respondents. One of the ways in which we are addressing this concern is by studying changes to the questions to be less burdensome while maintaining high-quality data. The ACS questions were first evaluated for the level of burden using several metrics including sensitivity, complaints made to interviewers, median time to answer, and item level response rates. Items scoring “high” on these metrics were reviewed with stakeholders across multiple federal agencies to explore the source of the burden and potential ways to address it, while still meeting federal requirements for the data. These options included modifying question text or instructions, changing the response type from continuous to categorical or changing the categorical ranges, and modifying question order, for example. An initial round of cognitive testing was then done to study the impact of the changes to the questions on respondents, both in terms of comprehension as well as burden. In this paper, we will discuss the metrics used to identify high-burden questions, the options explored for addressing the burden, and results of a first round of cognitive testing.