PRB

Each year, a representative sample of 3.5 million addresses nationwide is randomly selected to participate in the American Community Survey (ACS), which provides information on the economic, social, housing, and demographic characteristics of communities on a continuous basis.1 Data from the ACS influences how more than $675 billion in federal and state funds are spent each year. For the ACS to be a rich, valuable source of information for local communities, it must occasionally be revised to reflect our changing nation. A revised question can be clearer, easier to understand, and yield a higher response rate; revisions may also remove outdated and irrelevant phrasing in questions and instructions. For those who study trends, changes to survey questions—like the changes to the measurement of disability in the 2008 ACS—pose challenges if the revised questions result in estimates no longer being comparable across years.2  For this reason, changes are not made lightly but to improve the quality of the data.

The process of revising the content, including adding a question, typically takes place over a five-year period and involves content tests to ensure that each question and any revisions contain accurate wording, optimal formatting, and appropriate placement.3

The 2019 ACS questionnaire and forthcoming versions of the survey include multiple content changes. The Census Bureau’s proposed changes to the questions were posted in a December 2017 Federal Register notice and open to public comment through mid-February 2018.4 In March 2018, the Census Bureau posted the planned questions of the 2019 survey, including changes to several questions.5

Relationship Revisions Will Improve Estimates of Coupled Households

The Relationships question includes new response categories to improve estimates of coupled households, specifically same-sex couples. For respondents who answer electronically, the survey also includes an automated relationship/sex consistency check to improve accuracy. For example, if an inconsistency exists with the reported sex of an individual and whether their relationship was reported as “Opposite-sex” or “Same-sex” husband/wife/spouse or unmarried partner, the respondent is prompted to make a revision so their sex and relationship responses are aligned.

• The “Husband or wife” and “Unmarried partner” response categories, previously near the end of the response list, are moved to the top and split into four categories:
  • Opposite-sex husband/wife/spouse.
  • Opposite-sex unmarried partner.
  • Same-sex husband/wife/spouse.
  • Same-sex unmarried partner.

• The category “Roomer or boarder” is dropped from the response list.

Telephone Service Question Is Redesigned for Greater Clarity

The increase in cell phone usage has changed how people use and view telephones in households. The telephone question was redesigned to provide clarity and reduce confusion. The question’s intention is to measure a household’s ability to communicate and access services.

• The telephone service question is now a standalone question (previously asked as part of a broader housing question).
• Instruction is added to clarify the type of telephones respondents should include in their response.

New Health Insurance Question Will Improve Understanding of Health Exchanges and More

Following the health insurance coverage question, a new query asks about health insurance premiums paid and subsidies received. Its aim is to gather additional information about coverage categories and improve understanding of health insurance exchanges, Medicaid, and the Children’s Health Insurance Program.

• This two-part question first asks if there is a premium for the plan the respondent selected in the prior question.
• If the respondent answers yes, he or she is then asked if they receive a tax credit or subsidy based on family income to help pay the premium.

The 2016 American Community Survey Content Test tested several additional changes to the health insurance coverage question.6 For example, response categories were reordered, and new instructions were added to clarify that the question only asked about comprehensive health insurance coverage. However, none of these changes were implemented in the 2019 questionnaire.

Journey-to-Work Question Revised to Reflect Course-of-Day Travel

To reflect changes in public transportation infrastructure, transit mode categories in the journey-to-work question were reworded and reordered. These improvements will provide more accurate data on how people travel throughout the course of a day.

• Instructions are simplified to ask respondents to select the one transit mode he or she uses to get to work for most of the distance traveled.
• Several categories include revised wording to increase relevance:
  • “Subway or elevated” is changed to “Subway or elevated rail.”
  • “Railroad” changed to “Long-distance train or commuter rail.”
  • “Streetcar or trolley car” changed to “Light rail, streetcar, or trolley.”
  • “Work at home” changed to “Worked from home.”
• “Trolley bus” is dropped as a mode of transit.

Time-of-Departure Question Adjusted to Be Less Intrusive

Knowing what time of day people commute to work is useful information for transportation planners. However, in the past this question has raised privacy concerns.

• The word “home” is removed from the question to ask it in a less intrusive way.

Weeks-Worked Instructions Revised to Capture Better Details About Workers

The question instructions concerning weeks worked were revised to better understand characteristics of full-time and part-time workers. The revisions will allow analysts to calculate summary statistics of the number of weeks worked (for example, means and medians). Additional clarification is provided to define what to count as “work.”

• Part A is revised to ask if the respondent worked “EVERY week,” where it previously said “50 or more weeks.”
• A clarifying sentence is added to state that paid vacation, paid sick leave, and military service all count as work. Previously, the question stem only stated to “Count paid time off as work.”
• In Part B, respondents who did not work every week are asked to write in the number of weeks worked; the response was previously a check box.
• The reference period (PAST 12 MONTHS) is repeated in Part B, along with additional instructions about what to count as work.

Class-of-Worker Series Revisions Enhance Clarity

The class-of-worker question series was edited to provide clarity, refine definitions, and improve overall layout. The instructional text was edited, and simplified general headings were added to group class-of-worker categories.

• The formerly separate questions are changed to a single series with subquestions.
• “Active Duty” is added as a response category to categorize employment, and the “Active Duty” checkbox is dropped from the Employer Name question.
• An updated definition of unpaid family workers is provided.

Industry and Occupation Questions Are Updated to Improve Quality of Information

Research results suggested that the industry and occupation questions were unclear to respondents and did not provide accurate data. Examples were added or revised to improve the quality of the information received.

• The number of characters allowed for write-in response is increased.
• Question wording is modified and new and revised examples are provided.

Retirement Income Questions Now Encompass More Common Plans

Over the past few decades, the number of defined contribution retirement plans, such as 401(k) plans, increased in popularity while pensions became less common. To improve income reporting and response rates for the retirement income questions, examples were edited to include more common plans.

• The question stem is expanded to read “Retirement income, pensions, survivor or disability income” (formerly “Retirement, survivor, or disability pensions”).
• The examples list is expanded to explicitly reference income held in contribution plans.

Race and Hispanic Origin Questions Are Modified

Respondents self-identify racial and ethnic origins on the ACS questionnaire. The 2016 ACS Content Test included a combined race and Hispanic origin question.7 While this change was not implemented, the 2019 ACS and 2020 Census questionnaires will include slightly modified questions about Hispanic origin and race.

• The list of additional “Hispanic, Latino, or Spanish origins” is modified to include different examples.
• Black and white respondents are now asked to write in their specific origins.