The Census and American Community Survey: Time for Vigilance and Preparation
Terri Ann Lowenthal
With the 2010 census receding in the rearview mirror, civil rights advocates could be forgiven for putting goals such as a fair, accurate and comprehensive enumeration on a back burner. But they shouldn’t.
Yes, the population count portion of the census—required every 10 years by Article I of the U.S. Constitution—is finished. Congressional and legislative redistricting based on the new population numbers is essentially done for the decade, and the Census Bureau has issued scores of data products to help guide public and private investment decisions and the allocation of fiscal resources through program formulas. But there are important reasons to keep the census in the forefront of the civil rights agenda:
- With the advent of the American Community Survey (ACS), the census now extends beyond a moment-in-time snapshot of the population. The decennial enumeration features only the most basic demographic information, including race and Hispanic origin, sex, age, household relationships, and housing tenure (own or rent), while the ACS gathers vital socio-economic characteristics throughout the decade.
- The window of opportunity to research and test new enumeration methods for the 2020 census is small.
- And hanging over these critical activities is a cloud of small-government sentiment and fiscal austerity that threatens to decimate our nation’s vital data infrastructure and undermine a fair and accurate population count less than eight years from now.
The American Census: More than just one, two, three...
The 2010 census was the first since 1940 to gather only the most basic population and housing data from all of America’s households. Previously, the census had used the so-called “long form” to collect a wide range of social, economic and housing information. The long form provided detailed profiles at the community level on important indicators, such as educational attainment, housing conditions and costs, labor force participation and occupation, veteran’s status, ancestry, local transportation patterns, income and poverty, and disability. Congress directly or indirectly required most of the data to administer, monitor and evaluate federal laws and to allocate hundreds of billions of dollars annually in federal grants to states and localities. Long-form data also were used to implement sections of the Voting Rights Act, notably calculations of the voting age population for redistricting and the designation of places requiring language assistance for elections.
But Congress grew weary of the long form after the 1990 census posted the highest disproportionate undercount of people of color (Blacks, Hispanics, Asians, and American Indians) ever recorded. The 1990 census also was the first to be measurably less accurate than the prior census. Legislators urged the Census Bureau to find alternate means of gathering vital data needed for programs and policymaking, to reduce response burden in the decennial population count, and to focus the enumeration on accuracy for the constitutional purposes of congressional apportionment and fair redistricting.
And so was born the ACS, an ongoing canvass of roughly 3.5 million homes each year that asks the same or similar questions as those on the traditional long form. Implemented nationwide in 2005, the ACS compiles enough data over five years to produce statistical profiles for areas as small as an average census tract of 4,000 people. Unlike long-form data, ACS data are updated annually and therefore are more useful in helping policymakers monitor trends and changing conditions at the community level. According to a Brookings Institution analysis, Congress allocated $416 billion in federal grants in Fiscal Year 2008 based on ACS data. The Census Bureau has since added several questions based on new legislative requirements, such as the Affordable Care Act.
With so much riding on ACS data and seemingly inherent congressional interest in the data for policy purposes, you would think lawmakers would support a scientifically robust survey by providing an adequate, consistent funding stream. But you would be wrong.
In the 112th Congress, as pro-small government voices gained more sway with the governing majority in the House of Representatives, a previously slow-moving effort to turn the ACS from a mandatory to an optional survey took hold among growing numbers of primarily Republican members, including committee and party leaders. The House by voice vote adopted an amendment offered by Rep. Ted Poe, R. Texas, to make ACS response voluntary. But that seemingly simple change would have devastating consequences. Previous Census Bureau testing showed that survey response rates would plummet, especially among historically hard-to-reach population groups, and costs would soar by at least 30 percent, making it difficult—if not impossible—for the Census Bureau to produce statistically valid estimates for urban neighborhoods, rural and remote communities, and the diverse range of smaller population subgroups such as Chinese, Dominican, and Haitian Americans.
Emboldened House Republicans didn’t stop there. By a largely party-line vote, they also passed a subsequent amendment to eliminate funding for the ACS entirely. Imagine: No current socioeconomic indicators to determine the distribution of federal and state resources; to guide trillions of dollars in business investments in new stores, services, and manufacturing plants; or to implement key sections of the Voting Rights Act and other important civil rights laws.
Fortunately, neither the Poe amendment nor the measure to wipe-out all funding for the ACS became law during the 112th Congress. But risks to the ACS remain real, with the survey’s critics likely to try again in the 113th Congress.
The Importance of Early Planning for Census 2020
The decennial enumeration is the nation’s largest peacetime mobilization; it takes a full decade or more to conduct thorough research and testing, draw up design plans, and prepare to enumerate 310 million (and counting!) residents in 134 million (and growing!) housing units, as well as in group facilities such as military barracks, prisons, college dorms, and nursing homes.
Yet Congress generally has lacked foresight when it comes to early investment in census research and planning, despite strong evidence that groundwork early in the decade could save taxpayers hundreds of millions, if not billions, of dollars in census execution. Soon after the 2010 census, Congress scolded the Census Bureau for not offering an Internet response option and ordered the agency to develop secure electronic response methods for the next count. Substantial reliance on electronic response in 2020 could save significant resources by reducing the need for paper questionnaires and costly field follow-up, allowing the bureau to devote more resources to hard-to-count populations that might not be able or inclined to respond via the Internet.
But the Census Bureau can’t—and shouldn’t—implement major design changes in the census without comprehensive testing and allowing time for meaningful stakeholder feedback. For example, advocates for communities of color have urged the Census Bureau to revise the way it gathers data on race and ethnicity. A number of Black leaders are proposing more choices for African-American respondents (such as Haitian, Nigerian, Somali, Jamaican) that would better capture this population’s diversity. While the Census Bureau started evaluating alternative questions on race, ethnicity, and ancestry in the 2010 census, much work remains to be done before agency experts can recommend new questions for 2020.
Several states—including New York and Maryland—have taken the lead in changing the way they count prisoners for purposes of state legislative redistricting, preferring to place them at their pre-incarceration addresses—often predominantly urban—instead of at the prison—often hundreds of miles away in a rural area. If the Census Bureau is even to entertain the possibility of changing its census “residence rules,” it must conduct thorough research and testing of alternative data collection methods, which could take years.
And research into a broader use of administrative records—such as the Supplemental Nutrition Assistance Program and Medicaid rolls—could help the Census Bureau identify uncounted members of low-income households. But use of these government databases could raise sensitive privacy issues. Technical issues, such as how to fill in information missing from administrative records (such as race), are also a challenge. The Census Bureau needs to invest in research and testing early in the decade to overcome substantial barriers to incorporating government records into its 2020 census design.
Of course, adequate funding is a continuing issue. For Fiscal Year 2013, President Obama asked Congress for a modest 3 percent increase over his Fiscal Year 2012 budget proposal for the Census Bureau. Senate appropriators were supportive of the request, which would fully fund core programs like 2020 census planning, the ACS, and the 2012 economic census. Their House counterparts weren’t as thoughtful, knocking $92 million from the administration’s proposal. The Census Bureau has a reprieve from impending budget Armageddon, thanks to election-year budget gridlock that produced a temporary measure funding the federal government at FY2012 levels. But census stakeholders should not let down their guard. Congress will continue to grapple with discretionary spending restrictions and must complete action on current year funding bills early in the 113th Congress.
Preserving Census Data: A Civil Rights Issue for the Long Haul
Civil rights advocates put the census and the ACS on a back-burner at their peril. Sure, nonprofit resources are stretched thin and there are myriad important issues to monitor and champion. But think how difficult it would be to make the case for social justice, equal employment, educational opportunity, affordable housing, and access to health care without the comprehensive, timely, and accurate data from the decennial census and ACS.
The decennial census is the foundation of our democratic system of governance; from an accurate census flows the means to ensure fair redistricting and strong implementation of the Voting Rights Act. The ACS puts flesh on the bones of the basic population count, giving policymakers the tools to assess community conditions, identify needs, and set priorities. And all of these publicly available data allow Americans to hold elected officials and business leaders alike accountable for their actions. If we want to promote continued transparency in decision-making, we must fight throughout the decade to preserve and strengthen the only reliable, objective tool available to ordinary citizens—comprehensive, accurate data from the U.S. Census Bureau.
Terri Ann Lowenthal is co-director of The Census Project and a consultant for the Funders Census Initiative.