Use of the American Community Survey in the Context of the Voting Rights Act

When the 2010 U.S. Decennial Census data start to flow in February 2011, the cycle of redistricting always accompanies the release of these data. More than 10,000 jurisdictions in the United States will have to redraw district boundaries to meet the one-person/one-vote criteria. In addition, as redistricting activity unfolds, experts will have to be aware of the Voting Rights Act, which protects the voting rights of minorities.

In the context of the Voting Rights Act, the Census Bureau’s American Community Survey (ACS) will play an important, but not widely recognized, role. The American Community Survey is an annual survey of the U.S. population that is replacing the social, economic, and housing data that used to be collected through the decennial census long form. This article discusses a number of issues related to how ACS will be used in monitoring and enforcing the Voting Rights Act over the next few years.

A LEXIS search found five cases where ACS data had been citied in legal opinions regarding the act:

  • League of United Latin America Citizens et al. v. Rick Perry, Governor of Texas.
  • United States of America v. the School Board of Osceola County, Florida.
  • United States of America v. Euclid City School Board.
  • Benavidez v. City of Irving, Texas.
  • Benavidez v. Irving Independent School District.

The most intriguing of these cases are the two from Irving, Texas. These are very similar cases: one dealing with the City of Irving and one dealing with the Irving Independent School District (which is largely coterminus with the City of Irving). In both jurisdictions, residents charged that the at-large election system in Irving violated the voting rights of Latinos. In the City of Irving case, the court accepted the ACS data as an appropriate replacement for decennial census data, but in the other case the court rejected the ACS as a substitute for decennial census data. While these cases turn on the use of the ACS in relation to the decennial census data (an issue that will probably not be faced immediately after the 2010 Census data are released), they show that the courts are not yet certain how to treat ACS data relative to the once-a-decade enumeration of the U.S. population. This uncertainty is compounded by somewhat conflicting messages from the Census Bureau. While the Census Bureau reports that the ACS is part of the 2010 Decennial Census Program, they also warn users not to rely on the ACS for population counts, asking data users to focus instead on ACS social, economic, and housing characteristics.

As we look to the post-2010 environment, there at least three ways in which ACS data will be used in voting rights’ cases.

  • Determining citizen voting age population.
  • Determining socioeconomic status of groups.
  • Determining Section 203 (Language Minority) coverage.

Citizen Voting Age Population

The ACS is the only source of local information on the citizen voting age population (CVAP). Such data are critical in being able to satisfy the first prong of the threshold conditions needed to win a “Section 2” voting rights case. Section 2 of the Voting Rights Act prohibits practices that discriminate on the basis of race, ethnicity, or membership in a language minority group. Previous case law (Gingles v. Thornberg) identified three conditions that must be met in order to prevail in a Section 2 voting rights case:

  • The minority group is sufficiently large and geographically compact to constitute the majority in a single-member district.
  • The minority group is politically cohesive.
  • The white majority votes sufficiently as a bloc to enable it to usually defeat the minority group’s preferred candidate.

The first threshold condition means the minority population must constitute the majority of potential voters in a district. In practical terms this means they must make up the majority of the citizen voting age population or CVAP. The table shows data from the 2009 ACS, and indicates that nationwide, 37 percent of Hispanics ages 18 and older are not citizens, compared with 33 percent of Asians, 6 percent of American Indians/Alaskan Natives, 5 percent of blacks, and 2 percent of non-Hispanic whites. These figures vary widely for local areas, but show that CVAP figures are crucial for Hispanics and Asians, who are less likely to be citizens than other racial/ethnic groups.

Percent of Persons Ages 18 and Older Who Are Not U.S. Citizens, 2009

White (alone) 6
Non-Hispanic whites (alone) 2
Blacks (alone) 5
American Indians/Alaskan Native (alone) 6
Asians (alone) 33
Native Hawaiian or other Pacific Islander (alone) 16
Other Race (alone) 43
2 or more races 8
Hispanics 37

Source: 2009 American Community Survey.

However, the only local data for CVAP that will be available prior to the fall of 2011 are five-year estimates from the 2005-2009 ACS. Jurisdictions that have to redistrict early in 2011 because they have elections in the fall of 2011 will have no choice but to rely on these ACS figures.

There are as least five problems with using 2005-2009 ACS data in the context of the Voting Rights Act. First, the 2005-2009 estimates are delivered using the geography from the 2000 Census while the redistricting data (Public Law 94-171), which will be the basis for drawing new districts, will be delivered using geographic units from the 2010 Census. Many of the census tract and block boundaries used in the 2010 Census will be different than those used in the 2000 Census.

Second, the 2005-2009 ACS estimates rely on 2000 Census-based population control totals. In other words, the ACS sample data are inflated to match population estimates that are derived from the 2000 Census population figures. When the new 2010 Census data become available and are used as control totals for the 2006-2010 ACS, there are likely to be significant differences between the 2005-2009 estimates and the 2006-2010 estimates. Some demographers expect the 2010 Census results to be quite different than the population estimates at the national level.1 Without a doubt, many of the county and sub-county estimates from the 2005-2009 ACS that rely on the 2000 Census-based control totals will also differ from 2010 Census based estimates.

Third, the CVAP data from the ACS will not be available at the block level like the data used for redistricting (Public Law 94-171). The regular tabulations available from the ACS through the Census Bureau’s American FactFinder website will not provide CVAP below the census tract level. However, more advanced data users can access CVAP data for block groups through raw data files available for download from the Census Bureau. Block groups are clusters of blocks within census tracts and vary in size from about 600 to 3,000 people. Therefore, to use local ACS data in the voting rights context, one would have to use census tract or block group data to estimate CVAP at the block level. There are different methods available to produce such estimates, so competing sides in legal cases are likely to generate different estimates of block-level CVAP, and courts will be asked to decide which numbers are better.

Fourth, the race categories used in the ACS are not as detailed as those used in the 2010 Census redistricting data (the Public Law 94-171 files). The 94-171 files provide data on 63 race groups (all possible combinations of the five alternative race groups in the questionnaire) crossed by Hispanic and non-Hispanic, which produces 126 cells of data. The ACS voting-age population data are only available for nine racial/ethnic combinations (see table).

The Census Bureau plans to issue a separate set of block group tables in early 2011 with more detailed race groups than are currently available in the ACS tables. This should be helpful, but these tables will still not be as detailed as those in the redistricting files (P.L. 94-171).

Fifth, data for small geographic areas such as census tracts are not controlled to any population estimates. So ACS estimates for small geographic areas are likely to be quite different than the data from the 2010 Census for the same geographic areas. These differences are compounded by the smaller ACS sample size, relative to the decennial census long form of the past, and large margins of error associated with ACS estimates for small geographic areas and population subgroups.

Socioeconomic Status

If the three conditions of the Gingles test are satisfied, courts are asked to consider the “totality of circumstances.” This often involves looking at several factors listed in the Senate report that accompanied the extension of the Voting Rights Act in 1982. One of these factors is an examination of socioeconomic status of affected minority groups. If plaintiffs can show that the minority group has depressed levels of socioeconomic status and political participation, this connection is taken as potentially probative factors. Since the ACS provides local data on income, occupation/employment, and educational attainment, it is the most likely source of data for assessing the socioeconomic status of groups involved in voting rights cases.

Language Minorities

Section 203 of the Voting Rights Act protects language minorities of Hispanics, Asians, and American Indians. On July 26, 2002, the Census Bureau posted a set of covered jurisdictions based on the 2000 Census results in the Federal Register.2 Several hundred counties in 30 states were on the list. Some counties were included for more than one minority group. For example, Broward County, Florida (the Fort Lauderdale area) is on the list for Hispanics and for American Indians (Seminoles). Texas had the largest number of counties covered.

Section 203 requires that a covered state or political subdivision must “provide registration or voting notices, instructions, assistance, or other materials or information relating to the electoral process, including ballots, it shall provide them in the language of the applicable minority group as well as in the English Language.”3

A jurisdiction is covered under Section 203 of the Voting Rights Act if it meets a complex set of criteria, including the number of people who speak the language, the share of the population in a jurisdiction, and the illiteracy rate. Decisions about Section 203 coverage are made at the county level. The ACS is the only possible source of information to systematically assess jurisdictions in terms of the factors listed above. The Census Bureau is working with the U.S. Justice Department to determine how best to provide data for Section 203. Plans for updating the list of covered jurisdictions are uncertain at this point. One important decision is whether to use the 2005-2009 ACS five-year estimates or the 2006-2010 ACS data to make determinations of who is covered.

Using the 2005-2009 ACS data would provide covered jurisdictions with information sooner in the election cycle, which would help advocates and election officials prepare. However, if a jurisdiction has a rapidly growing minority population (Hispanics in Georgia, for example) that is not captured adequately in post-2000 population estimates, then the ACS would underestimate the size of the minority population. Under this scenario, some counties that should be covered would potentially be omitted from the list. Next year’s 2006-2010 ACS, which will be weighted to the 2010 Census results, are more likely to capture those fast-growing areas, but data from the 2006-2010 ACS will not be available until fall 2011, and it would be several months later before the Justice Department could certify which jurisdictions are covered under Section 203.

In summary, there a number of ways in which the ACS data will be used in upcoming voting rights activities. And there are a number of questions about how courts will view the use of ACS data. Will the relatively large standard errors for small area estimates from the ACS lead courts to dismiss this source of information? Will the inconsistencies between ACS and decennial census data cause the court to undervalue ACS data? It may take several years and many court battles to sort out the answers to these questions.