Chapter 4: Using Census Data Sources to Prove Citizenship in Voting Rights Litigation

Two factors combined to make citizenship an important element in voting rights suits. First, in Thornburg v. Gingles, a case arising out of North Carolina and involving a claim by African-Americans, the Supreme Court established a threshold test that plaintiffs must meet to maintain a claim under section 2 of the Voting Rights Act. To be able to maintain a suit, a plaintiff group must prove that (1) it is sufficiently large and geographically compact to constitute a majority in a single-member district, (2) it is politically cohesive, and (3) the Anglo majority votes sufficiently as a bloc to enable it, in the absence of special circumstances, usually to defeat the preferred candidate of the minority group. After Gingles was decided, the lower courts began to analyze and refine its threshold test. An initial question was what sort of majority did the Supreme Court have in mind when it spoke of being able to constitute a majority in a single-member district. Fairly soon, the courts concluded that the majority should be measured by voting-age population rather than total population. The court in Brewer v. Ham, for example, noted that the purpose of the Gingles test is to measure the potential to elect candidates in the absence of the at-large system, and “[o]nly voting age persons vote.” Using this same rationale (i.e., that the purpose of the Gingles test is to measure potential electoral strength), courts concluded that citizenship was also a necessary component of the test, at least in those areas where the racial and ethnic groups did not have approximately equal percentages of noncitizens. As a result, where the issue has been raised because of disparate citizenship rates, the circuits appear to be uniform in finding that the relevant measure in applying the Gingles test is citizen voting-age population.

Second, increasingly, groups that had disproportionate citizenship rates were seeking relief under the Act. Only 67.3 percent of adult Hispanics residing in the United States are citizens. For Asians, the citizenship rate is 69.07 percent. Because the citizenship rates of Anglos and Blacks are virtually identical and approach 100 percent, the very slight variation in the citizenship rates of those two groups has minimal impact on electoral strength. Where, however, one group has an almost 100 percent citizenship rate while only about two-thirds of the other group are citizens and potentially eligible voters, the impact is dramatic. The Hispanic or Asian electoral strength will be well below its percentage of the population, while the Anglo electoral strength will be higher. Additionally, Hispanic electoral strength is further reduced because the median age of the Hispanic population (27 years) is much lower than the median age of the general U.S. population (37 years), meaning that there is a greater percentage of Hispanics who are too young to vote. In those situations, it is critical to be able to determine how many of each group are voting-age citizens.

If being able to determine the number of adult citizens by racial and ethnic group is a necessary element of a voting-rights case, where do we find the data that will give us that information?

I. The Census Long Form and the Sample-Detail File

II. The American Community Survey (ACS)

III. Sample Size

IV. Sampling Error

V. Non-sampling Error

A. Reporting Error

B. Error in the Control Number

C. Over-Reporting Citizenship

VI. Can Citizen Voting-Age Population Numbers Be Used to Draw Districts?

In Burns v. Richardson (1966), the U.S. Supreme Court considered a case in which Hawaii chose to use an apportionment base when drawing legislative districts that reflected the eligible-voter population rather than total population. Because of Hawaii’s status as a major base supporting the military’s presence in conflicts in Asia and because of the large number of tourists who spend time in Hawaii, there were large transient populations primarily located in Oahu who might be included in the census but who did not qualify to register to vote. Accordingly, the state chose to apportion on the basis of citizen population. Because citizen population figures were not readily available, it actually drew its districts on the basis of registered voters. While the Court had concerns about using registered voters as the basis for apportionment, it upheld the districts because the distribution of registered voters was not dissimilar to the distribution of the citizen population. The Court made it clear that in its earlier opinions it had never suggested that the states were required to include “aliens, transients, short-term or temporary residents, or persons denied the vote for conviction of crime in the apportionment base.” It stated that the decision either to include or exclude such groups was not one with which the Court had been shown any constitutionally founded reason to interfere.

While Burns suggested that a citizen-based apportionment was permissible but not required, virtually no one adopted that option but instead opted to use a total-population apportionment base. There have, however, been efforts in the last quarter of a century to persuade the Court to reconsider that question and, in at least two of the cases coming to the Court, to rule that the Constitution required an apportionment base consisting of adult citizens rather than total population, and in each case the Court denied the petition for a writ of certiorari. In 2015, though, the Court noted probable jurisdiction in Evenwel v. Abbott. If the Court rules for Ms. Evenwel and finds that an eligible-voter apportionment base is constitutionally required, it seems likely that districts would need to be drawn on the basis of either registered voters or citizen voting-age population. The Court in Burns expressed reluctance to permit the use of registered voters—at least where it did not approximate the distribution of citizens or another permissible population base. The Court was particularly concerned that the use of a registered voter base was “susceptible to improper influences by which those in political power might be able to perpetuate underrepresentation of groups constitutionally entitled to participate in the political process.”

If the Court does rule for Ms. Evenwel but declines to require apportionment on the basis of registered voters, the alternative is to use the American Community Survey (ACS), which is the only source for citizen voting-age population. As one respected voting-rights scholar explained when addressing the possibility of a requirement to balance districts using citizen voting-age population or a similar measure, “the only relevant citizenship data available from the census gives ballpark figures, at best, and misleading and confusing estimates at worst.” Depending on the outcome of the Evenwel case, it is possible that states and local jurisdictions will be required to conduct redistricting with tools that are wholly inadequate for the job.

VII. Conclusion

Citizenship is an important and often critical element in voting rights litigation and analysis. The basic tool that we have to determine citizenship is the ACS. In using this data source, though, we are typically required to rely on a five-year sample that is subject to large margins of error, especially when applied to small geographic areas and to subgroups of the population. The ACS also is subject to non-sampling error including possible coding errors, faulty estimates of the population that form the base for the numbers reported, and inaccurate responses to the survey questions on naturalization status. Voting rights litigators must be aware both of the information that can be gleaned from the ACS and of its limitations and flaws.