June 2, 2015
Many Americans believe that someone, somewhere in Washington, must be in charge of tracking who is and who isn’t a citizen of the United States. Apparently, so does the U.S. Supreme Court, which just accepted a voting rights case that turns on the government’s ability to count the number of citizens in each voting district. But despite all the talk these days about government and Big Data, the justices, like the rest of us, might be surprised to learn that the most basic information as to who is an American citizen cannot actually be found in any publicly available government data set — anywhere.
The case, Evenwel v. Abbott, poses a question: whether the Constitution’s long-standing “one person, one vote” principle requires equal numbers of voters per district instead of equal numbers of people, as is current practice. Most commentary on the case has focused on its implications for political parties and racial groups. But focusing on the politics, or even on the merits of the constitutional argument, ultimately distracts from a much bigger problem: The data necessary to draw districts with equal numbers of eligible voters does not exist. We have no national citizen database that tells us how many citizens live in each district around the country.
“What about the U.S. Census?” you might be wondering. It’s true that the census releases a data set that provides the building blocks of redistricting plans for Congress, state legislatures, city councils and school boards. But that data set counts just two things: the total number of people, and the number of people over the age of 18, in every community in the country. The data file has no information about which of those people are citizens and which are not. Voter registration lists, another alternative, are notoriously unreliable and highly variable depending on whether an election is coming up — and some states don’t keep track of voter registration at all.
For centuries, unequal districts were as American as apple pie. But in 1964, Supreme Court Chief Justice Earl Warren began striking down maps of electoral districts that had radically different numbers of people than others. Warren was drawing on his experience as governor of California, where a sparsely populated rural county and densely populated Los Angeles each received only one seat in the California state Senate. Warren viewed “one person, one vote” as a critical move to ensure equal political representation. In fact, he considered the redistricting cases to be the most important of his tenure on the Court — even more significant than Brown v. Board of Education. Give people equal representation, he argued, and they will be able to defend themselves and protect their civil rights in the political process.
But while the Court instituted the rule of equality in redistricting over 50 years ago, it has never specified equal numbers of “what” or “whom.” That’s what the Evenwel case is about. Does the Constitution require the use of one statistic — citizens — instead of another — residents — in determining how districts should be drawn?
The Evenwel plaintiffs make a hardly radical argument that ensuring equal voting rights for all must mean that districts should have equal numbers of eligible voters — not just residents. Otherwise, they argue, voters in an area with large noncitizen communities are given an unfair advantage. They point out that certain Texas state Senate districts have roughly twice the number of eligible voters as others, despite having equal numbers of residents. In their view, some Texas voters, particularly Latino voters who live in districts with large noncitizen populations, have much greater power to elect their representative than those in other districts. “One person, one vote,” they argue, should really mean “one voter, one vote.”
As appealing as the logic may be, that transformation in constitutional law would be based on data that simply doesn’t exist. Of course, the census does make available certain estimates of the total number of citizens and noncitizens for states and localities — just as it also estimates the size of the labor force or the total number of bathrooms per household. But the high court is mistaken if it believes these numbers can be used for redistricting.
These census citizenship estimates come from one survey — the American Community Survey — that is based on reporting from just 2 percent of American households. The census releases these estimates for each year, plus averages for the preceding three- and five-year periods. Sometimes the surveys dating from these different years contradict one another, revealing quite different citizenship rates — either because of population migration, or because the survey design isn’t fine-tuned enough to accurately count citizens at the local level. And even when the ACS data remain consistent, they are accompanied sometimes by a large margin of error, just like any public opinion poll based on a small sample set.
What does the Constitution say on the matter? Well, the 14th Amendment explains that apportionment of Congress — that is, the number of members each state gets for the U.S. House of Representatives — is to be determined “according to their respective numbers, counting the whole number of persons in each State.” Persons — not citizens or voters. Now, the Texas case is not about Congress. It concerns districts for the state Legislature, and the Constitution is conspicuously silent on districting for state legislatures. (In fact, the whole notion of districting — for Congress or state legislatures — never appears in the Constitution at all, but that is another matter.)
In general, the court has deferred to the states on which statistics to use to count people. Unsurprisingly, virtually everybody uses the official U.S. Census data — data that count residents, not voters. After all, although the data have their flaws, like omitting many hard-to-count populations, they are far and away more accurate than the ACS numbers. The ACS surveys are also discretionary — they could be defunded and terminated by Congress at any time (the census, on the other hand, is mandated in the Constitution).
Legitimate philosophical arguments can be made in favor of using one set of statistics over another — just as one could argue that some concerns in redistricting, such as keeping counties or communities intact, should override the concern for precise equality. A redistricting plan based on equal numbers of people, the system we have now, also ensures that the workload and constituent-related burdens of representatives are roughly equal. In effect, equal representation may be more important than equal voting power itself.
But the question that confronts the court as it prepares to hear the Evenwel case in the fall is whether the Constitution mandates the use of one statistic to the exclusion of all else. The fact that no accurate count of citizens exists should be the end of the matter. Unless the justices are prepared to mandate a new kind of citizen census — one never contemplated by the Constitution — then they should leave it to the states to draw their districts using the most accurate data available. The one person, one vote rule isn't broken, and the Supreme Court shouldn't try to fix it.
Nathaniel Persily is James B. McClatchy professor of law at Stanford Law School.