You are using an outdated browser.
Please upgrade your browser
and improve your visit to our site.
Skip Navigation

I Did Not 'Recant' on Voter ID Laws

Richard A. Posner on judges, data, and consequences

A month or so ago, a new book of mine, called Reflections on Judging, was published by the Harvard University Press. I have been a federal court of appeals judge since 1981, and over this extended period I have become acutely conscious of certain deficiencies of the federal judiciary, and those deficiencies are the principal focus of the book.

To my considerable surprise, one sentence—I should have thought it entirely innocuous—in the book has received unusual attention in the media and blogs, much of it critical. The sentence runs from the bottom of page 84 to the top of page 85, in a chapter entitled “The Challenge of Complexity.” The sentence reads in its entirety: “I plead guilty to having written the majority opinion (affirmed by the Supreme Court) upholding Indiana’s requirement that prospective voters prove their identity with a photo ID—a type of law now widely regarded as a means of voter suppression rather than of fraud prevention.” (The footnote provides the name and citation of the opinion: Crawford v. Marion County Election Board, 472 F.3d 949 (7th Cir. 2007), affirmed, 553 U.S. 181 (2008).)

To understand the sentence in context, one must read the portion of the first sentence of the paragraph in which it appears. I say that “the Supreme Court and the lower federal courts have managed to enmesh themselves deeply in the electoral process without understanding it sufficiently well to be able to gauge the consequences of decisions.” I did not say that my decision, and the Supreme Court’s decision affirming it (written, be it noted, by the notably liberal Justice Stevens), were wrong, only that, in common with many other judges, I could not be confident that it was right, since I am one of the judges who doesn’t understand the electoral process sufficiently well to be able to gauge the consequences of decisions dealing with that process. I may well have been wrong in Crawford, because laws similar (I do not say identical) to Indiana’s represent a “type of law now widely regarded as a means of voter suppression rather than of fraud prevention” (emphasis added)—“now” referring to the fact there has been a flurry of such laws since 2007, when my opinion in the Crawford case was issued, and they have been sharply criticized.

If you go back and read my opinion in Crawford, you’ll see that at least I was conscious of the fact that a voter ID law is two-edged. On the one hand, the more, and more-reliable, identification that it requires, the more it reduces the likelihood of voter fraud (impersonation of the person actually eligible to vote). On the other hand, the more such identification the law requires, the more it tends to disenfranchise eligible voters: for example voters who don’t drive, hence don’t have a driver’s license (which would have their photo on it), and may find it difficult to obtain the required photo ID.

The opinion states (citations omitted): “The new law’s requirement that the would-be voter present a government-issued photo ID, such as a passport or a driver's license, is no problem for people who have such a document, as most people do. Nor is it a problem for people who vote by absentee ballot or who live in nursing homes—and anyone 65 or over can vote by absentee ballot. But what about people who do not have photo IDs and must vote in person, if they vote at all, because they don’t live in nursing homes and are ineligible to cast absentee ballots, though the eligibility requirements are not stringent? They can get a photo ID from the Indiana motor vehicle bureau by presenting their birth certificate (or certificate of naturalization if they were born outside the United States) or a certified copy, plus a document that has their name and address on it, such as a utility bill. Both the indigent and the nonindigent who does not have (or have with him) a photo ID can, if challenged, cast a provisional ballot and then has 10 days either to file an affidavit of indigency or to procure a photo.” The conclusion? “The Indiana law will deter some people from voting.”

But this was conjecture. We judges weren’t given, in Crawford, the data we would have needed to balance the good and bad effects of the Indiana law. We pointed out that “the principal evidence on which the plaintiffs relied to show that many voters would be disfranchised was declared by the district judge to be ‘totally unreliable’ because of a number of methodological flaws; and we accept her finding.” Given such empirical uncertainty, we naturally were reluctant to invalidate the law in the name of the Constitution; to have done so would have plunged the federal courts deeply into the management of the electoral process—a managerial responsibility that sections 1 and 4 of Article I of the Constitution actually consign to the states.

It’s not as if there had been good evidence about the frequency of voter fraud in Indiana. Our opinion pointed out that “as far as anyone knows, no one in Indiana, and not many people elsewhere, are known to have been prosecuted for impersonating a registered voter. But [this could not be decisive, because] the absence of prosecutions is explained by the endemic underenforcement of minor criminal laws (minor as they appear to the public and prosecutors, at all events) and by the extreme difficulty of apprehending a voter impersonator.”

Given our uncertainty about the effects of the Indiana law, a majority of our three-judge panel declined to invalidate it. One judge, Judge Terence Evans (since deceased), dissented, but not on the basis of empirical data, rather because “as far as anyone knows, no one in Indiana, and not many people elsewhere, are known to have been prosecuted for impersonating a registered voter.” A recent article on the case in Legislation and Public Policy, by Richard Trotter, “Vote of Confidence: Crawford v. Marion County Election Board, Voter Identification Laws, and the Suppression of a Structural Right,” remarks: “The evidentiary gaps that proved decisive in Crawford were a product of the relative novelty of voter identification laws and the lack of mainstream scholarly and journalistic attention dedicated to its potential effects.” Precisely.

The point I was making in my book in mentioning the Crawford case was not that the decision was right or wrong (I was not, as one journalist has it, “publicly recanting”; I had not “switched sides”), but that in many cases judges can’t have any confidence in the soundness of their decisions if they do not have empirical data concerning the likely consequences of deciding the case one way rather than another. Some judges will disagree with this. They will say that judges can use some combination of careful parsing of authoritative texts with conjecture or common sense or life experiences to decide cases soundly. That is often true but often false, as the Crawford case illustrates.

The main criticisms that commenters on the “I plead guilty” sentence from my book make are two, apart from partisan comments (Indiana’s voter ID law is clearly good or clearly bad). The first is that a judge shouldn’t admit a mistake, or even the possibility of a mistake, because that undermines faith in the judicial process. That’s ridiculous. Obviously judges are not infallible. Much of American law is quite vague, and even when it is clear, the information that judges would need to apply it with any confidence in the soundness of the application often is missing. In these respects Crawford is typical. My book gives many other similar examples and particularly emphasizes the challenge of modern technology, advancing with blindingly rapid strides, to judicial understanding. As Peter Lee said in a recent Yale Law Journal article on patent law, “As a general matter, lawyers and science don’t mix.” And for “lawyers” one can substitute “judges.” I consider this deplorable.

The second criticism of “I plead guilty” is that judges shouldn’t state publicly how they would, or would be inclined to, vote in a future case. This becomes another reason for thinking that judges shouldn’t admit to mistakes or even uncertainty, as such an admission could have implications for how they may vote in the next, similar case. I think it’s true that a judge shouldn’t announce his votes in future cases, because that might create in the judge an emotional commitment to vote that way even if the lawyers or the other judges present good reasons, unanticipated by him, for voting differently from how he planned to vote. But whether judges (like me, and before me more illustrious judges such as Holmes and Cardozo and Learned Hand and Henry Friendly) publish books or articles about law (as have both Justice Breyer and Justice Scalia on the current Supreme Court), or merely publish opinions giving their reasons for voting one way or another in a case, they will be providing a great deal of information about their inclinations, their policy insights, their liberal or conservative leanings in various areas of law, in short their priors. And whatever judges may say self-servingly about being merely “umpires,” just calling balls and strikes and indifferent to who wins and who loses, their priors influence judicial decisions in areas in which the law is in flux because authoritative sources of law, such as the Constitution and statutes and precedents, do not yield clear answers to new legal questions. An experienced judge is not a tabula rasa. My book argues that judges need a better understanding of the consequences of their decisions—including decisions that may have a profound impact on elections. I stand by that argument.

Richard A. Posner is a judge on the U.S. Court of Appeals for the Seventh Circuit and a senior lecturer at the University of Chicago Law School.