Proportional Representation—Brooding and Omnipresent to the End

Guest Blogger

This post was prepared for a roundtable on Voting Rights, convened as part of LevinsonFest 2022—a year-long series gathering scholars from diverse disciplines and viewpoints to reflect on Sandy Levinson’s influential work in constitutional law. 

Nicholas O. Stephanopoulos 

Like Pam Karlan, I want to say a few words about Sandy Levinson’s influential 1985 article, Gerrymandering and the Brooding Omnipresence of Proportional Representation: Why Won’t It Go Away?. Sandy wrote this piece before the Court decided its first partisan gerrymandering case in 1986. Sandy nevertheless anticipated one of the central themes of the Court’s partisan gerrymandering jurisprudence—a concern that eventually helped to doom the cause of action in 2019. This concern was that, if the Court recognized partisan gerrymandering claims, it would eventually be compelled to impose proportional representation on American legislatures. Only proportional representation, went the argument, ensures fairness for partisan, racial, and other political groups. As Sandy put it, proportional representation is “the only political system that offers substantively equal promise to participants in electoral struggles.” 

However, proportional representation (supposedly) has its own flaws that make it an unattractive candidate for judicial (or other) imposition. One problem is that it typically requires multiple parties to ally in the legislature to create a governing coalition. This feature “makes it practically impossible for a strong government to be formed out of the congeries of political parties, each organized around a narrow base of issues.” A second defect is that parties under proportional representation often get to choose their own slates of nominees, without any popular participation. Granted this discretion, parties tend to “pick[] their slates more in response to the imperatives of internal party politics than by consideration of something so abstract as the public good.” And third (and stemming from the second flaw), proportional representation is said to change the political culture for the worse—so that it fixates on political parties instead of nonpartisan virtues like civic republicanism. Per Sandy, it’s “hard to imagine anyone sympathetic to the republican ideal endorsing an electoral scheme that . . . relies on institutionalized parties even more than does the single-member district system.”

Inexcusably, the Supreme Court’s partisan gerrymandering cases never cited Sandy’s argument about proportional representation being the logical endpoint of the Court’s involvement in this area (and this being a bad thing). Despite its lack of citation, though, this argument ran through the Court’s cases like a red thread. Here’s the opening of Justice O’Connor’s separate opinion in Davis v. Bandemer, the 1986 decision that first recognized partisan gerrymandering as a distinct constitutional claim. “It is predictable that the courts will respond by moving . . . toward some form of rough proportional representation for all political groups. The consequences of this shift will be as immense as they are unfortunate.” These are Justice O’Connor’s words, but it’s Sandy’s thesis. 

The plurality that would have reversed Bandemer in the 2004 case of Vieth v. Jubelirer also echoed Sandy. Any legal test for partisan gerrymandering, asserted Justice Scalia for the plurality, must “rest[] upon the principle that groups (or at least political-action groups) have a right to proportional representation. But the Constitution contains no such principle.” And when the curtain finally fell on the partisan gerrymandering cause of action, in the 2019 case of Rucho v. Common Cause, Chief Justice Roberts again began his analysis from Sandy’s perspective. “Partisan gerrymandering claims invariably sound in a desire for proportional representation.” Proportional representation exerts a “gravitational pull.” But “[t]he Founders certainly did not think proportional representation was required.” And the Court’s precedents “clearly foreclose any claim that the Constitution requires . . . that legislatures in reapportioning must draw district lines to come as near as possible to allocating seats to the contending parties in proportion to what their anticipated statewide vote will be.” 

It’s impressive enough that Sandy’s article pioneered an argument that persuaded multiple Justices over multiple decades. But the piece was actually better than that. It correctly distinguished between two ways in which we might talk about proportional representation. One is proportional representation as an electoral system, distinct from our own first-past-the-post, single-member districts. Systems like closed-list proportional representation, open-list proportional representation, and multimember districts with ranked-choice voting all ensure (more or less) that parties’ seat shares are close to their vote shares. The second meaning of proportional representation is parity between parties’ seat and vote shares under any electoral system (including our own). On this view, first-past-the-post, single-member districts are perfectly capable of yielding proportional representation. District lines just have to be drawn so that each party’s vote share translates into approximately the same seat share. 

Sandy grasped this distinction—but the Justices who channeled him didn’t. All their references to proportional representation used the term in its second sense, as the equivalence of parties’ seat and vote shares under the usual American electoral system. When these Justices maintained that proportional representation is the only benchmark relative to which district plans can be evaluated, they simply meant Democratic or Republican seat share equal to vote share. They never contemplated federal courts requiring states to switch from first-past-the-post, single-member districts to some other electoral system. (Though maybe they should have imagined this possibility. As Sandy pointed out, “there is something almost comic about assuming either judicial caution or the inherent legitimacy of the way we have chosen to structure political contests in this country.”)           

Sandy’s article was prescient in one more way. He observed that many votes are “wasted” in first-past-the-post races in that they don’t contribute to any candidate’s election. “[T]he underrepresentation is real, in any single election, for all losing minorities.” He further floated a standard based on wasted votes for assessing partisan gerrymandering challenges. “One way to read [the Court’s one-person, one-vote cases] is to suggest that a state must adopt an electoral system that minimizes the ‘worthlessness’ of votes.” Thirty years after Sandy’s article, a coauthor and I introduced a gerrymandering metric—the efficiency gap—that capitalizes on Sandy’s insights. The efficiency gap is the difference between the parties’ wasted votes in an election, divided by the number of votes cast in that election. A large efficiency gap indicates that one party’s supporters waste many more votes than do the other party’s backers, while a small efficiency gap means that both parties’ voters are about equally cracked and packed. Since we proposed the metric, the efficiency gap has been used extensively by scholars, in litigation, and even in legislation. And its intellectual roots, it turns out, reach beyond what we thought was our own creativity, to a casual aside that Sandy offered a generation earlier. 

The efficiency gap brings me to the one nit I want to pick with Sandy’s article. (Though, to be fair, my complaint involves only the second sense of proportional representation, while Sandy’s piece mostly addressed the first.) The nit is this: It’s just wrong that proportional relationship is the only baseline that can be conceived for evaluating partisan gerrymandering claims. The efficiency gap, for example, idealizes equal wasted votes—not equal seat and vote shares—for parties. Arithmetically, in all circumstances other than a perfectly tied election, parties’ wasted votes aren’t equal if their seat and vote shares are the same. In other words, minimizing the efficiency gap isn’t merely consistent with—doing so actually requires—allocating seats to parties that are disproportional to their votes. 

The point is even clearer with respect to the newest method for detecting partisan gerrymandering: comparing an enacted plan to a large number of maps generated randomly by a computer algorithm without considering partisanship. These computer-created maps might mostly exhibit proportional representation. Or they might not. It all depends on the interaction of a state’s nonpartisan redistricting criteria with the state’s political geography—how Democratic and Republican voters happen to be spatially distributed. As Justice Kagan explained in her dissent in Rucho, this approach “does not use any judge-made conception of electoral fairness—either proportional representation or any other.” Instead, “it takes as its baseline a State’s own criteria of fairness, apart from partisan gain.” Consequently, replacing an enacted plan with a map like those churned out by a computer “could have led to proportional representation.” “Or it could have led to nothing close,” if disproportional representation would be the usual result of nonpartisan redistricting. 

To reiterate, my grievance here is primarily with the Supreme Court, not with Sandy’s subtler understanding of proportional representation. The Court conflated the two meanings of proportional representation while Sandy kept them separate. So maybe this is why the Court never cited Sandy’s article. If the Court had acknowledged the piece, it would have had to grapple with the fact that proportional representation is a they, not an it. 

Nicholas O. Stephanopoulos is the Kirkland & Ellis Professor of Law at Harvard Law School. You can reach him at nstephanopoulos@law.harvard.edu.