Justin Levitt is a professor at Loyola Law School, Los Angeles; he runs the website “All About Redistricting.”
When the Supreme Court last addressed partisan gerrymandering in detail, all nine justices agreed that “an excessive injection of politics” into the redistricting process violates the Constitution. The court splintered, however, on two pivotal issues: how to know when an injection of politics is excessive, and who should decide.
Four justices proposed various tests; four others wanted federal courts to abandon the field. Justice Anthony Kennedy, occupying his customary center square, rejected all of the proposed tests, but refused to slam the courthouse door entirely on partisan-gerrymandering claims. He expressly invited redistricting litigants to serve up additional standards. And for the last 13 years, they have been trying to solve Kennedy as much as they have been trying to solve excessive gerrymandering.
In Gill v. Whitford, the issue is once again squarely before the Supreme Court, on direct appeal from the first lower-court decision in at least 30 years to strike a redistricting map as a constitutionally excessive partisan gerrymander. Much of the hubbub around the case centers around the “efficiency gap,” a bit of math devised by political scientist Eric McGhee and then refined as a litigation tool with law professor Nick Stephanopoulos (who is also co-counsel for the challengers).
To hear the media tell it, Gill v. Whitford rises or falls on the efficiency gap, held out as a magical new formula to tell us all exactly how much partisanship is too much. It is, ostensibly, the objective and judicially administrable answer to Kennedy’s invitation. That Kennedy and the formula were both born in northern California is, I suspect, just a happy coincidence.
The efficiency gap attempts to track the process of a partisan legislative operative. One way to secure partisan advantage in a relatively evenly divided state is to “pack” opposing supporters into a few districts and “crack” the remainder so that they can’t win the rest, ensuring that your own party’s supporters are more efficiently distributed than the opposition’s. The “efficiency gap” attempts to quantify that disparity. It’s not an estimate of how much district-based partisanship is ideal (overriding all other factors) or “natural” (there is no such thing). But it is a handy flag for particularly skewed results, indicating when an operative with the keys to the redistricting software might have gotten especially greedy. And through historical analysis, it can help signal when that operative’s handiwork might prove particularly durable.
The trial court in this case found the efficiency gap helpful – and as with the winner of last year’s playoffs, everyone’s now gunning for the standard. Like any metric, it has some significant limitations: Among other things, it is quite sensitive to uncontested races, turnout differences and narrow electoral victories. (Sensitivity tests, recommended in the user’s manual, can help with the last bit.) It also measures past performance (because Kennedy said that’s what he wanted); as investors are constantly reminded, though, past performance is no guarantee of future results. Competing theories have emerged to vie for Kennedy’s approval: alternative measures of symmetry, shape or vote weight; deviation from simplified simulations purporting to establish a neutral baseline; or deviation from sophisticated simulations available to those with an in-house NASA-grade supercomputer.
And so current partisan-gerrymandering discourse finds itself in a bit of a math fight.
Which is missing the point.
The efficiency gap is a measurement, not a rationale. It is the readout of a speedometer, or a thermometer, or a scale: meaningless unless you know why you’re measuring. Each of the quantitative alternatives is similar in this respect. They offer different versions of “how much.” But no metric can resolve how much is “too much.”
The legal team in this case is smart enough to know this. And so, despite public acclaim for the measure as the latest silver bullet, the plaintiffs in this case have not actually offered the efficiency gap as the essential element of their legal claim.
Instead, the plaintiffs have urged that legislators engage in “too much” redistricting partisanship when one party attempts to entrench itself durably in power by packing and cracking the opposition. This is a qualitative standard built on democratic theory: It entails elements of invidious intent and discriminatory effect. The efficiency gap, particularly at historically extreme outlier levels, offers one diagnostic flag (among many) for the requisite intent and effect. It can be supplemented, or supplanted, by other evidence – including but not limited to the other quantitative tests mentioned above. (In the lower court, it was supplemented by plentiful evidence about the redistricting process Wisconsin legislators actually pursued.) It can be rebutted as well. The measure is an aid to normative legal principles, not a substitute for them.
* * *
The plaintiffs in this case, and the lower court, have developed an intriguing way to determine “how much partisanship is too much.” That is the question that Kennedy has asked. And it is probably the wrong question.
This case follows Kennedy’s lead in presuming a need for a plaintiff in a partisan-gerrymandering case to show not merely injury, but a constitutionally cognizable burden on voters’ representational rights. This assumes that the Constitution contains a baseline substantive preference for the representation a party “should” have: a threshold number of districts that a particular party “should” win. Affirmative substantive rights are not unknown in constitutional law: Substantive-due-process doctrine is full of them. But locating a threshold of proper partisan representation in the Constitution, and not merely in political science, has been … elusive.
I suspect that Kennedy has gravitated to the “representational burden” threshold because he seeks a screening mechanism. Courts need some way to distinguish an “excessive injection of politics” from business as usual, to throw out the extremes but preserve most maps. A cognizable quantum of burden provides that separation.
There is an alternative, though (described in more detail here). When evaluating excessive politics that may underlie a redistricting map, the Supreme Court should be asking not “how much,” but “what kind.” That slight change of question provides the distinction the court has been seeking, one firmly grounded in the Constitution.
Asking “what kind” focuses on why legislators are acting. This is a familiar judicial inquiry. And it allows the Supreme Court to try to isolate state action undertaken in order to injure citizens on the basis of their partisan affiliation. Partisanship comes in a wide range of flavors; this “tribal” partisanship is among the most pernicious. And it is unlawful, no matter how big the effect.
The constitutional infirmity of tribal partisanship is only hard to spot because we’ve become numb to partisanship in the redistricting process. But imagine a tax that subjected Republicans to an extra payment specifically because of their party affiliation.
That tax is unlawful. Clearly unlawful. And it is unlawful no matter how high it is. A two-cent tax on being Republican is just as unlawful as a two-million-dollar tax, and it is unlawful even if it does not deter a single voter from registering as a Republican. The impact doesn’t make it unconstitutional. What makes it unconstitutional is that it’s unlawful for the state to attempt to injure voters because of their partisan preferences.
Indeed, this is not novel doctrinal ground. In 2016, the court held in Heffernan v. City of Paterson that the intent to injure a state employee based on his partisan affiliation usually amounts to constitutional harm, even with no proven burden on the exercise of First Amendment rights. In the redistricting context, Cox v. Larios, from 2004, says much the same: A small population deviation isn’t ordinarily a constitutional problem, but it’s unlawful if the deviation is driven by intent to injure based on partisan preference.
A test that really turns on motive does the work the Supreme Court wants. It separates unconstitutional partisanship from all of the other properly political choices – like which communities or county lines or roads to follow, choices that may reflect different priorities in different parts of the jurisdiction. (Those political choices are what make it impossible to “just” have computers draw the lines. Computers could create consensus tax plans, too – but only if we agreed on how to program the computers.)
That is, this test doesn’t take politics out of the process, any more than precluding a legislature from punishing voters because they’re Republicans removes the politics from tax legislation.
And particularly given all those other properly political choices, proving that a legislature acted not just despite, but “because of,” an intent to injure based on party will be hard, in all but the really extreme cases. Intent cases aren’t easy to prove. (Just look to the success rate of Batson challenges, alleging intentional discrimination in jury selection.) This is just the sort of screen the Supreme Court seeks.
A test that really turns on motive still leaves space for the efficiency gap. The efficiency gap is designed to pick out anomalies. Many of its quantitative competitors can be calibrated (or are already designed) to do the same. Identifying maps that are extreme partisan outliers is one good way to flag maps that might have been intended to wreak partisan damage – subject, of course, to rebuttal. That helps find the intent that should be at the center of partisan-gerrymandering claims. It’s just not the magic math that “solves” the problem.