A guest post by Chris Elmendorf and Eric McGhee:
During oral argument in Gill v. Whitford, the partisan gerrymandering case, Justice Gorsuch complained that the plaintiffs’ proposed test for unconstitutional gerrymanders was too much like a steak rub: “I like some turmeric, I like a few other little ingredients, but I’m not going to tell you how much of each.” The implication is that adjudicating partisan gerrymandering cases would be like judging a Top Chef contest, with jurists relying on their personal preferences to deem the map at issue yummy or unpalatable.
One of us is the creator of the “efficiency gap,” a measure of partisan gerrymandering that has played an important role in this case. Together we filed an amicus brief that outlined the properties and uses of both the efficiency gap and a variety of other partisan gerrymandering metrics. As such, we have an interest in making sure that the social science of this case is understood and used properly. Although Gorsuch might make an excellent steak rub, we don’t think his metaphor caries well to the evidence or proposed standards in this case.
The metaphor is apt for totality-of-circumstances balancing tests, such as the constitutional test for procedural due process, and, arguably, the test for racial vote dilution under the Voting Rights Act. But the tests on offer for partisan gerrymandering claims—including the plaintiffs’ test, and the test suggested in an influential amicus brief by biostatistician Eric Lander—do not invite or require balancing.
There is, first, an objective, well-defined question to be answered: Is the legislative map substantially asymmetric with respect to the conversion of votes into seats—meaning that each party is likely to receive quite different seat shares for a given share of the vote? To ask whether a map is asymmetric in a partisan gerrymandering case is akin to asking, in a toxic torts case, whether a chemical released in an industrial accident causes cancer. A judge in the torts case might consider epidemiological evidence, lab experiments on mice, and biomechanical studies of cell division. But the question to be answered is not whether these three types of evidence, considered together, show the plaintiffs to be morally deserving of compensation (a steak-rub question). Rather, the question is objective: does the chemical cause cancer?
Similarly, the three measures of partisan symmetry introduced by the plaintiffs in Gill—the Efficiency Gap, Gelman-King bias, and the mean-median difference—each serve to answer the objective question of whether a map of legislative districts yields an asymmetric votes-to-seats curve. The measures are extremely highly correlated in competitive states like Wisconsin. They diverge somewhat in politically lopsided states, but the reason for the divergence is well understood and points to a clear choice among the metrics.
If a legislative map were shown to have substantial asymmetry, then under the plaintiffs’ proposed test, the court would ask whether that degree of asymmetry can be explained by neutral factors, such as the geographic distribution of each party’s supporters. No balancing is involved: the court would not weigh the size of the asymmetry against the likelihood that it arose by chance, or against the weightiness of the state’s official (legitimate) redistricting criteria. Indeed, to minimize judicial discretion at this stage, judges could use redistricting simulations to determine whether the map at issue is an outlier relative to the range of algorithmically generated maps.
Courts applying this approach would eventually have to settle on quantitative thresholds for “substantial” asymmetry, and for “outlier” status relative to simulated maps, but this is no different than what the courts did in malapportionment cases after Reynolds v. Sims. Also, while the plaintiffs in Gill formulated the substantial-asymmetry question as a two-part inquiry into magnitude and durability, these steps could easily be collapsed into one if courts focused on the expected rather than the observed level of asymmetry, where expected asymmetry is an average taken over the range of historically plausible partisan swings.
Ironically, the only serious subjectivity in the plaintiffs’ proposed test lies in the intent prong—whether the map was adopted to benefit the favored political party. This inquiry may turn on a judge’s priors in cases where the legislators worked hard to conceal their motives. The irony is that no one disputes that the intent prong is manageable. Intent tests are ubiquitous in constitutional law. But to the extent that the Supreme Court worries about judges simply voting for their party in gerrymandering cases (or being perceived to do so), the Supreme Court could implement the intent prong via conclusive presumptions based on the composition of the legislature (partisan intent presumed if the advantaged party held a majority of the seats when the map was enacted), or based on the results of computer simulations (partisan intent presumed if the map is an outlier relative to the distribution of simulated maps).
Again, our purpose here is not to argue for any particular outcome for the Wisconsin plan. The Supreme Court must decide whether this gerrymander is too extreme. But the Justices need not worry that the available metrics are too variegated for manageable adjudication. Steak rubs are great at the grill, and perhaps in some cases they should season the law too. But partisan symmetry is not a steak rub concept, and Gill is not a steak rub case.