A First Amendment test for identifying when a map is functionally equivalent to a facially discriminatory statute.
After listening to the oral arguments in the Supreme Court’s two gerrymandering cases last week, one from North Carolina and the other from Maryland, I think there may be a way to simplify the First Amendment analysis that is causing the Court (and the litigants) so much trouble. The simplification would be based on a distinction between (a) the applicable standard in principle and (b) the statistical evidence required in practice to show a violation of the applicable standard.
The applicable standard would be based on the recognition that it would violate the First Amendment for a state legislature, when enacting election laws, to engage in any facially explicit discrimination on the basis of partisanship. Suppose, for example, that Maryland’s legislature enacted a law explicitly providing that to win an election a Republican candidate must receive a majority of votes (more than 50%), but to win the same election a Democratic candidate must receive only a plurality of votes (more than any other candidate). This law, in other words, would make any votes for third-party, independent, or write-in candidates work to the advantage of the Democratic candidate at the expense of the Republican candidate. Under this hypothetical statute, there would be a runoff only if and when the Republican won a plurality but not a majority; the Democrat would be declared the winner without a runoff based on a mere plurality. This facially asymmetrical state law would be undoubtedly unconstitutional under the First Amendment, as a form of blatant viewpoint discrimination on the basis of partisanship.
It would not matter that the degree of facial asymmetry was relatively small. This hypothetical is not as extreme as if the legislature required Republicans to win elections by a two-thirds supermajority (otherwise, the Democrat would be seated). But that fact would make no difference. Any degree of explicit partisan asymmetry would be a form of unconstitutional viewpoint discrimination.
Justice Alito recognized this point during the oral argument in the Maryland case. He asked the plaintiff’s counsel (at 34): “if your claim is based on the First Amendment, doesn’t that necessarily mean that partisanship cannot be taken into account at all, not one iota?” When the lawyer demurred, Justice Alito pressed: “Have we ever said that there’s such a thing as benign viewpoint discrimination under the First Amendment? . . . Or, you know, you can discriminate on the basis of viewpoint, but it just has to be small?”
The difficulty with gerrymandering litigation under the First Amendment, as I’ve discussed previously, is that it is not a form of facially explicit discrimination. Redistricting laws are not written explicitly in terms of partisanship, but instead in terms of facially neutral lines on a map or words that demarcate geographical boundaries. The challenge is knowing when those facially neutral laws function as a form of viewpoint discrimination on the basis of partisanship.
To be sure, there may be rare cases in which a state legislature brazenly acknowledges that it is drawing the lines in order to discriminate on the basis of partisanship. Indeed, the two pending cases before the Court may be rarities of this kind. But, as several of the Justices observed during the arguments, if the Court condemns self-declared partisan gerrymanders as unconstitutional, then legislatures will not be so brazen in the future, and the Court will need a way to determine which maps actually operate as a form of unconstitutional viewpoint discrimination.
Here is where it may be useful to distinguish between (a) the standard itself and (b) the evidence that shows a violation of the standard. We can state the standard this way: a redistricting map violates the First Amendment if it is the functional equivalent of facial discrimination on the basis of partisanship. A plaintiff is able to win a claim based on this standard if, but only, if the plaintiff presents evidence demonstrating that the map is in fact the functional equivalent of facial discrimination based on partisanship.
How is a plaintiff able to do that? In the absence of an explicit admission, the only way is the kind of statistical evidence that the plaintiffs relied upon in the North Carolina case. If thousands of alternative maps randomly generated by computers demonstrate that the actual map is an extreme statistical outlier insofar as it advantages a political party more than all the alternative maps do, then one can confidently say that the actual map is the functional equivalent of a state statute that contains a facially explicit preference for that political party.
But there still remains the question of how much partisan advantage the statistics must show. The actual North Carolina map is superficially extreme, without regard to statistical analysis, insofar as it regularly produces a 10-3 split in favor of Republicans when the state is more evenly balanced between the two major political parties. Statistical analysis, in fact, confirms that the actual map would not be chosen except by a legislature deliberately wanting to create a structural advantage in favor of one political party—making the map the functional equivalent of facially discriminatory legislation on the basis of viewpoint.
Suppose, however, that statistical analysis would also show that a 9-4 map, or an 8-5 map, would be an extreme outlier relative to all the randomly generated maps; would this showing, then, demonstrate the legislature’s chosen map to be the functional equivalent of facial viewpoint discrimination?
Although the plaintiffs seemed to waffle when faced with this kind of questioning at the oral argument, I think upon reflection that the correct answer clearly must be yes. As a practical matter, this situation seems unlikely. The nature of the statistical analysis is such that a relatively small deviation partisan balance—an 8-5 map, instead of 7-6, for example—is unlikely to end up an extreme outlier in the tail of a statistical distribution. But if it did, it would be demonstrable evidence that the 8-5 map was the product of partisan favoritism, and not viewpoint-neutral legislation. To Justice Alito’s crucial point, if a little bit of viewpoint discrimination is unconstitutional under the First Amendment, then legislation deliberately designed to produce an 8-5 partisan advantage (when viewpoint-neutrality would yield a 7-6 map) is just as improper as legislation deliberately designed to produce a 10-3 map.
In an effort to be clear on this point, if the statistical analysis does notdemonstrate an 8-5 map to be an extreme outlier, then the map cannot be challenged as the functional equivalent of facial viewpoint discrimination. But if the statistics does show an 8-5 map to be an extreme outlier, then it is the functional equivalent of a statute that is facially discriminatory on the basis of partisanship. It all depends upon what the statistical analysis actually shows. The principle remains the same: maps that are statistical outliers are the functional equivalent of facial discrimination. But it depends on the actual evidence of the random maps generated by the computer simulations to know whether a particular map is, in fact, a statistical outlier.
It seems also important to acknowledge that point that several Justices emphasized: deviations from proportionality are relevant to determining whether a map is unconstitutionally discriminatory. In this regard, consider another example: suppose in a given state, the actual map regularly produces an evenly balanced 5-5 split between Republicans and Democrats, but statistical analysis shows that this map is an extreme outlier because most computer-generated maps based solely on nonpartisan geographical considerations yield 7-3 splits in favor of Democrats. Suppose, further, that in this state the statewide balance between Republicans and Democrats is roughly 50%-50%, and thus the 5-5 split in the actual map corresponds to this statewide balance, thereby offsetting the “disproportionality” caused by geographical considerations. If in this situation the actual map would be valid, as it would seem to be under the holding of Gaffney v. Cummings, the reason is that the map does not discriminate against either political party relative to the statewide baseline of partisan balance, and it is not discriminatory for a state to endeavor to replicate this statewide balance when drawing district lines even when doing so is to offset a geographical advantage that one party otherwise might have.
Another way to make this point is to say that if a facially neutral statute achieved proportionality it would not be unconstitutional. For purposes of determining representation in a state legislature, there would be nothing wrong with a state statute explicitly providing that each political party receives the same number of seats as its share of the statewide vote. Although the federal Constitution does not command this kind of strict proportional representation, it also does not condemn it. (This point, by the way, answers Justice Alito’s separate concern about access of speakers to a public park. Making shares of legislative seats turn on a party’s share of votes is not the same as the government making its own viewpoint-based determination, distinct from any content-neutral criteria, of which speakers get to use a public park.)
But a state statute could not, consistent with the First Amendment, explicitly state: “When Republican candidates receive 60% of the votes, they shall receive only 40% of the seats.” That would be another form of facially explicit viewpoint discrimination. Thus, when a map is the functional equivalent of this facial disproportionality, it is equally problematic under the First Amendment. Acknowledging this point does not mean that the First Amendment requires strict proportional representation. It does not. At the same time, however, it can be recognized that deviations from proportionality is relevant to determining whether viewpoint discrimination, or its functional equivalent, is occurring.
Does all this analysis mean that any partisan favoritism in the drawing of district lines is unconstitutional viewpoint discrimination? I don’t think so, and the reason is that partisan favoritism as the motivation underlying a statute is not necessarily the functional equivalent of facial discrimination on the basis of partisanship. A facially discriminatory law has both a discriminatory intent and a discriminatory effect imbedded in the text of the statute itself. The discriminatory effect follows from the operation of the facial discrimination. For example, a tax law that explicitly sets a different rate of taxation for Democrats and Republicans inevitably has the effect of discriminating on the basis of the facial distinction. By contrast, a law that is not facially discriminatory but is motivated by bias might not actually have a discriminatory effect in operation: the example of a facially neutral law that is premised on an inaccurate hunch that it might favor one party is a law that has a discriminatory motive but not necessarily a discriminatory effect.
The “functional equivalence” standard prevents laws motivated by partisanship from being unconstitutional unless they also exhibit a distinct discriminatory effect. A facially neutral law is invalid viewpoint discrimination if, but only if, it operates to have a discriminatory effect that makes it the functional equivalent of facially explicit viewpoint discriminatory.
This standard, in other words, does not invalidate a state legislature’s redistricting map just because it was motivated by partisan favoritism. Instead, it invalidates the map only if that partisan favoritism is tantamount to facially explicit discrimination on the basis of partisanship. But defining the scope of constitutionality liability in this way is just as it should be given the First Amendment. What is impermissible is for the legislature to enact election laws that explicitly turn on viewpoint discrimination, and this standard invalidates those laws—but only those laws—that are essentially the same.