Say you’re trying to determine whether a district plan that has been used in multiple elections has been highly skewed in a party’s favor. Would you look only at how the plan performed in its first election? Or would you consider all of the results of the elections held under the plan? I think you would choose the latter approach. It wouldn’t be sensible to ignore the outcomes of the subsequent elections. After all, those outcomes provide additional, valuable information about the plan, either confirming or undermining the tentative conclusion drawn from the first election result.
The state of Wisconsin, however, has the opposite intuition. Last week, it filed an amicus brief in Benisek v. Lamone arguing that Maryland’s congressional plan has not been especially asymmetric, despite recording one of the largest average skews in modern history over the 2012, 2014, and 2016 elections. In Wisconsin’s view, only the 2012 election is relevant to assessing the plan. And in that election, the plan was only moderately tilted in a Democratic direction. It was not yet apparent that the plan would go on to exhibit one of the worst average asymmetries of the last half-century.
The flaw in Wisconsin’s position should be obvious: It’s now 2018, and we can’t pretend that the 2014 and 2016 elections didn’t happen. Those elections did take place, and their outcomes need to be taken into account in evaluating Maryland’s congressional plan. Perhaps one could quibble over whether the plan’s average or median skew should be the focus of the analysis (though they happen to be almost identical). But it’s untenable to try to turn the clock back to 2012—to act as if voters didn’t cast ballots, and candidates didn’t run campaigns, in subsequent years.
This point, that more information is better than less, may be a bit banal. More interesting is why Wisconsin felt the need to intervene in Benisek, and to take such an odd stance about the data that should be considered. The state’s reasoning seems to be as follows: In a different case involving North Carolina’s congressional map, the plaintiffs’ expert identified a partisan asymmetry threshold that, if exceeded by a plan in its first election, means that the plan will likely go on to exhibit an average skew of at least one congressional seat. In 2012, the Maryland map’s partisan asymmetry fell below this threshold. Yet certain amici now contend, based on election results from 2012, 2014, and 2016, that the map has indeed been highly asymmetric. This contention is thus inconsistent with the map’s performance in its first election.
Wisconsin’s argument reflects a number of misconceptions that are worth correcting about thresholds in partisan gerrymandering litigation. First, no one has asked the Supreme Court (or any other court) to endorse a particular numerical cutoff, above which a plan would presumptively be invalid. Instead, the plaintiffs’ position in all of the recent cases has been that any setting of thresholds should be deferred until later, when courts have gained more experience with measures of partisan asymmetry. It’s premature now, before a single federal case has reached a final judgment, to try to draw the line between excessive and tolerable asymmetries.
Second, the threshold the expert suggested in the North Carolina case doesn’t apply to plans, like Maryland’s, that are challenged after they have been used in multiple elections. The expert asked what first-election asymmetry has historically been associated with a lifetime-average asymmetry of at least one congressional seat. This approach makes perfect sense when (as in North Carolina) one election is all we have to go on. Then we’d like to know what that one election tells us about a plan’s probable future performance. But the approach is unnecessary when (as in Benisek) a plan has already been in place for the better part of a decade. Then we don’t have to predict how the plan is likely to fare in its second election, its third election, and so forth. We can just examine those elections’ actual results.
And third, the expert’s first-election cutoff may be even less relevant in future cycles, if suits start to be filed before rather than after plans have been used. In this decade, all of the partisan gerrymandering litigation that has enjoyed any success has been ex post litigation, launched after plans have been in effect for one (or more) elections. But if the plaintiffs prevail in either gerrymandering case now pending before the Supreme Court, it’s possible that future litigants will bring their claims ex ante. Ex ante suits, after all, both resolve plans’ constitutionality more quickly and prevent candidates from winning under unlawful maps and subsequently benefiting from their incumbency. Ex ante suits, though, would have no reason to rely on a threshold set on the basis of first election results. Indeed, there would be no such results for them to cite. Instead, ex ante suits would probably assess plans’ performances under a range of electoral scenarios, weighted according to their likelihood. This isn’t hard to do—in fact, today’s gerrymanderers do it all the time—but, understandably, it isn’t what was done in this cycle’s ex post cases.