We conclude this section by discussing and clarifying ten misunderstandings about symmetry and partisan bias that should be avoided. We have drawn our illustrations of these misunderstandings primarily from the briefs of various of the Appellees and amici in Jackson v. Perry, one of the cases consolidated under the name LULAC v. Perry. We believe the discussion below can be helpful to courts reviewing expert witness testimony and legal arguments about the nature of partisan bias and partisan fairness.
First, it is sometimes claimed, as in the Brief of Appellees Tina Benkiser, Chairman, Republican Party of Texas, and John DeNoyelles in Jackson v. Perry60 that the methods described above cannot be applied because results will be so dependent upon which particular elections and other data are used that no meaningful calculations about the magnitude of partisan bias can be provided. Given modern technology, that claim is false. Experts can disagree about which set of input data is relevant for a given case, but the resulting measures of partisan bias normally will not differ to any significant degree because the best current methods are quite robust to changes in model specification. Because social scientists studying partisan bias rely on the symmetry standard as the baseline for attempting to calculate the degree of bias, differences of opinion between experts will be limited in scope. In practice, differences across experts in the academic literature, and even those on opposite sides in redistricting litigation have usually been minor, especially when considering the margin of error properly calculated. We expect that the methodology of ascertaining the seats-votes curve from real data, and then measuring partisan bias (and electoral responsiveness, if desired) will continue to improve, but the current state-of-the art methodology has now matured to the point where the vast majority of the available relevant information is now incorporated in the methods, and so future improvements will likely be incremental, such as helping to handle special cases or unusual situations.
Second, some make the claim that the symmetry standard is yet another version of a proportionality rule.61 This assertion is factually incorrect.62 Those making such a claim are conflating two very different concepts, proportionality and symmetry, by confusing electoral responsiveness with partisan bias. Perfect proportionality is allowed as one possible fair system under the symmetry definition of fairness, but symmetry does not require proportionality. Fairness in the form of symmetry can exist in electoral systems that are far from proportional. In particular we can have the absence of partisan bias (or its presence at only very low levels) in systems such as plurality elections in the U.S. that have electoral responsiveness values that are very far from one. On the other hand, limiting the level of partisan bias in a system has no necessary effect on the system's level of electoral responsiveness, and thus will have no necessary effect on the degree to which a system is proportional.
Third, not all methods of analyzing seats-votes curve are equally useful to the courts. As demonstrated in the discussion of the evolution of the social science literature on this subject, the statistical technology to investigate seats-votes relationships has now matured to the point where it can be counted on to provide reliable information for public policymakers, the courts, and others involved in redistricting. However, when laypersons are discussing data on seats-votes curves they sometimes present and analyze that data in a misleading way. For example, in the State Appellees Brief in Jackson v. Perry, and in the Republican Party of Texas Brief in Jackson v. Perry, data are reported on statewide vote shares and on statewide seat outcomes and the two numbers are compared to provide an indicator of partisan fairness.63 But, this simplistic mode of analysis makes it virtually impossible to distinguish discrepancies between seat share and vote share that are caused simply by levels of electoral responsiveness higher than one from discrepancies that are caused by actual partisan bias.
A fourth empirical misunderstanding is found in the statement in the plurality opinion in Vieth, that, “even if a majority party could be identified, it would be impossible to assure that it won a majority of seats unless the States’ traditional election structures were radically revised.”64 This point is linked to an important insight into a statistical feature of plurality elections discussed above, namely that plurality-based elections cannot be expected to generate proportional outcomes because of what is sometimes referred to as the “bonus effect,” and hence it is inappropriate to use proportionality in the outcomes of plurality-based elections as a test for partisan fairness or of the equality of treatment of the voters who are the supporters of the (two) parties. But, as we emphasized earlier, from the fact that we cannot expect plurality-based plans to be proportional, it does not follow that we cannot apply the quite different concept of partisan bias to evaluate plans that use plurality based elections (see further discussion of crafting of judicially manageable standards in Section 4 below). In fact, since partisan symmetry does not imply proportionality, the problem identified is not a problem.
A fifth misunderstanding is found in the plurality opinion in Vieth that begins with the fundamentally sound insight into real world politics that “a person’s politics is rarely as readily discernible--and never as permanently discernible-- as a person’s race. Political affiliation is not an immutable characteristic, but may shift from one election to the next; and even within a given election, not all voters follow the party line. We dare say (and hope) that the political party which puts forward an utterly incompetent candidate will lose even in its registration stronghold.”65 Considerable literature in political science supports this claim of the four justices in the Vieth plurality that voter choices may vary from election to another, and that there are idiosyncratic reasons why one candidate of a party may do well and another do badly. But they go too far when they also claim that these facts make “it impossible to assess the effects of partisan gerrymandering, to fashion a standard for evaluating a violation, and finally to craft a remedy.”66
That claim does not follow from the empirical regularity noted, four sub-points about which are critical.
A. As noted earlier, the empirical findings necessary for experts to demonstrate partisan gerrymandering are directed at characteristics of the electoral system, and thus based on “if, then” scenarios, not on predictions about vote shares.
B. The second point has to do with drawing the correct implications of the fact that partisan propensities are not fixed, immutable features of human beings. Far from this being a problem for using the methodology used to calculate partisan bias, it is precisely this mutability that allows us to make use of the standard social science methodology necessary for measuring partisan bias. 67
C. Regarding the mutability of partisanship, the partisan bias methodology we propose is makes only a very weak assumption, namely that different units of geography can be approximately rank ordered in terms of their average partisan propensities.
D. Finally, because measuring deviations from symmetry requires understanding hypothetical scenarios about the consequences of changes in partisan vote shares, in order for partisan bias calculations based on such scenarios to be meaningful, these changes in vote shares must, at least in principle, be feasible ones.68 For example, consider a state where the Democrats consistently receive 80% of the average district vote in election after election, decade after decade, and where we therefore have no experience of the Republicans ever coming close even remotely to a majority, much less 80%. In this situation, even though the concept of partisan symmetry still applies in theory, ascertaining what would happen to the seat division in symmetric situations where each party in turn received some fixed percentage of the vote would be so far from the historical experience that no empirical method could be reliably used to ascertain the degree of partisan bias.69 Thus, the methodology we propose is intended only for jurisdictions where the politics is competitive enough that it is empirically feasible to develop reliable expectations what each party would receive in seats if it won a given sized majority of the votes.70 Because we are not proposing to apply this methodology in every situation, but only in potentially competitive jurisdictions, where the consequences of gerrymandering might be especially onerous in thwarting the will of the majority, the burden on courts of seeking to apply our methodology is correspondingly limited.71 Which settings are appropriate (in a statistical sense) for application of the methodology used to establish partisan fairness is appropriately a topic for expert witness testimony, but in most jurisdictions which go to litigation this issue is likely to prove completely unproblematic.
A sixth and related error about the partisan bias approach has to do with claims that it requires us to consider empirically implausible as well as purely hypothetical outcomes. State Appellees in their Brief72 claim that the symmetry standard is “designed to address only the rare if not unprecedented situation of an electorate shifting near instantaneously from a majority (of 58% or more) of one party to a comparable majority of another.” Of course, this clever phrasing is a complete mischaracterization of how deviations from partisan symmetry are established. The symmetry definition does not, in any way, envision instantaneous shifts of huge magnitude. Rather, it considers possible outcomes in the range where elections are reasonably competitive (such as from 40% to 60% for a party). Moreover, as emphasized earlier, we only propose to apply the methodology to jurisdictions where it is factually reasonable to assume that elections can be competitive somewhere within this range.
A seventh and related mistake was made where they claim that ”[b]y definition, [the symmetry standard] does not assess the electoral results that will actually be produced by a plan, but makes its validity turn on theoretical results that might occur if political circumstances significantly change.''73 As we have emphasized, there is no way to determine the “electoral results that will actually be produced by a plan” without knowing how voters actually voted. Even when there has been an election under a plan, court determination about that redistricting plan still involves case facts that turn on hypotheticals about how the plan will operate in future elections.
The key assumption behind using hypotheticals in calculating partisan bias is simply that it is possible to approximately rank order units of geography in terms of their partisan propensities.74 But, that is exactly what gerrymanderers must do when they decide how to construct their gerrymanders. When we use the methodology described above to calculate partisan bias we are merely positing that it is realistically possible to construct a partisan gerrymander by putting together different units of geography.75
An eighth error is found in the State Appellees claim76 that the symmetry standard does not “account for the heterogeneous distribution of population and political preference.” This assertion is false. The symmetry definition of fairness applies directly and straightforwardly no matter how heterogeneous are the voter populations or their partisan preferences. Moreover, the methods that have been developed to measure partisan bias explicitly take into account heterogeneity in district votes, and cause no difficulties for either conceptualization or measurement.
A ninth point of confusion is found in the complaint of the State Appellees77 that “no standard exists in the [social science] literature as to how much partisan bias is unconstitutional.” While we have no quarrel with that assertion, this brief is obviously confusing issues that are properly in the domain of the courts with those that fall within the expertise of political scientists and other social scientists. Consensus among political scientists about issues in constitutional law is not of legal relevance;78 consensus among political scientists about issues of conceptualization and measurement of partisan bias would, on the other hand, be highly relevant were courts to adopt a test that makes use of information about levels of partisan bias, because the fact of such a consensus informs the Court about the likelihood that reliable empirical evidence can be introduced into the record by competent experts about the magnitude of partisan gerrymandering effects.
Similarly, the fact that there is no consensus among political scientists as to what level of partisan bias is unconstitutional does not, as the State Appellees claim79 “merely restate the fundamental quandaries that left the Court in Vieth searching for a substantive measure of fairness.” Rather, as we show in the next section, if partisan bias were to be adopted as a key component of a legal test for unconstitutional partisan gerrymandering, the Supreme Court could draw on its jurisprudence in other voting rights areas to establish appropriate constitutional thresholds and to evaluate empirical evidence for unconstitutional effects in a fashion that is readily judicially manageable. Having an unambiguous, generally accepted, and easy to apply definition of partisan fairness in redistricting will make devising a legal standard for unconstitutional partisan gerrymandering much more straightforward.
A final and related source of confusion about the partisan bias methodology is the claim made by the State Appellees in their Brief in Jackson v. Perry, at p. 47 that “adopting an overall goal of ‘symmetry’ would require constructing maps that take no account of other, more traditionally meaningful redistricting values.” That claim is flatly wrong. As we discuss in Section IV, by drawing on well-established approaches in other domains of voting rights case law, standard redistricting criteria can easily be incorporated into legal review of partisan gerrymandering claims, especially as they involve consideration of defenses against a finding of partisan bias that rely on the State’s claimed need to satisfy standard districting criteria, and issues of constructing judicial remedies if unconstitutional partisan gerrymandering is found. More generally, we would note that the fact that there are competing legitimate (or even constitutionally grounded) criteria which courts must balance when considering partisan gerrymandering is no different from the task confronting courts in many other areas of constitutional and statutory jurisprudence involving various aspects of the Bill of Rights or the Civil War Amendments.