Princeton Election Consortium

A first draft of electoral history. Since 2004

Meta-Margins for control: House D+3.5% Senate R+6.5% Find key elections near you!

One last gerrymandering case…Abbott v. Perez in Texas

April 23rd, 2018, 11:16am by Sam Wang


This week, the Supreme Court hears one last case on gerrymandering for the term. This one’s a racial-gerrymandering case, Abbott v. Perez. The history of the case is long and tortured – see Ian Millhiser’s summary. Also, here’s great coverage from Alexa Ura at the Texas Tribune.

This case is in a separate category than the partisan-gerrymandering cases (Whitford, Benisek, and Rucho), where there is an opportunity to create new guardrails. Instead, Abbott seems more a case study in how a dispute can be slow-walked – in this case since 2011. I agree with Millhiser that the slow-walking is upsetting. But it could arise from caution in the face of complexity (or in the partisan cases, the absence of settled doctrine). So one could justify that aspect of how gerrymandering cases are handled.

Millhiser is quite negative about the Court’s approach to gerrymandering. At the risk of sounding Pollyanna-ish, let me take a more positive view of where the Court may head in the future. Indeed, I see potential for substantial improvement, by expanding gerrymandering from race into the domain of party.

Expansion into the partisan domain is timely because polarization between the parties has escalated so much starting in the mid-1990s. The erosion of minority voting rights is an area of obvious concern, but even there, the Court’s record in redistricting is not all bad – see the Cooper v. Harris decision in North Carolina.

How can racial and partisan gerrymandering be combined? From the standpoint of best forensic practices, a suitable distinction might be between how to address cases involving (a) a very outnumbered minority (e.g. North Carolina African-Americans, or Maryland Republicans) or (b) two groups at near parity (e.g. statewide partisans in Wisconsin, Pennsylvania, or North Carolina). These require different approaches.

I wrote about the possibilities in the Harvard Law Review Blog and in our working paper. Our main focus in the piece was separating the two concepts of unequal opportunity and unequal outcomes. Under either concept, it is important to understand that the concept (whether opportunity or outcome) should be examined in multiple ways. Racial/partisan minorities require examination of single districts, whereas when the two parties are at near-parity, a statewide approach is necessary. If the Supreme Court adopts this logic, it will define a new right: the right of a whole party to be represented. That would be huge!

In the past, I have written about detailed math. But I think the math is beside the point. Indeed, a big intellectual step for the court will be to see that the many formulas all fit under a few principles. To read more, see our SSRN working paper.

Tags: Redistricting

7 Comments so far ↓

  • Pechmerle

    Here is ScotusBlog’s summary of oral argument today in this case.
    http://www.scotusblog.com/2018/04/argument-analysis-redistricting-again-divides-justices/#more-269441
    Lots of questions about the procedural posture of the case. My guess is that the Court will send this back to the lower court on procedural grounds, rather than making any law on the merits. The lower court was hostile to the Texas Legislature’s shenanigans, so a remand might not be a bad thing. But probably no new law from the Supremes.

  • LondonYoung

    “How can racial and partisan gerrymandering be combined?”
    It would certainly be good if the court tried this approach, but I have a legal question that maybe Sam or Pechmerle could comment on.

    Racial discrimination in voting was fought by the Voting Rights Act of 1965 – congressional legislation – rather than by court decision as in, say, Brown v. Board. Later on, the Supreme Court said that much of the VRA was really just a restatement of the 15th Amendment.

    However, one might say that the 15th Amendment (and, by the way, the 19th Amendment too) is really just a restatement of the 14th Amendment’s equal protection clause. If this were to be the case then a lot of gerrymandering problems would be (legally) blown away since every group would enjoy the same collective protections.

    However, while I have seen the court say “this legislation means nothing because the constitution covers it”, can they say “this amendment doesn’t matter”? I vaguely recall some prohibition against interpreting law as meaning nothing …

    • LondonYoung

      By the way, I just realized this: One central point of the VRA was that if a minority group tends to vote together thou shalt not crack them to prevent them from electing a candidate of their choice.
      This is very clearly what was done to republicans in Maryland.
      Wisconsin is more complicated since packing comes in as well.
      While gerrymandering nerds may group cracking and packing together, maybe the court won’t …
      Just a thought

    • Sam Wang

      I agree with your point regarding the VRA and Maryland Republicans. See my “tea leaves” post today. I believe this to be the easiest thing that the Supreme Court can do, and it would come in a decision on Benisek. I think they may draw the connection.

      I also agree that Wisconsin is more complicated. There is uncertainty about what they will do – again see the “tea leaves” post.

      Curious to know if you think my HLRB post and SSRN working paper address these issues, or if not, what needs to be clarified.

    • LondonYoung

      I think the law blog and SSRN (and SLR too) could use more discussion before the math is rolled out to address Wisconsin. Two points come to mind.

      1) Judges hate math because they can’t understand xkcd comics:
      https://xkcd.com/1132/
      So I’d make this appeal: I don’t need any math to know that a cracked voter is suffering because I can see they are represented by someone they don’t want. But is a packed voter suffering? I think this is why Kennedy sees this as a first amendment issue, freedom of association. A successful gerrymander will increase the number of people not represented by the party of their choice. They are harmed because of their lack of ability to associate with the packed voters. Thus there is a collective right involved here. So – step one – get the court to agree to this collective right – they haven’t yet. Then, whenever we discuss anything collective argue that we have to use some math. Let’s remember that voting is based on using math to calculate the winner. Not everybody gets to be represented by the party of their choice (in the US system) so the suffering is a matter of degree and we must use math to measure matters of degree – sorry judges.

      2) Intellectuals of both parties dislike extreme partisan gerrymandering. However republicans don’t care about partisan symmetry while democrats love it. This has nothing to do with math or rights of association – it is just personal bias because everyone knows the democrats are naturally packed into cities and that without any partisan intent randomly drawn lines will hurt them. Every time someone says flipping 51/49 and 49/51 should reverse the results you know that person is a democrat. This makes the topic radioactive. The way to deal with it is by not starting with partisan symmetry, but rather by starting with the work that has be done to estimate the democrat disadvantage and then factoring it out before rolling out more math.

      The dems can go after partisan symmetry in the future – as Reagan said “I’m always happy to take 80% of what I want and come back for the other 20% later”.

      All these topics *are* covered in your works, but your question was about how to clarify the arguments …

    • Sam Wang

      I agree that we might need to make a clearer verbal argument. However, your argument is basically Brunell’s argument. He thinks gerrymandering is good because in a packed district, more voters get representation that matches their preference.

      Imagine a state with four districts, where the electorate is evenly divided between Democrats and Republicans. Consider the following patterns of outcome:

      Pattern 1:
      80% D, 20% R
      40% D, 60% R
      40% D, 60% R
      40% D, 60% R
      In this example, Brunell might tell a court that the first district is really good because it’s 80% Democratic. However, somehow he never does the calculation: what fraction of statewide D’s got someone to their liking, and was it comparable to the corresponding fraction for R’s? I consider this to be lacking in intellectual honesty.

      If one calculates that, one gets:
      Fraction of D’s who get a D = 80/(80+40+40+40) = 0.4
      Fraction of R’s who get an R = (60+60+60)/(60+60+60+20) = 0.9
      Overall fraction = 0.65

      Pattern 2:
      55% D, 45% R
      55% D, 45% R
      45% D, 55% R
      45% D, 55% R
      Fraction of D’s who get a D = (55+55)/(55+55+45+45) = 0.55
      Fraction of R’s who get an R = (55+55)/(55+55+45+45) = 0.55
      Overall fraction = 0.55. This is less than Pattern 1!

      Which is fairer? I’d say Pattern 2. Why? Because….it’s more symmetric (0.55=0.55), and represents the overall wishes of the voters more fairly (2 D’s, 2 R’s). These match our opportunity and outcome tests.

    • LondonYoung

      You are taking me wrong because you are using too much math!
      I do not automatically think that pattern 1 is better than pattern 2 because 0.65 > 0.55. Let me restate at the risk of sounding repetitive – we are just looking for clarity here!

      A voter has a 14th/15th amendment individual right to a “fair chance” to elect someone of their choice. When a legislature deliberately cracks a voter from other like-minded voters to put that voter in a district without a rep of their choice the legislature is acting illegally because they deny this 14th amendment right.

      J. Kennedy is ripe to be persuaded that under the first amendment voters also have a right to associate with each other in order to advance their common interests. When voter A is intentionally denied the right to associate with like-minded packed voters and ends up with a rep not of their choice, voter A has been denied a 1st amendment right.

      In pattern 1 35% of the voters might consider bringing suit, and in pattern 2 45% of voters might considering bringing suit. Just because pattern 2 results in more people with a possible claim of harm is really irrelevant to the judges in terms of whether the constitution has been violated. Even a single voter intentionally denied a constitutional right is good enough to reject the map on 1st amendment grounds. The math is a tool to measure a suspect map and – as per your work – you need *many* tools in the toolbox to verify malice.

      I have no opinion at all on whether pattern 1 or pattern 2 is better without more information about how they came about. If Pattern 2 is created by cracking city democrats such that suburban democrats can then build bridges to the city rather than subways within the city then it is illegal. If Pattern 1 is created by building a salamander shaped district of voters selected only because of their party, then that would be illegal.

      Pattern 2 is more proportional, but the court has already ruled that this does not matter. I don’t think the court demands a tanh function any more than a straight line – but maybe that is the case you really do want to bring? Deviation from a tanh function is suspect, but maybe it should just be illegal? But I don’t see what the constitutional argument for that is??

Leave a Reply to Pechmerle (Cancel)