Teague: Math and Redistricting: Diagnosing a problem

The Op-Ed Page

An image from the presentation Lynn links to in the last graf.

An image from the presentation Lynn links to in the last paragraph.

By Lynn Teague
Guest Columnist

Brad drew my attention to an article in The Washington Post about mathematics and
redistricting. This brought to mind some important math about South Carolina’s current
legislative districts. The majority of South Carolina’s legislative districts are non-competitive in
the general election. The winning candidate is selected in the primary in June. This makes
November elections meaningless in many cases and encourages polarization, since highly
engaged and often extreme voters are especially likely to participate in primaries. It also seems
odd, since we know that in S.C. statewide races the majority party draws about 55% of the vote,
while they now control a super-majority in the Senate and House. The most common
explanation for this disparity in proportions is partisan gerrymandering.

However, you can’t fix a problem if you don’t know what the problem really is, and guessing
isn’t good enough. Even well-designed districts can look odd, and gerrymandered districts can
look okay. An eyeball test isn’t enough. So, League of Women Voters of South Carolina board
member Matthew Saltzman supervised a Clemson grad student thesis to evaluate whether our
districts meet mathematical tests of partisan gerrymandering. Anna Marie Vagnozzi used a test
originally employed in the 2017 case League of Women Voters v. Commonwealth of Pennsylvania. Vagnozzi generated millions of maps and found that the current S.C. maps do not fall at the extremes of the resulting distribution and do not seem to have been pushed to extremes by partisan bias. Instead, they fall very much where the presumably fair post-litigation Pennsylvania maps do.

So, why are South Carolina’s maps so non-competitive? Some of this arises from demography.
South Carolina retains a significant level of racial polarization in voting, so the tendency of black
voting-age populations to be concentrated in some areas, especially urban centers, is a major factor. Votes are wasted when a group of voters are clustered together in such high numbers that their district would have been won by the same party without many of them. White populations are more evenly distributed throughout the state and this provides an automatic electoral advantage, giving them greater voting strength with fewer votes wasted.

The other big factor is incumbent protection. Incumbents of both parties have engineered their
districts to be “safe.” They have amplified the differences caused by demography to create even
more extreme differences by carefully choosing their boundaries to include neighborhoods favorable to them and exclude others. This bipartisan process, repeated through successive redistricting cycles, has led to some excessively predictable districts. (Bipartisanship is not always the Holy Grail of good politics.)

A very recent presentation on redistricting in South Carolina is posted on-line at the League website. It includes maps of our noncompetitive districts as well as a short summary of Vagnozzi’s research and discussion of where we are in the redistricting process.

Lynn Teague is a retired archaeologist who works hard every day in public service. She is the legislative lobbyist for the South Carolina League of Women Voters.

30 thoughts on “Teague: Math and Redistricting: Diagnosing a problem

  1. bud

    Careful Lynn, you used that dirty word “mathematics” in front of Brad. Of course districts should be drawn by non-partisan means. How can anyone really argue that. And that process will invariably require mathematics.

    Reply
    1. Brad Warthen Post author

      Well, bud is joking (I hope), but to make sure others are not confused, I’ll repeat part of what I said yesterday about what the Brits call “maths:”

      The thing is, I understand it well enough to see its limitations, and I respect it enough to argue that math shouldn’t be disrespected or abused by trying to apply it to things to which it is ill-suited.

      Which is what I’m trying to say here so often. But I get misunderstood.…

      Reply
      1. bud

        I don’t misunderstand you. I just respectfully disagree. Math, or more broadly science, should be used to address practically every problem. Your road example is a great illustration of that. Traffic patterns, immigration, economic measures, census projections, environmental concerns and so forth all have their basis in mathematic analysis. Those factors, not political chicanery, are what should determine where and when to build or widen roads.

        Reply
      2. Bob Amundson

        Come on Brad! You are a wordsmith; please recognize your bias against math. I hope I don’t have to explain more. 😉

        Writing, music, statistical probability – probability is a tough concept FOR MANY to understand (bud knows). 2 out of 3 is 66.666666666 and so on %.

        Reply
  2. Lynn Teague

    I’m here for the informed use of math and science point of view, and for knowing the difference between scientific assessments and those based in values. (Not saying either of you aren’t, Bud and Brad, just want it explicit.) For example, I was sent a link this morning about a new interstate segment that will go through part of South Mountain near Phoenix, damaging places sacred to the O’odham. Math and science can tell us how to build it, but values are what decides whether it will happen. It turns out our society values roads and population growth leading to more money for some folks more than it values the spiritual beliefs of the O’odham. No surprise there. Or, another example, science doesn’t tell is where in the life cycle of Homo sapiens we should draw a line and say “this is what we call a person and assign the associated values,.” Some assign moral values to zygotes (usually on the religious teaching that it possesses a soul at that point) while others assign moral values to independently viable sentient stages of the life cycle. Science doesn’t dictate either choice. But back to Redistricting.

    There is no pure math in Redistricting either. Math is the servant of values-based decisions. We have the districts we do in part because many legislators value their own easy reelection and their continued power above just about anything else. The League values an undistorted representation of the will of the voters, including fair representation of minority voters. Before you do any Redistricting math, you must define your criteria based on what matters to you. It is the most crucial decision node in Redistricting. The math comes later.

    And by the way, it is autocorrect that thinks Redistricting should be capitalized, not me.

    Reply
    1. Brad Warthen Post author

      Lynn and I are about as much in agreement as two people can get — on this subject, anyway.

      This all started when I — not bud, not someone else — sent her the WashPost piece about the mathematician helping with the Virginia process. Lynn’s response began, “Thanks for drawing my attention to this, Brad. Of course mathematicians can help.” And that’s exactly where I am. Math can help, but of course it isn’t everything.

      And of course, one must understand “the difference between scientific assessments and those based in values.”

      I may have confused the issue earlier when I used the term “politics” to describe what is involved in deciding whether to build a road to Florence, in addition to math.

      As a retired highway department guy, I think bud thinks of politics as the weird, perverse, meddling that subverts the rational decisions made by people who work at the highway department.

      Which, of course, is not what I mean by “politics” at all. It’s the process of public decision-making. Math helps you determine whether you have the resources necessary to build a road, and helps you build it straight. Politics is the necessary process of deciding whether to build it. It’s how you decide, for instance, whether you’re concerned about “damaging places sacred to the O’odham.”

      Lynn is calling it values. I call it politics…

      Reply
      1. James Edward Cross

        I would be more inclined to consider politics as the *process* by which those values are put in play. People with conflicting values come together and through politics either one set of values or another triumphs or a compromise is reached.

        Reply
      2. bud

        No, values and politics are not the same thing. Values exist outside the realm of politics. Politics is the means by which values can be administered. But even when it comes to values math can and should play a role. Everything can be quantified. Sacred land is certainly valuable, but not infinitely. And it’s value can be quantified. Perhaps that road through sacred land would save lives by making the trip safer. At the DOT (not the highway department) we decided cutting trees made road travel safer. That offended conversationists. So two values collided. If a ship is sinking and the only way to save it is by flooding the engine room then math comes into play. That’s making a tough decision that values life but not infinitely so.

        Pure politics ics on the other hand ignores values. It is strictly about jamming a particular project through in the absence of any mathematical determination. What Brad is saying is that math has zero role in road building until the decision to build the road is already made. That’s nonsense.

        Reply
        1. Brad Warthen Post author

          No, actually I’m saying something very, very different from that, but I’m just going to stop saying it for now and move on to other matters. Because saying it isn’t helping…

          Reply
      3. Bob Amundson

        I disagree Brad. Math is a tool that has no value judgement built in; clearly imperfect. But combining Math (Scientific Method is a more accurate term) with values is how I work; I am extremely value driven, but those values are informed by math/science.

        I am just having a hard time of saying politics and values in the same sentence. I am glad a value driven scientist like Lynn is influencing the outcome. Go get ’em Lynn. Let me know if I can do anything.

        Reply
  3. bud

    As for redistricting math and values are not mutually exclusive. If I was going to develop a values based districting map I would run away from politics. Rather it would rely mostly on geography which is essentially about math. Simply draw the maps by treating all voters as equals. I would give no consideration to subjective considerations.

    Reply
    1. Lynn Teague

      Which geography? County and municipal lines? River drainages? Areas where people go to the same restaurants (don’t laugh, it has been suggested)? If you try to get around decisions by simply imposing a grid, you will be far from meeting the Constitutional standard of one person one vote because they will be wildly dissimilar in population numbers. The League is developing maps that prioritize not splitting counties or cities, but that still is a decision, and it leads to the need for more decisions.

      Reply
  4. Mark Stewart

    The map is not viewer friendly; it does not convey the necessary information to achieve buy-in. Sales is not always a dirty word.

    Reply
    1. Lynn Teague

      The map is one in a series, and the same data are presented much more dramatically in the map that follows in the series.

      Reply
      1. Mark Stewart

        Thanks, Lynn. I did review the presentation. First of all, it is good to know that SC is only moderately gerrymandered – and that this is more about the normal push and pull of historic incumbency than some ongoing Atwaterish conspiracy to throw the vote.

        That said, the graphics do not do the research justice. It might just be me being dense, but the importance of the data, both visually and by value, is missing the punch of significance which the study deserves. When I was starting out in sales & financial analysis, my mentor came into my office one day with a few Edward Tufte books and said “do this.” It was a mind-bending introduction to what one of the books was titled, the visual display of quantitative information.

        That said, I believe the congressional map highlights an important issue for the Columbia MSA relative to other South Carolina regional hubs. Columbia is uniquely hobbled by shared representation by two disparate political office holders. If SC isn’t extremely gerrymandered, then it would seem that this dual, off-setting representation issue is the most significant negative of the current district boundaries. No matter which way the political winds blow, Columbia is guaranteed to be marginalized legislatively.

        Reply
        1. Lynn Teague

          Tufte is great, although I’ll admit that I haven’t looked at his books since retiring from archaeology. The redistricting maps are generated, with some minimal changes for readability and the addition of logos, from Maptitude, which is the software that we (and the General Assembly) use. On redistricting we are doing with a handful of volunteers what the houses of the General Assembly are doing with millions of dollars and a lot of tech and comms staff, and doing it under extreme time pressure. They expect to convene for a final vote in mid-October, so we must get our maps (in the form of data files, no graphic issues) submitted very soon if they are to have any chance of impact. I’m afraid I can’t promise any major graphic improvement in our public presentations in the little time we have left.

          As to the Columbia area — we are working on maps that we hope will fix some of the crazy here, not just in the Congressional but in other maps as well. I can’t share them because they are still in draft and under review by our own independent review panel. They will become public soon.

          Reply
          1. Mark Stewart

            Thanks for the insight. I won’t add to your burden, and agree that time is short to make an impact. Carry on and fair winds!

            Reply
  5. Ken

    Another example of the competition between values vs. mathematics?

    What from one perspective is seen as a problem: “voter clustering” (or “packing,” in redistricting parlance) leading to “wasted” votes is seen from another perspective as the answer to “voter dilution” (or “cracking”). Because, if memory serves me correctly, it was the dilution of the Black vote in the South that came in for constitutional criticism in the 1970s and after.

    So what’s the proper balance? It’s not clear in the presentation how the League proposes squaring the goal of ensuring “effective representation of racial … minorities” while also pursuing the other goals (esp. i and ii). The citations in the concluding slides speak to the issue but don’t explicitly state how the League seeks to address it in SC.

    Reply
  6. Lynn Teague

    Both of our primary technical people are well versed in this issue, on which there is abundant case law making it anything but simple. The League member actually doing much of the drawing has three decades of experience as a redistricting consultant and expert witness in litigation on the issue. We didn’t attempt to address this is in the presentation because it is simply too complex for a short presentation. Courts have determined (wisely) that setting a single percentage goal for minimum BVAP is not acceptable. Also, as the presentation notes, other smaller minority groups should be considered in order to avoid a packed district. However, while Hispanics are the fast growing young population group in SC, they vote in lower percentages than Black voters. Presumably part of this is because some fraction are legitimate members of the Census population but are not qualified electors. That must be taken into account. And, sometimes the percentages are elevated above what is necessary for a minority to select a candidate of their choice simply because that is who lives there, a heavily minority population. Also, we need to consider local voter polarization figures. These figures are very complex and time-consuming to compute. The League is using racial polarization figures provided to us by our friends in the national ACLU and NAACP. These figures can be significant; even within South Carolina, there are areas where white voters are more willing to vote for a Black, Asian, or Hispanic candidate than in other areas.

    What we are doing with all this will be revealed on September 29, when we publicly present our own redistricting maps for SC in a Zoom presentation, aided by our technical experts. You can register for that event for free at https://www.eventbrite.com/e/what-would-fair-maps-look-like-tickets-168718235785.

    Reply
    1. Ken

      Thanks for your response. I’ll be interested to see what the League proposes. And how, if any, that differs from what the legislature decides.

      Reply

Leave a Reply

Your email address will not be published. Required fields are marked *