We were wrong about redistricting.

In 1994, Gary King and I published an article in the American Political Science Review, “Enhancing democracy through legislative redistricting,” which began:

We demonstrate the surprising benefits of legislative redistricting (including partisan gerrymandering) for American representative democracy. In so doing, our analysis resolves two long-standing controversies in American politics. First, whereas some scholars believe that redistricting reduces electoral responsiveness by protecting incumbents, others, that the relationship is spurious, we demonstrate that both sides are wrong: redistricting increases responsiveness. Second, while some researchers believe that gerrymandering dramatically increases partisan bias and others deny this effect, we show both sides are in a sense correct. Gerrymandering biases electoral systems in favor of the party that controls the redistricting as compared to what would have happened if the other party controlled it, but any type of redistricting reduces partisan bias as compared to an electoral system without redistricting.

I think we were correct in our analysis, but it was an analysis of past data (state legislatures in the 1970s and 1980s). Since then, it’s my impression that gerrymandering—extreme partisan redistricting—has gotten much worse. I’m only speculating here because I haven’t looked at the numbers, but I’ve heard of five explanations for why gerrymandering has become more of a problem:

1. Better data and computers. The technology is now available to draw some really biased districting plans using local voting information. There are just more people who can draw these sorts of biased maps, and it’s easier to do.

2. Increased partisan polarization, part 1. Voting patterns are more stable. One of the risks of gerrymandering was always that you could be too clever by half: give your party a bunch of seats that it can win 60-40 and you’re at risk of being buried under a landslide. But elections are more predictable now, so it’s easier for gerrymanderers to assess these risks.

3. Increased partisan polarization, part 2. There’s more of an anything goes attitude and it’s easier for a legislative majority to keep the cohesion to steamroller the minority on all sorts of things, districting included. Gerrymandered districts then make a party with a legislative majority less likely to lose power.

4. Decline in incumbency advantage. Incumbency advantage isn’t as high as it was back in the 1980s, so, from the standpoint of partisan control, incumbency protection isn’t such a big deal. The traditional retirement of incumbents after redistricting does not shake up the system so much—and, as we explained in our article, that was a bit reason why redistricting increased electoral responsiveness in the past.

5. Reduced judicial scrutiny. I’m not sure about this, but I remember back when we were doing the research that led to our 1994 article, we had the impression that the most extreme gerrymanders didn’t happen because legislators were afraid that maps that were too obviously biased would be thrown out by the courts. That doesn’t seem to happen so much anymore.

It would be a useful research project for someone to look into this. From my part, I feel bad about writing that paper, if it made anyone complacent about the ways in which gerrymandering can threaten democracy.

Even back in the 1990s, we always said that nonpartisan or bipartisan redistricting was preferable to partisan redistricting: our argument was in favor of redrawing the district lines, and we just said that the partisan aspect of it wasn’t such a big deal because extreme partisan plans would be constrained by the courts. But in retrospect, yeah, back when gerrymandering wasn’t such a problem, that would’ve been the time to do something about it.

So the title of this post is a slight exaggeration. I don’t think any of the claims in that 1994 paper were wrong. But I think we were wrong to not at least try to anticipate future changes in an era of increasing partisan polarization.

20 thoughts on “We were wrong about redistricting.

  1. “We demonstrate the surprising benefits of legislative redistricting (including partisan gerrymandering) for American representative democracy. ”

    IMO this claim likely is far too strong no matter your data. For “democracy”? Really? That’s a pretty…er…substantial claim. Without reading your paper, sounds like you might be showing that redistricting improves the representativeness of individual elections which is probably a good thing, but what it means for “democracy” as a whole is *far* beyond the data of individual elections! Did Americans become happier or more wealthy because of redistricting? Were there fewer wars, less famine? :]

    How do we know whether gerrymandering is a net benefit or not for “democracy” as a whole? I can’t imagine how you could answer this question with election data. Without the electoral college we might never have had a democracy, so while it doesn’t accurately represent the populations’ vote, isn’t it a net benefit to democracy? There might also be benefits to gerrymandering: although it distorts representation, it also may create more stability and continuity in government that’s a net benefit to the population.

    So, yes, I think you were wrong. It would have been appropriate to claim that redistricting improves the accuracy of representation in elections. But it wasn’t appropriate to claim that it was a “benefit to democracy”.

    • Jim:

      By talking about democracy we were referring to the concern that redistricting would create a class of safe seats so that no incumbents would be threatened. We were not just talking about “accuracy of representation” or closeness to proportionality; we were talking more generally about our finding that redistricting created more competitive elections. I do think competitive elections are a part of democracy in the U.S., but I agree that competitive elections are not absolutely necessary. I don’t know anything about Japan, but I have the impression that the same party wins all the elections there but it’s still democratic.

      In any case, we said that redistricting was “enhancing” democracy and that it provided “benefits” to American representative democracy. I think that’s fair. Creating more competitive legislative seats, while introducing only a small amount of partisan bias, seems like an enhancement or benefit to democracy.

      But, as noted in the above post, that was then, this is now. I’d be very interested in seeing a research project that would explore the prevalence of redistrictings that distort the vote and reduce democracy. We hear about this in the news but I don’t have a sense of the systematic impacts.

  2. Things do change. I don’t believe #1 is really one of them though: the Karl Rove-types have made this a science long ago. But, the biggest change: the Supreme Court has said this is a state-issue and there is no federal constitutional basis to over-turn legislative districting (only a handful of districts have been re-written in Wisconsin, Maryland and North Carolina).

  3. Are you going to write up a new paper? You’ve noted how academics treat anything published as presumably true and disregard the “post-publication peer review” of speculation on blogs.

  4. ” give your party a bunch of seats that it can win 60-40 and you’re at risk of being buried under a landslide.” Isn’t the idea to give the other party a few very safe seats?

  5. I have a question about #1. Better algorythms and data allow optimal gerrymandering. However, won’t these hypergerrymandered distrists breack quicker. I have this intuition that by the end of the 10 years the hypergerrymandered districts won’t hold up as well as the old fashioned gerrymandering because they are more artificial. Is there any research on how long these gerrymandered district hold up?

    • > However, won’t these hypergerrymandered distrists breack quicker. I have this intuition that by the end of the 10 years the hypergerrymandered districts won’t hold up as well

      I believe a component of contemporary “hypergerrymandering” is more accurate forecasting of demographic/political shifts within a given district and the creation of districts that are robust to those shifts. As Andrew wrote, this also seems to be less of a problem due to more stable voting patterns over the last however many years. With voting patterns being more stable, instances where a political shift is large enough to break a gerrymander by the time the next districts are drawn don’t seem to occur terribly often (although it’s hard to support this robustly over time since district maps are drawn so infrequently).

      By my recollection, the most notable instance of a state breaking their gerrymander due to shifts in the political landscape is Virginia, but their rapid blue-ward shift was somewhat exceptional. It’s noteworthy that Virginians supported non-partisan redistricting on a ballot measure that originated during a pro-GOP gerrymander, and now that Democrats are firmly in control of the state legislature, some are lamenting the fact that they are unable to enshrine a pro-Democrat gerrymander.

  6. A number of your speculations seem not at all clear to me – I know this is more your area of expertise than mine, and that you have carefully studied some of these areas and I have not. But a few things strike me as questionable:

    2. Elections are more predictable now? I have two issues with this. Some are, some are not (e.g. Trump 2016). Also, if they are more predictable, isn’t that at least partially due to gerrymandering itself? The argument seems a bit circular to me.

    3. Anything goes attitude – I totally agree with that one – but it borders on irresponsibility (actually goes over the border). Again, isn’t this partially a result of gerrymandering itself?

    4. Decline in incumbency advantage? Really? Not from what I perceive – but again, I know you have studied these things and I have not. But, if the raw data reveals such a decline, has it accounted for a number of reasons why people resign rather than getting voted out of office?

  7. Related to number 4 (and number 3 to an extent): it seems to me that if incumbency isn’t worth much and you have nothing to fear from the other party then the logical outcome of this is an increase in pandering to the largest voting block within your party, because in such a situation the thing an incumbent has to worry about most is a primary challenger. I don’t know if this warrants another point, but it would seem to explain some of the tendency toward more extreme policies of incumbent politicians who might previously have been described as “moderate.”

  8. From what I understand (which is little in this area), gerrymandering, by definition, is the attempt to establish an unfair political advantage. Assuming that’s the case, wouldn’t it be safe to assume that gerrymandering could never be a good thing, whether for democracy or for society in general? If I’ve misunderstood something here, please let me know.

    • David:

      I agree with you that all gerrymandering is bad. Some would argue that a small amount of gerrymandering is fine as it represents the spoils of the system, but I wouldn’t say so; I agree with your position. One of the findings of our 1994 paper was that the partisan bias introduced by partisan redistricting in the 1970s and 1980s was pretty small. So, yes, gerrymandering was a problem back then, but a small problem. Since then, it appears that gerrymandering has become a much bigger problem—how much bigger, I’m not sure, as noted above I think this would be a good research project.

      • Andrew

        Ahh, cool cool cool. What is it about this topic that you find interesting? I ask because it’s not immediately obvious to me what about this theme is compelling enough to inspire research into it.

        • David:

          Back in 1990 our motivation was to understand the decline in electoral responsiveness from the 1950s through 1980s, which was considered a big concern in American politics—more safe seats, fewer close districts, less motivation for legislators to appeal to the median voter . . . indeed, a big contributor, I think, to the modern era of hyper-polarization. Gerrymandering was one explanation that was sometimes given for this decline in responsiveness, but in our research we found that this explanation didn’t hold up, that redistricting actually increased responsiveness during the period we studied (1970s-1980s).

          Nowadays our interest in the topic comes because gerrymandering seems to be a big problem in some states, with impact on national politics as well. Again, to the extent that legislatures are insulated from the will of the voters, that’s a concern.

        • Andrew:

          Cool! Now that sounds really interesting! When I read “hyper-polarization” it reminded me of an interesting paper I just read by Koc-Michalska et al. (2020) Public Beliefs about Falsehoods in News – have you read it? If you have (or if you haven’t but you decide to check it out) I’d actually really love to know what you think about it; it seemed like a really good study from its presentation, but I can’t tell if the methodology or the conclusions drawn were good.

          This got me thinking, can we see the principle of gerrymandering playing itself out anywhere else in society? Like, is the media being gerrymandered? Or something.

          By the way I’m new to the site and was wondering if there’s a way for me to get a notification when someone replies to my messages?

Leave a Reply to Andrew Cancel reply

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