Navigating Debates about Redistricting

People feel passionately about redistricting.  They don’t like how it’s done, or how it’s disadvantaged their party, or both.  So when political scientists come along to say “redistricting might matter less than you think”—for the outcomes of the 2012 House elections, for party polarization, for declining electoral competitiveness—people get cranky.  For example:

 Suck on it Monkey Cage and prove me wrong with maps.

So there’s clearly room for more thinking and discussion about the effects of redistricting.  Here are 4 things I think are important to discuss or at least mention.

1) To say that something matters less than is often suggested is not to say that it doesn’t matter at all.  This should be obvious, but apparently it’s not sometimes.  So I am making it clear, in hopes that people will stop arguing with a straw man.

2) A lot depends on the counterfactual. As we’ve discussed before, making a causal claim—“the 2011 redistricting caused the Republicans to retain a majority after the 2012 election they otherwise would have lost”—entails making a claim about what the “otherwise” is.  The challenge  becomes not only agreeing upon some relevant and realistic counterfactuals, but agreeing upon what would have resulted under those counterfactual scenarios.  Which is not easy.

In Eric’s original post, he asked a specific question: what was the effect of the 2010 redistricting?  So his counterfactual was to compare the 2012 boundaries to the previous district boundaries.  People fussed about this—see the comments thread—because who’s to say that the previous boundaries were fair?  But Eric’s question was about the most recent redistricting, something many people were discussing, so his counterfactual was relevant.  His analysis suggests that the 2011 redistricting may have hurt Democrats a bit, but not that much.  It explains less about why the Democratic gains in the House were small in 2012 than many commentators seem to think.

But we can also consider the effects of redistricting against other baselines.  Nick Goedert’s post does this, by comparing the 2012 outcome to the historical relationship between votes and seats from 1972-2010.  He finds that the 2011 redistricting played a larger role.

And there could be other baselines or counterfactuals. Others in that comments thread seemed to want to go back before other GOP gerrymanders, perhaps before 2001. Or what if there were no Republican tide in 2010 and Democrats ended up controlling more state legislatures? Or what about other possible maps?  The Daily Kos diarist who told us to “suck it” drew up a set of maps that he argued were fair and non-partisan and would have produced a 22-seat majority for the Democrats.  Compared to that, the current set of districts undoubtedly favors Republicans.

At the same time, Jowei Chen and Jonathan Rodden drew thousands of maps using similar criteria for fairness and still came up with a pro-GOP bias.  Across a very wide range of counterfactual scenarios, the geographic concentration of Democrats will produce votes-seats discrepancies without deliberate partisan gerrymandering.  Nick’s analysis also shows evidence of this as well.

My synthesis of all of this is that, yes, the state of the districts as of 2012 is tilted in the GOP’s favor relative to a bipartisan baseline, and this may be the outcome of multiple rounds of redistricting (2001 and 2011, plus the in-between redistrictings, e.g., in Texas). But this GOP tilt could not be completely eliminated via redistricting.

3) A lot depends on what standards for drawing congressional districts you value. “Oh no,”  you might be thinking. “The GOP tilt absolutely could be eliminated via redistricting. Why not spread out those geographically concentrated Democratic votes?” This is possible, as Michael McDonald noted, although perhaps not everywhere, as Jonathan Rodden noted.  You could probably do it while drawing districts that would respect the standards of compactness and contiguity, at least enough that no court would object.  But then you’re doing at least some violence to another standard for districts, which is that they should respect naturally occurring boundaries or communities of interest.

David Butler and Bruce Cain’s book, Congressional Redistricting, makes this important point: you can’t have it all.  It is very difficult to achieve equal district populations, respect compactness and contiguity, respect communities of interest, avoid diluting minority voting strength, and create perfectly proportional representation or at least minimize seats-votes discrepancies.  As Butler and Cain write:

…almost all the generally accepted principles of redistricting can come into conflict with each other.

So claims about partisan bias in the current set of districts, or claims about what would be true under various counterfactuals, also depend on (often unstated) priorities about which standards are most important and what trade-offs we should make.

4) To say that redistricting matters less than many people think is not to say that the current system of redistricting is a good one. This is another assumption people sometimes make in reaction to Monkey Cage posts like the ones I’ve linked to above.  But the assumption doesn’t follow.  There may be many very good reasons to reform redistricting—such as by putting non-partisan commissions in charge in more states—even if that reformed process doesn’t transform politics.  At a minimum, one can still make the argument on purely normative grounds, and I think such arguments matter.

We’ll continue to report on research about this topic.  My hope is that with these ideas stated, the resulting conversation will be more productive.

34 Responses to Navigating Debates about Redistricting

  1. Andrew Gelman December 12, 2012 at 1:55 pm #


  2. W.M. December 12, 2012 at 2:47 pm #

    This is all so very nice, but let’s get real here:

    Alot of this discussion involves completely talking past one another and, frankly, much of this particular post involves “addendums” onto the theory in order to make it more satisfactory.

    Take point 3 specifically: you’re saying here that it is possible for maps to not have anti-Democratic bias, but you couch this in language suggesting that this can only be the case if you ignore other types of considerations for redistricting.

    The problem here is that one that even you admit: it doesn’t matter what your considerations for redistricting are, because you’ll always end up breaking each and every one of those considerations to some degree somewhere in the map. Therefore, it is possible to stay within the letter of the law (which at this point simply means being VRA compliant and with population equity in line with jurisprudence) while still creating maps that reflect the partisan disposition of the state.

    (I’ll have more on this particular above point in a second).

    The more specific critique that many people have with the theoretical discussions here is that it becomes quite obvious very quickly that not a single one of you has actually dealt with these problems through GIS software integrated with election and demographic data. If you’d actually go through and try to do such, many think that you might change your tune to some meaningful degree.

    Anyway, what I think is the most important thing to consider is that you all seem to really be partial to this “there are many considerations that come into play during redistricting” when there really aren’t. This is a political act where the goal is to maximize your party’s chances for a majority at the ballot box irrespective of if you’d garner the majority of votes cast. You seem to rely way too much on political geography to tell us that Democrats are at an inherent disadvantage, but that’s simply false as any time on on GIS software would show you that it is actually very easy to draft maps that balance partisanship with other issues (including having a map that “looks pretty,” which is so en vogue) that reflect a state let alone actually draw one that would result in more Democrats.

    The truth of the matter is that what causes the Democratic disadvantage here is the virtual requirement that Democratic voters, which yes are naturally concentrated, be packed into few districts as a result of VRA jurisprudence. What happens, then, is that although there may be excess African Americans or excess Hispanics (and the problem is worse here, because of the need to have a higher threshold to create an ability district for Mexican-Americans) these excess voters are not used to create coalition, crossover, opportunity, or influence districts (because those are, again, not required by the current jurisprudence and when they are only in very specific circumstances) but instead are packed even further to create super effective vote sinks.

    The problem is not the geography per se, but problem is two-fold: 1) the jurisprudence ont he VRA and 2) this stupid fetish with pretty looking totally compact districts.

    All one has to do is look at states like Ohio, Michigan, Pennsylvania, Texas’s legislatively passed map, and others to conclude that the overriding concern was politics and that if these maps had not been passed (well, if Texas had gotten their way the problem would be even worse) that Democrats would have a majority in the U.S. House. Republicans drew these maps, and they drew them to elect Republicans even despite the fact that these states are overall not as Republican as their delegations would suggest. The maps caused these things … outside of the districts required by the VRA (and even then, remember, there is excess packing going on, which I’d argue along the lines of the Hajnal piece that you linked to a week or so ago despite the fact that that piece was specifically with regard to Presidential administrations not legislative majorities is worse for minorities themselves than using excess minority population areas that aren’t needed for VRA compliance to create other minority heavy if not majority districts that might not elect minorities themselves, but that’s not required so Republicans can get away with not doing it) … not the political geography per se.

    After all, Republicans drew the maps this way so that this would happen. That’s really all you need to know.

  3. W.M. December 12, 2012 at 3:01 pm #

    Also, it is a travesty that the only comment you single out from that thread here is “Suck on it Monkey Cage and prove me wrong with maps” as if that and that alone is representative of the discussion that took place there.

    It isn’t.

  4. W.M. December 12, 2012 at 3:22 pm #

    “At the same time, Jowei Chen and Jonathan Rodden drew thousands of maps using similar criteria for fairness and still came up with a pro-GOP bias. Across a very wide range of counterfactual scenarios, the geographic concentration of Democrats will produce votes-seats discrepancies without deliberate partisan gerrymandering. Nick’s analysis also shows evidence of this as well.

    My synthesis of all of this is that, yes, the state of the districts as of 2012 is tilted in the GOP’s favor relative to a bipartisan baseline, and this may be the outcome of multiple rounds of redistricting (2001 and 2011, plus the in-between redistrictings, e.g., in Texas). But this GOP tilt could not be completely eliminated via redistricting.”

    Sorry for the delay, but I wanted a way to respond to this concisely:

    This is what I referenced above when I said that “people are talking past one another”. Although it may be true that if you systematically create hundreds of thousands of possibile maps and then look at the average discrepancy between those maps and the overall partisan vote, this is not the same thing as saying that redistricting doesn’t matter and when it does matter only matters some. Concluding such from that data is erroneous.

    It is completely possible to draw maps that do eliminate this bias (which you say cannot be done: “But this GOP tilt could not be completely eliminated via redistricting”), and the averages that you can get really don’t point to concluding you can’t eliminate the bias. The reason that we can’t eliminate the bias is because of the legal framework.

  5. sawolf December 12, 2012 at 3:27 pm #

    Since you have decided to take my comment out of context and used my somewhat poor word choice to condescendingly criticize my position without actually defend yours, I feel that I need to take the time to defend myself here if you’re going to talk past me.

    First, when I said “suck it” I didn’t mean it to be as if I were using profanity, but rather along the lines of “take that.” Perhaps this is due to my generation (college age) not finding the phrase that offensive, but regardless my intent was certainly not to denigrate you personally.

    Second, my WHOLE point was that your article here: makes a claim that is NOT supported by the evidence you presented. That was it. Period.

    I then went on to show why you need to provide maps in defense of your position, whether you think that gerrymandering didn’t ultimately change the outcome or if you did, because simply using the old 2007-2011 maps is inherently biased. I’m not saying my maps were perfect and have never claimed to, but there’s just no way to prove that gerrymandering didn’t change the outcome without hard evidence.

    I find it very offensive that you would bring up my objection only to point out the ONE instance of mildly offensive language and instead then try to castigate my position as pretending that districting doesn’t concentrate Democrats or doesn’t matter. Whether it does or not wasn’t relevant to the point I was making, it was that Democrats STILL would have carried the house despite of it, but thanks to gerrymandering they didn’t. If you had read the diary I posted on DailyKos you would see that the distribution of seats still skewed somewhat Republican even under the non-partisan scenario, but that it was a surmountable obstacle for Democrats whereas the actual gerrymandered districts weren’t.

    As your own post asserts, we’re clearly in a very gray area when it comes to counterfactual scenarios, but we do have commissions like the type California uses where as political scientists we could construct maps that are generally realistic and even use a few sets to construct a confidence interval. Your original post made no such attempt and THAT is what I was criticizing it for.

    So again, I’ll leave it to you to show me what set of maps following a California or Iowa-style NON-partisan commission would have allowed Republicans to win 218 or more seats last month, or even the seats you think they would have retained under the old maps. But don’t bring up my comment and then proceed to talk past me as it is quite rude.

  6. John Sides December 12, 2012 at 5:08 pm #

    WM : No one is saying that “redistricting doesn’t matter.” That’s my point #1 above. And, ultimately, I actually think we agree! You point out that because of certain aspects of the legal regime — like the VRA — some Democratic votes are wasted by drawing these districts around areas with preexisting concentrations of Democrats. The same thing will happen under many scenarios when districts are drawn that are compact and respect preexisting communities. That’s what I meant in point #4: there are conflicting goals and rules in drawing districts. If you are willing to trade off some of those goals — and if the courts would allow it — in the name of ensuring strict proportionality in seats-votes, I’m sure you could do it.

    sawolf: First, in matters of style: nowhere in this post did I “castigate” you, even considering that you referred to this blog with a crude metaphor for oral sex. In fact, I described the work you had done using your own terminology (“fair”), ignoring the doubters in the Daily Kos thread who suggested that your map was too optimistic, amounted to a Democratic gerrymander, etc. (In that sense, WM, I was even favorable to sawolf in how I quoted from that thread.) And I did not “talk past you.” I acknowledged that if your maps were the baseline (or counterfactual) then indeed the current set of districts looks quite biased toward the GOP. Far from criticizing your work, I took it very much at face value and cited it without disapproving of it in the slightest, even if I wished you had been more polite in referring to this blog.

    Second, you are not reading Nicholas Goedert’s post correctly. The baseline in that post is not the 2007-2011 maps. It is the historical relationship between votes and seats. And the post agrees with you! It argues that pro-GOP gerrymanders in 2011 cost the Democrats seats in several states.

    Third, all of the this blog’s posts thus far — both McGhee’s and Goedert’s — have used “hard evidence” for their claims. And the Chen-Rodden paper drew thousands of maps using similar fairness criteria as your maps to make the point that there may be some pro-GOP tilt because of Democratic geographic concentration.

    Finally, the question, as I noted and you agree, is what the appropriate counterfactual is. And also what standards you want to uphold in drawing districts. I have no doubt that one could draw districts that (a) would pass court muster and (b) elect more Democrats — leaving aside tricky questions about what are “fair” districts. I don’t have to draw 435 districts to know that. But, as I said in my point #3 and my comment above to WM, a lot depends on how you want to make trade-offs among these different standards and rules for drawing districts.

    Thanks to you both for your comments. I hope that you can discern from my tone in this post and in this comment that I’m not interested in denigrating you or winning an argument, but in presenting relevant research and discussing the challenges inherent in studying this topic.

    • aidian December 12, 2012 at 8:08 pm #

      I would suggest that ‘suck it’ has become sufficiently divorced from its original meaning to no longer constitute a ‘crude metaphor for oral sex,’ and has been for some time.

      As for the impact of our latest travesty of apportionment, I have little to add to the discussion except to note I’ve always wanted to do something so brazenly corrupt that my name becomes synonymous with the act itself, the way Governor Elbridge Gerry did after his redistricting. Sure he took a lot of heat, but he didn’t let it faze him, and 200 years later every time we use the term gerrymandering it’s like he’s telling all them haters to suck it.

  7. W.M. December 12, 2012 at 5:49 pm #

    This was a much more even-handed response. I hope you take this as such as well, as I don’t mean it in a negative way at all.

    I think the “talking past each-other” here, though, isn’t really what you’re thinking it is.

    What I’m suggesting is that both the idea that there may be some pro-GOP bias emerging from a multitude of factors (mostly because of the legal regime, which packs the already concentrated Democratic voters), this assertion really is not getting at the meat of the issue being raised by the DKE commenters (this is a distinction that I think needs to be made here, because DKE is the holdover community from when Swing State Project was absorbed by Daily Kos. The commenting community there is… I can’t put this in less rude terms… much more intelligent than those on the main site).

    What they’re saying is that the maps passed by the GOP caused them to hold on to the house. Responding by saying that “well, the average map is going to have a pro-GOP bias anyway” isn’t responding to the assertion that these particular maps are what caused the GOP to have a majority.

    That’s what I meant by talking past one another.

    I think you address this somewhat by referencing the counterfactual, but even this seems to not really be a response adequate to the assertion. For me, I think, it boils down to the quantitative approach and the qualitative approach, and I think quite obviously a strictly quantitative approach fails to capture what is really going on. And I think that any reasonably smart analysis (quant or qual) would concede that the GOP’s maps are what caused the GOP to have their majority. It really doesn’t matter what the counterfactual (in this case: “what other map could have been drawn”) is, because practically any less-gerrymandered and legal map in PA, MI, TX, OH, and VA would have resulted in more Democratic seats than were won this cycle even if they were still gerrymandered to some degree in favor of the GOP, and especially if they weren’t gerrymandered even if there was still some residual pro-GOP bias because of the relative distribution of voters.

    Also, perhaps we should re-evaluate our choices of words. I think it more appropriate to say, instead of pro-GOP, that districting will always have some level of pro-rural and pro-suburban bias. It just so happens that at this point this bias aligns with pro-GOP bias.

    • Nick Goedert December 13, 2012 at 4:29 am #

      I think what you are saying in your penultimate paragraph here is exactly the point made in my post, which is:
      (a) If Republicans hadn’t gerrymandered PA, MI, OH, VA, etc., the Democrats would have won more seats than they did; AND
      (b) If Republicans hadn’t gerrymandered these states, the outcome would have still been biased in favor of the GOP

      • W.M. December 13, 2012 at 9:39 am #


        You never said that, but you did say this: “That is, the Democrats’ loss in the House was caused largely not by gerrymandering, but districting itself.” Albeit, that is a summary of the Chen and Rodden work, but your entire piece was seem to be meant as a qualifier to theirs, to which you said that your work found only “slightly greater effects” for redistricting. In fact, you don’t say that if Republicans hadn’t gerrymandered their state Democrats could have gotten a majority, but that if Democrats could have gerrymandered their states better, they would have gotten a majority for that.

        And I also have to point out that there are a number of assumptions that you make that I don’t think reflect reality: Florida and New Jersey’s maps are not bi-partisan. They are partisan gerrymanders masquerading under the cloak of bi-partisanship. New Jersey’s tie breaking vote along with Republicans had complete control as that person was appointed by Republican Governor Chris Christie given that the parties by are are equally represented on this committee. Florida was still legislatively drawn, and completely controlled by the Republican party. Yes, they were hamstrung somewhat by the Fair Redistricting law, but assuming that they weren’t still drawing the most favorable lines possible is milquetoast-esque.

        I’d also possibly include Arizona in this (as a Democratic gerrymander for… evenness), but you’d have a situation where an independent or Republicans (I’m not sure we actually know) appointed by a Republican would have sided with the half of the panel that were Democrats instead of the party she was supposed to side with to the outrage of the entire legislature. I think that that lends itself to actual bipartisanship.

        And then there are all the other pieces published here:

        “We have looked at this question several times before (here, here, and here) and concluded that redistricting is a wash.”

        “It’s not just about who draws the districts, but also about where Democrats and Republicans live in the first place.”

        Only here is there any real concession, but it comes after the entire piece is steeped in insinuations that anyone who believes redistricting plays a major role is off their rocker:

        • Eric McGhee December 13, 2012 at 1:32 pm #


          Can I put some more systematic numbers on this discussion? My interest has always been in the national impact, and talking about one state versus another clouds those implications. I also think it’s useful to talk about presidential vote, since that’s how you get the biggest redistricting effects, so it biases things against my own point of view.

          Dave Wasserman of Cook has reported the presidential winner in each district. Obama has 206 and Romney 225, with 4 “TBD” (which I guess means they’re really, really close). For Obama to perfectly match his national vote share, he needs about 222 seats. To get the typical bump for a majority party, you might hand him an extra four seats (1%). So the total gap due to “districting” (to borrow Nick’s term) is a maximum of 20, and might be as low as 16 (if the 4 TBD’s go to Obama). Not a huge number, but not nothing either.

          How many of those 20 seats are a product of gerrymandering, such that a reform that respected communities of interest and compactness would eliminate them? All of them? Half of them? Keep in mind that you don’t get to pick a handful of states that turned out great for Republicans and adjust those. To keep this systematic, you need to think about all districts everywhere in the country.

          Personally, I don’t have a clear answer to this question. But the evidence I’ve seen–including Nick’s analysis, the Rodden/Chen analysis, and my own look at updated bias numbers in the post you link to above–tells me most of this difference (though not all) is just political geography.

          I’m always trying to be open to new systematic evidence, and to the extent that I’m not, it’s my own fault. Show me that evidence, and I’m happy to be proven wrong.

        • Nick Goedert December 13, 2012 at 3:59 pm #

          It is understandable that people feel like the maps in NJ and FL were partisan given that the result of those maps clearly favored Republicans. But the fact is that both states implemented reforms designed to reduce or eliminate partisanship in their process. Despite those reforms, the resulting maps were still noticeably biased.

          I think if there is policy implication of my post it is exactly this: passing reforms designed to eliminate partisanship will not necessarily eliminate bias in seats outcomes. They may eliminate bias in some cases, but this will largely depend on the interaction of these institutions with existing political geography. NJ and FL in this past cycle provide good demonstrations of this principle.

          • Eric McGhee December 13, 2012 at 6:15 pm #

            I’d also add that Obama won 57% of the vote in NJ and 66% of the seats. That seems eminently fair to me, since an SMD system typically gives a boost to the majority party. Of course, the fact that Democratic House candidates only won half the seats could certainly be a product of redistricting interacting with incumbency. But if you start bringing incumbency into the picture, many of the most generous estimates of redistricting effects go out the window.

            Similar thing is true for FL. Obama won 41% of the districts–so unlike NJ, the plan is biased–but Democratic *candidates* still only managed to win 37%.

            I would also add TX to this list. Obama won 41% of the vote and 31% of the seats. This is a lot like AZ (44% vote, 33% seats) and CA (62% vote, 77% seats). Where’s the bias?

            • Eric McGhee December 13, 2012 at 6:42 pm #

              A couple more, because I can’t resist.

              Obama got 61% of the vote in MA, and 100% of the seats. Should we proactively draw lines to give Republicans 2-3 seats there? What about Minnesota, where Obama got 53% of the vote and 75% of the seats? That’s only a slightly smaller premium than in the notorious gerrymander of North Carolina (which I would agree was a gerrymander, BTW), where Obama got 48% of the vote and 23% of the seats.

              My point is that reform requires coming up with a standard process that would apply to all states, regardless of the outcome. Any standard you pick–whether proportionality or symmetry or compactness and COI–is likely to gain Dem seats in some states and lose them in others. Pick your standard carefully, test it on the entire country, and then be prepared to defend it on grounds other than “this would elect the most Democrats to office.”

              • RED December 13, 2012 at 7:33 pm #

                I don’t think you can directly compare presidential vote percentage to the percentage of that parties house delegation. Some house candidates can run ahead of their presidential nominee. Look at Rep Jim Matheson in Utah.

                Regardless, your assertion that fair maps would yield a delegation that would look similar to the presidential vote isn’t really what people have in mind when they talk about fair redistricting. To create maps like that, could essentially be called Fairymandering (fair-mander). Most people’s ideas of fair maps include drawing districts that lump together communities of interest, like say keeping all of a college town in the same district, or areas that have similar interests, so that community’s interests can be represented at the federal level.

                This may yield lopsided delegations in some instances, but that really doesn’t matter. If Fair maps were employed in MA, dems would still have 100% of the delegation, and I would have no problem with it, the same can be said of any red state that would likely have a lopsided delegation when looking at the presidential vote and the number of reps in the state of that same party under fair maps.

                My argument, and the one a few on here have been making is that fair maps, specifically using the exact system that California used, which includes a panel of indies, dems and reps drawing maps purely to make them represent communities of interest and keep them somewhat compact, would have given the house to the dems.

                • Eric McGhee December 13, 2012 at 7:53 pm #


                  The points you’re raising have been addressed on this blog many times, including in this very post. The challenge is that none of the existing systematic evidence supports the argument you’re making in the last paragraph. Neutral redistricting criteria would likely reproduce an existing Republican bias and make it virtually impossible for Democrats to undo it. Moreover, bringing the realities of incumbency into the picture only clouds things further.

  8. Greg December 12, 2012 at 6:51 pm #

    The VRA is not the only thing causing Democratic packing in cities. There are lots of majority white urban areas not affected by the VRA that are packed into one district. I live in one (WI-02). I’m sure I could draw a map that creates 8 competitive districts in Wisconsin, but I’m not sure it would be a good map. I don’t want to be represented by the kind of person who’d be elected from a competitive district. I like that my congressman-elect is a liberal fire-brand, and he would never have run, let alone won if the district were competitive.

    Similarly, I’m sure you could draw plenty of districts that put black or Hispanic voters into coalitions with white voters and it would result in more Democrats being elected overall, but the types of Democrats those districts would elect would be very different from the ones that would come from majority black districts.

  9. W.M. December 12, 2012 at 6:55 pm #

    Greg: Well, to be fair, I was not saying the VRA caused packing of Democrats in the cities. I was saying that the VRA caused packing of Democrats into certain districts. Those are two completely different statements.

  10. Jon Denn December 12, 2012 at 10:31 pm #

    Madison felt so strongly that congressional size should be limited he wanted it to be the original first amendment. He felt the “people” would lose their voice in government, otherwise. Today, congressional districts are larger than States were back then. We now essentially have a Senate and a Super Senate. Redistricting is somewhat moot as the two party system is a monopoly. It’s a great grievance without a greater solution. Here’s what I consider to be the “ax” solution on the subject. With congressional districts limited to no more than 50,000 persons we would have a House with 6000+ representatives. Most would work from their districts by computer and advanced software programs. With congressional districts that small it would be much easier for the best policies to float to the top. Committee chairs would spend more time in DC, as would national security members. But the daily grind of rewriting the tax code or making sense of the farm bill could be done more efficiently by telecommuting with 100% transparency. You can read more at

  11. Andre Kenji December 12, 2012 at 11:07 pm #

    Frankly, I think that Democrats deserves it. They had several opportunities to increase the size of the House: smaller districts would make gerrymandering much more difficult – it would be impossible to create a district going from El Paso to San Antonio, or to distribute Austin voters among six districts – and the concentration of voters in the cities would be a much smaller problem(And Al Gore would have won the Electoral Vote if the House had 600 members).

  12. M James December 13, 2012 at 12:34 am #

    Just a quick comment regarding point #3. Given that the Supreme Court has yet to rule on the merits of a partisan gerrymandering case, and given that the Supreme Court is likely to strike down more and more of the 1965 VRA, it is possible that a very pro-Democratic gerrymander that nicely spreads out its voters to maximize Democratic advantage may well be possible in the future. The “traditional districting criteria” of compactness, contiguity, respect for political subdivisions, and community of interests have no constitutional foundation, except when a Republican appointed Justice fears that his aids Black voters too much.

  13. M James December 13, 2012 at 12:36 am #

    The last part of the last sentence should read “this aids Black voters too much.”

  14. Nick Goedert December 13, 2012 at 4:20 am #

    I just realized this evening that there was a sort of Twitter wat going on yesterday over my post from last month! John, I appreciate your defense of my work, and I think you’ve represented my thesis quite well.

  15. buddyglass December 13, 2012 at 10:08 am #

    Despite not having a large effect on the aggregate level (i.e. the entire House), redistricting can significantly affect an *individual voter* if he happens to live in a newly gerrymandered district.

    I’ve always maintained that we should pay some mathematicians to come up with an algorithm for drawing districts based on some number of inputs. It should accept some set of political boundaries (zip codes, counties, etc.) on a map and population statistics for each one then spit out districts according to some deterministic set of rules that minimize certain undesirable characteristics (variability in population between districts, longest distance between any two points in a given district, etc.)

    Enshrine the algorithm into law, then just re-run it every time there’s a census and everybody has to live with the results. None of the back-room scheming and “to the victor go the spoils” stuff that happens now.

    • John Sides December 14, 2012 at 10:59 am #

      buddyglass: That is very true about individual-level effects. I can refer you to some research by my colleague Danny Hayes and Seth McKee on how voters in a new district are less likely to vote in the House election:

      There is also an interesting effect on black voters in particular:

      • buddyglass December 14, 2012 at 10:29 pm #

        Thanks John. I’ll take a look at those. Mostly I just wanted to rant about the way things are done. I’m highly sympathetic to “once and done” solutions where laws are written in such a way as to be “self-adjusting” without the need for legislators to constantly tinker with them.

  16. David Nir December 13, 2012 at 8:25 pm #

    John, I thought your move in the introduction of this post was very unfortunate. You singled out one throwaway line from a lengthy comment in a move strikes me as incredibly dismissive—”nutpicking” at its very worst, since the line you cite was in no way representative of sawolf’s work as a whole.

    There was just no reason for that intro. You should strike that portion, apologize to sawolf (and all the other non-“political scientists” who’ve engaged with you on this issue), and stick to the subtance.

    • John Sides December 13, 2012 at 9:04 pm #

      David: sawolf’s quote illustrates only one particular point, which is that people react to our posts about gerrymandering by being “cranky.” I think “cranky” is a fairly mild way to describe sawolf’s language.

      Then, as is clear in this post, I focus on substance, making several arguments about how we can think more usefully about the effects of redistricting. In so doing, I link to sawolf’s much longer post where he outlined his districts. And, as I noted in my response to sawolf’s comment in this thread, I did so to illustrate how different baselines lead to different conclusions about the effects of redistricting. As such, I took sawolf’s work as a reasonable baseline — albeit one of many potential baselines — even though other DKE commenters in that thread were skeptical of his work.

      To be honest, I am quite surprised that when a DKE diarist uses crude language to dismiss this blog without actually engaging the arguments raised by the political scientists who have posted on this blog — including a paper by Chen and Rodden that drew thousands of maps and arrived at a conclusion different than his — somehow the fault lies with me.

      • W.M. December 13, 2012 at 10:48 pm #

        This is the classic case of generational difference.

        As was pointed out above “suck on it” is not something that people of my generation view as “cranky” at all. It’s a very much endearing turn of phrase meant to evoke friendly competition. Not some huge insulting “cranky” piece of language.

        The fault is even here. I agree that he should not be entirely dismissive of the arguments that political scientists make (ofcourse, being a grad student in this field, I can certainly see that some evidence points this way but with many qualifiers and with what I consider to be some perhaps erroneous conclusions from this data), but I also think that you should realize that most readers here are going to develop a certain preconception by reading that one quote nudged along by the way even you framed it and, subsequently (if they even bother), read his comments more negatively than they otherwise would have.

        That alone, John, is academically rude. You simply do not call out the most negative points of someone’s argument. You are biasing the entire audience against them. What you did was, in a way, build a straw man.

        Before that you were, honestly, one of my favorite scholars. Your work still is, but the way you’ve gone about this by treating someone I consider a friend in an exceptionally dismissive and, frankly, rude manner has me completely disenchanted with you as an individual.

        • John Sides December 14, 2012 at 10:45 am #

          WM: I am 38 years old, so hardly a geezer. And I have been blogging for 5+ years and am not particularly thin-skinned. So I was more amused than offended by his comment. If it was meant in the spirit of friendly competition, that’s fine. But it is certainly not a stretch to interpret it as both rude and crude.

          I’ll repeat to you what I said upthread to sawolf:

          “In fact, I described the work you [sawolf] had done using your own terminology (“fair”), ignoring the doubters in the Daily Kos thread who suggested that your map was too optimistic, amounted to a Democratic gerrymander, etc. (In that sense, WM, I was even favorable to sawolf in how I quoted from that thread.) And I did not “talk past you.” I acknowledged that if your maps were the baseline (or counterfactual) then indeed the current set of districts looks quite biased toward the GOP. Far from criticizing your work, I took it very much at face value and cited it without disapproving of it in the slightest, even if I wished you had been more polite in referring to this blog.”

          So, as I also said to David Nir, I did more than cite simply the “suck on it” quote, I did not build a straw man, and I was far less rude to sawolf than he was to us. Indeed, I engaged his argument much more than he engaged the arguments presented on this blog.

          Let me also respond to something else you said up-thread:

          “It really doesn’t matter what the counterfactual (in this case: “what other map could have been drawn”) is, because practically any less-gerrymandered and legal map in PA, MI, TX, OH, and VA would have resulted in more Democratic seats than were won this cycle even if they were still gerrymandered to some degree in favor of the GOP, and especially if they weren’t gerrymandered even if there was still some residual pro-GOP bias because of the relative distribution of voters.”

          That is a hypothesis to which you are entitled. But, as a social scientist, I cannot simply agree with that in the absence of systematic data and well-defined counterfactuals. You seem to agree with the need for such data. I think I’ve made clear that there are plausible counterfactuals under which the current state of the districts appear to have a pro-GOP bias, one that cannot simply be explained by the geographic concentration of Democratic votes. But that concentration may mean that redistricting alone is not enough to eliminate any GOP bias in district boundaries, at least without trading off against some other norms and laws that undergird redistricting. We seem to agree on that as well.

          Given that, I think I’ve said as much as I can say on this thread. Thanks for your comments and for reading the blog.

  17. Nate January 28, 2013 at 1:47 am #

    Would love to see more independent commissions… however, growth in polarization in the House isn’t all that different than in the not redistricted Senate (although the house GOP DWNom is shooting out a bit lately). Can the senate be a baseline? Also, homogeneity often leads to responsiveness through reps being very much like districts…something people kind of like.

  18. Peter Principle February 4, 2013 at 4:34 pm #

    To the extent that the core of the problem is the overwhelming concentration of Democratic voters in cities and inner suburbs, the redistricting solution would seem to be the creation of a lot of wedge shaped districts — each taking in a small, but densely populated section of city and a wider, less populated swath of suburb.

    In essence, major metro areas would have to be divided up like pies – with each slice heavily Democratic at the pointy end, GOP-leaning out by the crust.

    Not familiar enough with the topic to know if this is the kind of map favored by Dems in states where they’ve been strong enough to gerrymander. Seems like it would almost have to be — unless the state is covered by the VRA, in which case it would be easy to attack on disparate impact grounds.

    A more interesting question is whether such districts meet the “natural boundaries” and “community of interest” tests. (Which don’t appear to be requirements that any court in the nation is interested in enforcing at this point, but I guess there’s always a first time.)

    If standard metropolitan areas can be regarded as single “communities,” with shared interests stretching from the inner city to the outermost exurb, then pie-slice shaped districts seem perfectly logical. But if we’re talking socio-economic “communities,” then obviously they wouldn’t.

    Of course, I don’t actually give fig about whether districts are “natural” or not: I only care about winning — and GOP partisans unquestionably feel (and act) the same way about their maps. But, since the concentration patterns of Democrats and Republicans aren’t likely to change much, or very fast, going forward, it does seem like Dems would want to put some intellectual firepower behind the idea that metropolitan areas are single, coherent communities, and should be treated as such.