Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC

How Powerful Is Gerrymandering, Anyway? A Conversation With David Daley – Wednesday October 12th in Princeton

October 7th, 2016, 12:01am by Sam Wang

Do you want to move the needle on who controls the House? Use the cool app in the left sidebar to locate competitive House districts near you. Also, use the donation links, which lead to either major party. Democrats will gain seats – but they face a major barrier to gaining a majority.

Tonight (Wednesday), I’ll sit down for a conversation with David Daley, author of Ratf**ked:┬áThe True Story Behind the Secret Plan to Steal America’s Democracy. We’ll be at Princeton’s own Labyrinth Books at 6:00pm. In Ratf**ked, Daley outlines how an unprecedentedly large wave of partisan gerrymandering swept the United States after the Census of 2010 and consequent redistricting.

Gerrymandering has been around for centuries – but a combination of factors in recent years has made it more popular as a partisan team sport. Three factors have come together: powerful redistricting software, single-party control of state legislatures, and closely divided states in which the advantage of gerrymandering is maximized.

I have calculated that the net effect of partisan gerrymandering is currently 10-15 seats benefiting the Republican Party. Several analysts, including the insightful David Wasserman, think that gerrymandering, population clustering, and incomplete candidate recruitment make it impossible for Democrats to retake the House this year. Daley thinks so too.

I am not so sure. I think a national House popular-vote win margin of D+7-8% might be enough to do it, on the grounds that this would have been enough in 2012 or 2014. Who’s right? Come watch us hash it out on Wednesday evening.

Whatever is the case with gerrymandering, it is pretty certain that Republicans will lose Congressional seats this year. The national House generic ballot has fluctuated between Democrats +2% and Democrats +7.5%. It’s been following the movements of the Clinton-versus-Trump Meta-Margin. Clinton is very likely to win the November election. The size of her victory will determine what kind of House she will face.

Tags: 2016 Election · House · Redistricting

26 Comments so far ↓

  • TJHalva

    This is sort of an esoteric question, but are states required to draw congressional districts once every ten years? My naive reading of Article 1, Section 2, appears to imply that each state is apportioned seats once every ten years not necessarily required to draw once every ten years. Is there anything preventing a state from redrawing districts each 2 year cycle?

  • maye

    The title and subtitle of Mr. Daley’s book has invoked a rant in me that has nothing to do with gerrymandering. I submit: the primary reason Obama and the congressional Democrats lost America by 2010 was a failure to communicate. They could not (and still cannot) explain what they are doing and why they are doing it. The majority of Americans did not understand the nature of the economic collapse of 2007/2008. Obama understood. He knew the history of 1929, and he could either go Hoover or go FDR. He chose FDR bailed out the banks and the auto industry, trying to head off 25% unemployment. Problem was he and his communications team could not explain what they were doing or why they were doing it. Into the information vacuum stepped the right wing propaganda machine, and the rest is history. Next came the Affordable Care Act. Three years after its passage, the average American did not understand what it was or how it was going to work. Into the vacuum: the Republican disinformation apparatus. Time after time, the Democrats and their leader President Obama failed to explain the “what” and “why” of their actions. This was PR malpractice at an astonishing level. Gerrymandering? Sure. But failure to communicate is how the Democrats lost America.

    • Rob in CT

      My recollection is that they tried to communicate it. I thought at the time they could’ve done a better job of it (in particular, Dems need to get better about message discipline in such times – keep it fairly simple and hammer the points over and over), but there was more going on.

      The problems are several:

      1) Concrete changes (as opposed to vague promises of change) are harder to sell than tearing down proposals for change, especially when people are fearful. FUD – fear, uncertainty, doubt is an effective tactic;

      2) When one side is willing to lie, constantly, it sets up an asymmetry. Media both-sides-do-it-ism aided and abetted this; and

      3) The Dems were not only less than great about messaging, they were also caught off-guard by the willingness of the GOP to blow up various long-term norms (e.g., debt ceiling chicken, ramping up filibustering dramatically). In retrospect it’s clear they should’ve nuked the filibuster on day 1, and by doing so they could’ve gotten a lot more done in the 2009-2011 window.

      Remember also that the electorate in mid-term elections is, in comparison to Presidential elections, skewed R. D-leaning turnout drops dramatically in mid-terms (R-leaning turnout drops, but not nearly as much). This is not a new phenomenon. Getting D-leaning voters to show up in “off year” elections is like the holy grail of electoral politics for the Dems.

    • Panster

      I once agreed until I happened to hear O during an interview in which he explained His restraint after “the line in the sand.” My aha moment was that O *does* explain and I don’t always hear his explanations.

      Now…. The great Republican meme machine is truly impressive, particularly in the era of 24 hour cable in hyper-partisan, post-factual America.

      But they are not filling a vacuum is my point.

      Leftists don’t talk policy using shibboleths. They go on and on and on and on like the New Yorker or the Sunday Times magazine. The entire case against O foreign policy is “he’s weak”. Every Low information voter knows this. And the case *for* his foreign policy? How much time do you have to read my next fifty thousand words?

    • liberal

      “The majority of Americans did not understand the nature of the economic collapse of 2007/2008. Obama understood.”

      I don’t think that’s true. Everything I’ve read said they thought that a stimulus package of about $1T would be enough (IIRC). Dean Baker pointed out repeatedly that simple back-of-the-envelope calculations showed it wasn’t enough.

      Now, it’s true that even if they wanted to get a higher stimulus than the one they got, they’d likely not get it. But if the question is whether they understood the magnitude of the problem, the evidence is “no”.

    • alurin

      @Liberal: My understanding is that the administration knew they needed a larger stimulus package, and that the original ARRA was intended to be just the first wave. Then they discovered that there was too much political resistance to get the rest.

    • liberal


      My knowledge of what went on was based on Dean Baker’s blog posts, plus Ron Suskind’s _Confidence Men_. My recollection is that the actually legislated stimulus was perhaps 30% less than what they wanted, but what was actually needed was 100% more than they wanted.

      As for Obama getting another bite at the apple, I vaguely recall a Krugman column pointing out that they wouldn’t get a second chance; I think he wrote that column “in real time,” without the benefit of hindsight.

      My thoughts might be off, insofar as I haven’t read about these things in awhile. But given that _Confidence Men_ shows that Geithner was so awful, he made Larry Summers look great by comparison, I wouldn’t place much stock in the notion that the administration had Main Street’s best interests at heart.

    • liberal

      This piece by Krugman seems pretty good:

  • Peter

    The point here is that Democrats communicated effectively enough to win the 2012 popular vote by more than a point and still couldn’t take the House.

    To address communication, yes Democrats could do better. But perhaps cut them some slack and consider some other points of fact. “Is Obama a Muslim?” PPP found 59% of Trump supporters say yes. “Was Obama born in the US?” NBC News found 72% of Republicans aren’t sure. “Climate change is being driven by human activity.” Pew found 15% of conservative Republicans agree. (My language; exact poll language not checked.)

    These results show that there’s a lot more going on than poor communication by Democrats.

  • Roger

    Wouldn’t gerrymandering cause the party doing the gerrymandering to be overly susceptible to large margin. By overly susceptible I mean amplified.

    Many people think gerrymandering creates super districts for your party, but as I understand it, it is the opposite – it creates super districts for your opponent not your party.

    Basically gerrymander is concentrated voters of the other party into a few districts while spreading out your party’s voters into more district. That way a state with 10 districts that are gerrymandered by Republicans, would end up with a scenario of 3 districts with something like 75% Democratic voters and 7 districts with 55% Republican voters. In normal elections to ones with just a little advantage for your opponent that means 3 Democrat seats and 7 Republicans.

    But if there was a big margin election against the gerrymandering party, then instead of losing a few seats, it would lose a very high number.

    • Carey Sublette

      Republican gerrymandering works glove-in-hand with its extreme polarization of its base through the normalization of – let’s call it what it is – lying and hate-filled demagoguery and the establishment since the early 1980s a right-wing echo chamber (90% of all radio talk shows are right wing, then there is Fox News, and the demonization of “liberal” media). Remember it is an essential objective characteristic of those drawn to conservatism that they value “authority” highly. When their authority figures and sole news sources all agree that Democrats are evil the possibility of a significant shift in Republican voting towards Democrats simply will not happen. They locked down their base to ensure virtually automatic R-voting while Gerrymandering to maximize its impact. The sort of “big margin” that would materially shift Republican votes that you imagine simply will no longer happen. Witness the Republican base solidly behind a mentally disturbed, totally ignorant, wildly erratic Trump. If they will overwhelmingly support him, they will support anyone or anything.

    • Lorem

      As I understand it, the “ideal” gerrymander is something like 80%+ majority for your opponents in “their” districts, ~60% (or a couple percent over that) majority for you in “yours”. So, yes, if everything moves by 10%, you’d get blown out, but that’s a high enough number that it’s just not going to happen.

  • Michal

    Will the conversation be recorded/available online?

  • Rachel Findley

    If the current voters were legislatively districted into the current state boundaries, would it measure up as wildly gerrymandered?

    • Sam Wang

      I do not understand your question…but read my SLR article. If districts were drawn to have partisan composition drawn from a symmetric distribution, then the resulting delegation would follow the null hypothesis of each test. For example, D winning vote shares and R winning vote shares would generally not differ using a two-sample t-test. And the mean-median difference would be t-distributed with width parameter = 0.7*SD/sqrt(N) (0.7 is approximate – I forget the exact multiplier).

      Population clustering shifts these values slightly, but not enough to set off the tests. Gerrymandering causes a massive shift.

    • Sean Patrick Santos

      If I understand you correctly, you’re asking whether the historical boundaries between states “look like” gerrymandering, especially by benefiting one party over another (e.g. for Senate, or the electoral college).

      This is kind of a hard comparison to make, for two reasons that I can think of:

      1) Unlike congressional districts, states are not even remotely close to equal in population. Wyoming is more than 10x more powerful in the Senate than it would be in a perfectly proportional body, while California is more than 6x less powerful. This effect is fundamentally different from (and probably stronger than) anything we typically think of as gerrymandering.

      2) Philosophically, we think of gerrymandering as being due to how boundaries are drawn. Democrats are sometimes at a disadvantage due to the fact that they tend to live together in cities, and this may be unfair, but it’s not necessarily Republicans’ fault; it inherently takes more care to give the parties fair representation when one is much more urban. For congressional districts, we can try to distinguish between deliberate unfairness and the “natural” unfairness that is an accidental byproduct of where people live. But for states, this goes out the window, because virtually all of the boundaries were drawn in eras where politics was far different from today.

      (An example: if you redrew the borders so that Fort Collins was in Wyoming, Wyoming elections would become far more competitive. But the Wyoming-Colorado border is clearly not carefully “gerrymandered” for party convenience; it’s a straight line drawn purely for convenience. Colorado is a relatively liberal state because of Denver’s prominence as a city, and because liberals moving to the western U.S. tend to cluster where other liberals (and more liberal laws) already are.)

  • Ed C

    Hey Sam.

    The House generic Congressional preference plot is a nice way to see where the trends are going, but “likelier Dem” and “likelier GOP” do not give me a good sense of how likely. It would be great it there were some way to create a plot that is more like for the presidential and senate races that shows the distribution of possible outcomes. I’m sure there’s not enough polling out there to really do that well, but early in the primaries you found ways to extrapolate from the limited polling data. Maybe a new way of applying Google correlate. Or maybe the number of competitive races is small enough that you could focus on those – and those races probably have more polling.

    Related to this, I am seeing a disconnect between the idea that Dems could have any possibility of taking over the house and the number of “Competitive districts” that I see in the District Finder Map. It looks like there are only about 28 competitive districts but the Dems would need 33+ new seats to take over the house. They would have to win all of the competitive districts and more.

    • Sam Wang

      I agree that House prediction is more interesting now than even two days ago. But it is the problem I feel least comfortable with. Thoughts…

      1) Converting generic Congressional preference to seats. There is much uncertainty in this relationship. I could attempt to estimate the relationship. This has the advantage that if the GOP experiences a major downdraft, the information could be incorporated quickly.

      2) District-specific poll data. This would give a district-specific result, though there are terrible time delays. In 2012, had a great compilation in 2012…I wonder if anyone is doing that now.

      And then there is the question of how to fill in missing information. Possibilities:

      3) Partisan Voting Index and incumbency advantage.

      4) Expert evaluation. In past years, a high-quality independent estimate could be gotten using the Cook Political Report and Larry Sabato’s Crystal Ball. Their ratings are again somewhat time delayed – think of all they do to evaluate hundreds of races at once. Based on the current ratings at Cook Political Report, if Democrats won all Toss-up, Lean D, and Likely D races they would get to 208 seats. Then they’d need 10 out of 25 Lean/Likely R races. That is a hard lift…but the ratings can and do change. For example, NJ-05 just shifted the other day from Lean/Likely R to Toss-up.

      5) New innovation. Your suggestion of using Google Correlate is super-interesting! Google only makes the Correlate tool available at a statewide level. District-level information could help…in all my spare time I can open a channel of communication to work on it! ;-)

      Hmmm. I wish you had not tempted me with this – I am trying to finish a neuroscience manuscript here!

    • Josie

      Sam: You can’t possibly be working on a manuscript during all this. :)

    • Sam Wang

      Trying to – it is a bit challenging but things are moving forward. I marked two PDF proofs yesterday, so not all is lost!

  • Peter

    As a sanity check, is it correct that Democrats taking the House is the only plausible road to removing Lamar Smith as Chair of the House Science Committee? Is anyone tracking efforts to walk back the recent expansion of his subpoena power?

  • Dave Kliman

    Florida state house districts are so gerrymandered that republicans hold 81 of the 120 seats, in spite of the fact that Obama won Florida with 50.1% of the vote in 2012.

    What would you propose, to fix the problem in Florida? A federal lawsuit? What would be the most fair way to draw district lines, with all of the various factors that disparate groups such as minorities want to pay attention to?

    Also, will your talk tonight be on any live feed online? will it be podcast later? I would love to hear it, in spite of the fact that I’m in Florida tonight.

  • liberal

    …and while I’m not a USSC expert, it seems pretty strong evidence that Brennan was one of the best justices ever.

  • Amitabh Lath

    One can argue with other choices they’ve made such as weighting by previous vote. But as for sticking with this guy even though he makes them a deep outlier? They made the right call.

Leave a Comment