Gerrymandering: having it both ways?
October 20, 2013 8:25 AM Subscribe
It seems to me that there are two different things you can accomplish with gerrymandering that are not mutually compatible: 1) make "your" districts safer by concentrating your side's voters in them, or 2) win more districts by inducing relatively narrow victories in as many districts as you can. But the references to gerrymandering in the US House of Representatives seem to imply both things are happening. Which is it?
The discussions I've seen about the makeup of the House of Representatives tend to claim both that 1) extreme Republicans don't have to worry about general election fights due to gerrymandering and 2) Republicans have a greater than "deserved" representation in the House due to gerrymandering. But from a mathematical perspective these two things, while not necessarily mutually exclusive, certainly work against each other.
If you have control over the districting of a state with a relatively balanced voting population, you could try to make a few of your districts safe by moving more of your side's voters into them, but then you've made the races closer in other districts. Or you could try to win a lot of districts by making a couple of them blowout victories for the other side and getting a narrow advantage in all the others. But these two strategies are at odds.
I guess it is possible mathematically to get the best of both worlds by shoving all of the opposition into a single 100-0 district, but of course that doesn't work in the real world.
Have Republicans really managed to redistrict some states so that they have both a greater number of districts and a greater chance of winning each one? Or by redistricting to win more seats, have they spread themselves thinner and made those seats a bit more at risk? Links to actual statistical analyses would be welcome.
The discussions I've seen about the makeup of the House of Representatives tend to claim both that 1) extreme Republicans don't have to worry about general election fights due to gerrymandering and 2) Republicans have a greater than "deserved" representation in the House due to gerrymandering. But from a mathematical perspective these two things, while not necessarily mutually exclusive, certainly work against each other.
If you have control over the districting of a state with a relatively balanced voting population, you could try to make a few of your districts safe by moving more of your side's voters into them, but then you've made the races closer in other districts. Or you could try to win a lot of districts by making a couple of them blowout victories for the other side and getting a narrow advantage in all the others. But these two strategies are at odds.
I guess it is possible mathematically to get the best of both worlds by shoving all of the opposition into a single 100-0 district, but of course that doesn't work in the real world.
Have Republicans really managed to redistrict some states so that they have both a greater number of districts and a greater chance of winning each one? Or by redistricting to win more seats, have they spread themselves thinner and made those seats a bit more at risk? Links to actual statistical analyses would be welcome.
You can abandon some districts. So the republican suburbs each get paired with a small bite of the democratic downtown, then they make a couple of districts with what's left of downtown, and just let the democrats win with huge majorities there.
posted by Monday, stony Monday at 8:38 AM on October 20, 2013
posted by Monday, stony Monday at 8:38 AM on October 20, 2013
Let's say you have a state with 100 units of voters and five districts of 20 units each. Let's further say that you have 35% of voters from Party A and 65% from Party B. Today District 1 has 13 units of A and 7 of B, and D2 has 11 of A and 9 of B, D3 has 8 of A and 12 of B, D4 has 2 of A and 18 of B, and D5 has 1 of A and 19 of B. District 1 will be a fairly reliable win for A, Districts 2 and 3 will be fairly competitive, and Districts 4 and 5 will be reliable for B. Now, imagine that we redistrict so that D1 is 19A/1B, D2 is 5A/15B, D3 is 6A/14B, D4 is 5A/15B and D5 is 0A/20B. Now Districts 2 and 3 swing heavily for Party B and District 5 is a dead lock for B. Additionally, if Party B can move a lot of "undesirable minority voters" into District A, they can portray Party A as a bunch of communist Koosbanians and maybe get people to come over to their side.
posted by slkinsey at 8:52 AM on October 20, 2013 [2 favorites]
posted by slkinsey at 8:52 AM on October 20, 2013 [2 favorites]
Red or Green has it -- in fact the 80-20 Democrat, 60-40 Republican district phenomenon is extremely common.
However, partisan gerrymandering plays a distinctly modest role in achieving it.
Population patterns are the biggest factor: there is simply no analogy among Republican-leaning people to the concentration of Democrat-voting minorities in the inner cities or Democrat-leaning SWPL types in those city's nice neighborhoods or liberal suburbs. Or think of like this: precious few Republican bankers live in the ghetto, but plenty of Democrat school teachers live in the exurbs.
Secondarily is non-partisan gerrymandering to maximize "minority-majority districts." You would have more Democrat officeholders if you split up minority neighborhoods and used the votes to dilute Republicans in the suburbs and exurbs ... but you'd have many fewer minority Democrats, and that's an illegal result under the Voting Rights Act, and seen as a less desirable result by many minority interest groups.
posted by MattD at 9:06 AM on October 20, 2013 [2 favorites]
However, partisan gerrymandering plays a distinctly modest role in achieving it.
Population patterns are the biggest factor: there is simply no analogy among Republican-leaning people to the concentration of Democrat-voting minorities in the inner cities or Democrat-leaning SWPL types in those city's nice neighborhoods or liberal suburbs. Or think of like this: precious few Republican bankers live in the ghetto, but plenty of Democrat school teachers live in the exurbs.
Secondarily is non-partisan gerrymandering to maximize "minority-majority districts." You would have more Democrat officeholders if you split up minority neighborhoods and used the votes to dilute Republicans in the suburbs and exurbs ... but you'd have many fewer minority Democrats, and that's an illegal result under the Voting Rights Act, and seen as a less desirable result by many minority interest groups.
posted by MattD at 9:06 AM on October 20, 2013 [2 favorites]
You can see this happening very clearly in the Cook Partisan Voting Index. There are 20 congressional districts where Democrats hold at least a 30-point advantage, but only 3 like this for Republicans. MattD is correct that this has largely to do with big cities being very Democratic; every one of those D+30 districts I clicked on is contained or mostly contained in a big city, and I'll bet they all are. (Liberal suburbs are not so much of a factor: places like MD-8 in Montgomery County where I grew up are D+11, and even WI-2, where I live now (mostly Madison) is only D+17.)
On the other hand, there are 26 districts that are either R+1 or R+2, but only 8 that are D+1 or D+2. I don't believe that has to do with cities; I think that has to do with people drawing district boundaries extremely carefully to create a large group of districts with a slight Republican edge.
In other words, both things are happening.
posted by escabeche at 9:35 AM on October 20, 2013 [2 favorites]
On the other hand, there are 26 districts that are either R+1 or R+2, but only 8 that are D+1 or D+2. I don't believe that has to do with cities; I think that has to do with people drawing district boundaries extremely carefully to create a large group of districts with a slight Republican edge.
In other words, both things are happening.
posted by escabeche at 9:35 AM on October 20, 2013 [2 favorites]
My sense is that mostly what you've identified is just that critics of Congress who point to gerrymandering aren't terribly consistent or don't understand gerrymandering very well. I'm way too lazy to go look it up now, but I'd swear I've seen popular-press critics/pundits argue that gerrymandering means creating safe Republican districts and that it means maximizing the number of Republicans in the same article, even though these goals are diametrically opposed.
Across Congress, you can see some of both. There are some states with some very safe Republican districts that were probably created with that goal, or with more general incumbent protection if the state had divided government. In 2012, for example, Alabama only had pretty-safe districts. Other states, like PA, more or less maximized the number of expected Republican seats. As others have noted, this means that PA in particular is liable to a really large swing in representation if there's even a moderate Democratic wave between now and 2022. New York probably also came pretty close to maximizing the number of Republican MCs elected. If you want links, I'd just go straight to the presidential vote by congressional district.
posted by ROU_Xenophobe at 9:38 AM on October 20, 2013
Across Congress, you can see some of both. There are some states with some very safe Republican districts that were probably created with that goal, or with more general incumbent protection if the state had divided government. In 2012, for example, Alabama only had pretty-safe districts. Other states, like PA, more or less maximized the number of expected Republican seats. As others have noted, this means that PA in particular is liable to a really large swing in representation if there's even a moderate Democratic wave between now and 2022. New York probably also came pretty close to maximizing the number of Republican MCs elected. If you want links, I'd just go straight to the presidential vote by congressional district.
posted by ROU_Xenophobe at 9:38 AM on October 20, 2013
I would put less weight on critics/pundits being right in some careful, meaningful way that both are happening at once and a lot more weight on those pundits/critics basically pointing at gerrymandering from the logic of "Crazy Republicans are something I don't like, gerrymandering is something else I don't like, so crazy Republicans must be because of gerrymandering." Granted I'm pretty jaded about this because commentators in the mass media are just so wrong, so often.
The idea that the current state of polarization is because of gerrymandering has been considered and rejected; the actual paper by some combination of Poole, Rosenthal, and McCarty is somewhere on voteview.com.
The idea also doesn't make a whole lot of sense, because primary and general elections happen in sequence. If you're facing a crazypants Tea Party challenger in your primary, it doesn't really matter how marginal your district is in the general election -- you still have to do what it takes to win the primary first, or you don't get to compete in the general. Having a campaign and set of positions that would have won you the general election doesn't help you one damn bit if you've lost the primary, at least in states with sore-loser laws. If you want to think of it formally, preferences over primary vote share and general vote share are lexical. All you need to get Republican craziness is crazy primary voters, which they have in abundance.
posted by ROU_Xenophobe at 9:47 AM on October 20, 2013
The idea that the current state of polarization is because of gerrymandering has been considered and rejected; the actual paper by some combination of Poole, Rosenthal, and McCarty is somewhere on voteview.com.
The idea also doesn't make a whole lot of sense, because primary and general elections happen in sequence. If you're facing a crazypants Tea Party challenger in your primary, it doesn't really matter how marginal your district is in the general election -- you still have to do what it takes to win the primary first, or you don't get to compete in the general. Having a campaign and set of positions that would have won you the general election doesn't help you one damn bit if you've lost the primary, at least in states with sore-loser laws. If you want to think of it formally, preferences over primary vote share and general vote share are lexical. All you need to get Republican craziness is crazy primary voters, which they have in abundance.
posted by ROU_Xenophobe at 9:47 AM on October 20, 2013
Here's a good post from poli sci blog The Monkey Cage on the subject. The analysis here suggests (and I do mean suggests, not proves) that concentration of Democratic voters in big cities does build in a certain tilt towards Republicans (i.e. if 52% of the public wants a Democratic representative, the House will have a GOP majority) but that this tilt was substantially larger in states with district boundaries drawn by Republican majorities (and flipped in the other direction in states with district boundaries drawn by Democrats.)
In other words, gerrymandering isn't everything, but it isn't nothing, either.
posted by escabeche at 9:56 AM on October 20, 2013 [2 favorites]
In other words, gerrymandering isn't everything, but it isn't nothing, either.
posted by escabeche at 9:56 AM on October 20, 2013 [2 favorites]
Much of the discussion so far seems to be centered around this apparent contradition:
1) The party in charge of redistricting wants to create "safe" districts for their party.
2) The party in charge of redistricting want to pack the other party into a small number of districts and have the remainder of the districts be "close"-- taking a big hit in a small number of districts and having a close race in many other districts.
The truth is that #2 is greatly preferred. Even a district in which 50% voted for Obama can be an absolutely "safe" district for a Republican. Say the local republican is an incumbent. They have the built-in advantage of carefully cultivated patronage networks, and they also have the ability to portray themselves as "centrist." In a wealthy, urban district, say, Manhattan's Upper East Side, they can lean socially liberal and economically conservative. In, say, the rust belt, with its large proportion of blue-collar workers and Catholics, they can be pro-labor, but anti-abortion. And, on top of this, the incumbent has huge advantages in name-recognition, patronage networks, and carefully-cultivated relationships of trust and reciprocity with the most important local players on both sides of the political divide. Also, many incumbents have a legacy of years to decades of constituent services. (This was one of the reasons Dixiecrat Jesse Helms of NC managed to stay in power for so long.)
Another apparent contradiction: for a while now, approval of Congress has been at record lows, yet people re-elect House members at greater than 90%. They think Congress is terrible, but they like and trust their own representative. Certainly some of this follows from gerrymandering, but I personally believe much follows as well from notions of loyalty.
Furthermore, since incumbency is so important, another trick is to reshape districts to "merge" two districts of the opposite party, forcing two incumbents to run against one another in the same district. So now, rather than 90% expectation of re-election, you kick out one incumbent with 100% certainty.
So, even though #2 is the dominant, over-arching strategy, there are a number of other tricks that can be used. If your party has the ability to draw the lines, you can assume that they will use any number of strategies to secure the largest possible advantage over a range of outcomes.
posted by Maxwell_Smart at 10:22 AM on October 20, 2013 [1 favorite]
1) The party in charge of redistricting wants to create "safe" districts for their party.
2) The party in charge of redistricting want to pack the other party into a small number of districts and have the remainder of the districts be "close"-- taking a big hit in a small number of districts and having a close race in many other districts.
The truth is that #2 is greatly preferred. Even a district in which 50% voted for Obama can be an absolutely "safe" district for a Republican. Say the local republican is an incumbent. They have the built-in advantage of carefully cultivated patronage networks, and they also have the ability to portray themselves as "centrist." In a wealthy, urban district, say, Manhattan's Upper East Side, they can lean socially liberal and economically conservative. In, say, the rust belt, with its large proportion of blue-collar workers and Catholics, they can be pro-labor, but anti-abortion. And, on top of this, the incumbent has huge advantages in name-recognition, patronage networks, and carefully-cultivated relationships of trust and reciprocity with the most important local players on both sides of the political divide. Also, many incumbents have a legacy of years to decades of constituent services. (This was one of the reasons Dixiecrat Jesse Helms of NC managed to stay in power for so long.)
Another apparent contradiction: for a while now, approval of Congress has been at record lows, yet people re-elect House members at greater than 90%. They think Congress is terrible, but they like and trust their own representative. Certainly some of this follows from gerrymandering, but I personally believe much follows as well from notions of loyalty.
Furthermore, since incumbency is so important, another trick is to reshape districts to "merge" two districts of the opposite party, forcing two incumbents to run against one another in the same district. So now, rather than 90% expectation of re-election, you kick out one incumbent with 100% certainty.
So, even though #2 is the dominant, over-arching strategy, there are a number of other tricks that can be used. If your party has the ability to draw the lines, you can assume that they will use any number of strategies to secure the largest possible advantage over a range of outcomes.
posted by Maxwell_Smart at 10:22 AM on October 20, 2013 [1 favorite]
This NYTimes article explains how "Democrats received 1.4 million more votes for the House of Representatives, yet Republicans won control of the House by a 234 to 201 margin" through the use of "packing throwaway districts" and "arranging to win close victories" elsewhere.
posted by mdn at 10:35 AM on October 20, 2013
posted by mdn at 10:35 AM on October 20, 2013
Another apparent contradiction: for a while now, approval of Congress has been at record lows, yet people re-elect House members at greater than 90%. They think Congress is terrible, but they like and trust their own representative. Certainly some of this follows from gerrymandering, but I personally believe much follows as well from notions of loyalty.
This has been around since at least the 70s; Fenno's famous phrasing is that people run for Congress by running against Congress. So it's unlikely that the recent trend towards more obvious gerrymandering has much to do with it.
Hibbings and Thiess-Morse point to the other direction in Congress as Public Enemy and Stealth Democracy -- the puzzle might not be why people like their Congressman so much as why they dislike Congress. And their answer simplifies to "Because Congress is in the news a lot, and because people don't really like democracy very much." Democracy is messy and conflictual and there's compromise. What people mostly seem to want is a responsive, empathetic dictator.
posted by ROU_Xenophobe at 10:41 AM on October 20, 2013
This has been around since at least the 70s; Fenno's famous phrasing is that people run for Congress by running against Congress. So it's unlikely that the recent trend towards more obvious gerrymandering has much to do with it.
Hibbings and Thiess-Morse point to the other direction in Congress as Public Enemy and Stealth Democracy -- the puzzle might not be why people like their Congressman so much as why they dislike Congress. And their answer simplifies to "Because Congress is in the news a lot, and because people don't really like democracy very much." Democracy is messy and conflictual and there's compromise. What people mostly seem to want is a responsive, empathetic dictator.
posted by ROU_Xenophobe at 10:41 AM on October 20, 2013
It's important to remember that these super-safe Tea Party Republicans that can act with impunity are a minority even within their own party, as you can see from the recent debt ceiling vote. Most of the pro-Republican gerrymandering is focused on making a lot of districts that tilt Republican by ~10 points, typically a very strong position for an incumbent (especially during a midterm election).
Good news is that a "wave" election favoring the Democrats could be enough to overtop many of these electoral levees, resulting in a lopsided Democratic victory. But it would take a sizable wave given the current district lines.
posted by Rhaomi at 1:31 AM on October 22, 2013
Good news is that a "wave" election favoring the Democrats could be enough to overtop many of these electoral levees, resulting in a lopsided Democratic victory. But it would take a sizable wave given the current district lines.
posted by Rhaomi at 1:31 AM on October 22, 2013
This thread is closed to new comments.
Actually, it more or less does. You can make geographically contiguous districts that cover for example inner urban areas, and end up with better than 80-20 Democratic representation. Such districts exist, IIRC.
So your answer is really "both": you pack opponent voters into single districts so that they win a couple of seats absolutely overwhelmingly, then distribute your voters over 60-40 majority districts so that you win many more that are still very safe. (10 point swings don't happen easily.)
posted by RedOrGreen at 8:36 AM on October 20, 2013 [5 favorites]