Busting More Gerrymandering Myths
Andrew Gelman says that not only is gerrymandering not the cause of partisan polarization, it doesn’t even really make seats safer:
I can’t disagree with Cohen’s first sentence above, but I part company with him after that. When Gary and I looked at the data, we found that redistricting (“gerrymandering”) was not associated with a decline in competitiveness of elections in Congress or state legislatures. Legislative elections have been gradually becoming less competitive, but they are typically more competitive after redistricting.
I’m glad to learn of this empirical result, because I never really understood the theoretical basis of the gerrymander/uncompetitiveness link. Any constituency, no matter how you draw it, is going to have a median voter to whom one can appeal. The shape of the district ought to alter what kind of candidates are viable, but never make it impossible to field viable candidates. I would say that the biggest impediment to competitive elections is fundraising issues. If you had a public financing system that guaranteed a fully funded campaign to the major party nominees in every district, a lot of “safe” seats would suddenly start looking less safe, since it would make sense for both parties to do their best to find candidates well-suited to every district. And that, of course, is why we’re unlikely to see public financing of congressional campaigns.
39 Responses to “Busting More Gerrymandering Myths”
Register
RSSLogin
Topic Cloud
ArchivesBlog RollAbout Matt YglesiasContact Me
Donate
February 19th, 2009 at 2:08 pm
Not all redistrictings are created equal. Some are designed to make seats safer by putting more of one partisan group in them. Some are designed to become more dangerous to the non-redistricting party (gerrymandering to pick up a seat) which can create competitive elections.
February 19th, 2009 at 2:08 pm
I would also add that when times are good, people are less inclined to challenge the power structure in place. Afterall, why would they? Time were good.
Now, not so much.
February 19th, 2009 at 2:13 pm
Actually, they have funding for state legislative elections in Arizona and Maine, and it doesn’t have that big an impact on the retention rate. I’ll get back with a link.
February 19th, 2009 at 2:21 pm
My understanding is that gerrymandering isn’t meant to produce blow-out seats. Rather, you want to add more of the other party’s voters to districts they’re already winning (so that those extra votes don’t get them any extra seats). If you had a 50-50 state with 10 districts, an ideally gerrymandered map would provide one party with a single 95-5 district and the other with nine 55-45 districts. Obviously, shifting political winds will turn that 50-50 state into something else over time, but that’s not something under the control of mapmakers anyway.
February 19th, 2009 at 2:25 pm
If you had a public financing system that guaranteed a fully funded campaign to the major party nominees in every district, a lot of “safe” seats would suddenly start looking less safe.
How do you define major? I ask because funding as you describe it might make third party competition not difficult but impossible.
I’m writing from a Canadian context: 5 major parties, and in any given riding there tend to be 3 parties which have some chance to win if they had a strong enough candidate. Major parties phase into and out of existence evrry ten years or so; our current ruling conservative party only came into being in 2003, when two feuding right wing parties found a way to get back together. American politics would be very different if party-splitting was the norm; a Palin party/paleo-con party split would better reflect what has happened to the Republican party than trying to pretend that they are all on the same side.
This isn’t intended as a derail — the point you’ve made about gerrymandering is excellant. One worry: gerrymandered districts are likely to increase political polarization. All viable candidates are likely to be far to the left of the regional average in lefty-gerrymandered districts and far to the right in right-gerrymandered districts.
February 19th, 2009 at 2:26 pm
This isn’t the theory as I’ve heard it described. Say you have 64 people split into four districts, divided equally into two opposing groups, A and B of 32. In the normal course of things, eight people from each group would be in each district, and each group is represented more or less equally. Now redraw the boundaries so that two members of group A are taken from districts 1, 2, and 3, and put into district 4, while 6 members of group B are taken out of district 4 and equally distributed among the other districts. So for districts 1, 2, and 3, the ratio of A:B is not 8:8, but 6:10, while the ratio in district 4 is 14:2. Voila! Three of the four districts now reliably vote the interests of group B, even though the group strengths are the same.
That’s the theory of Gerrymandering as I’ve heard it, and I’ve seen no evidence that it doesn’t work that way in practice.
February 19th, 2009 at 2:26 pm
What about open primaries, does that make an election more competitive, or does it just produce bland centrist candidates? This is relevant for the state of California…
February 19th, 2009 at 2:27 pm
OK, there is a decline in incumbency, but its negligible — point remains there are clearly other forces at work.
From policyarchive:
“However, when a member of the House runs for the Senate, that candidate enjoys many of the benefits of incumbency. If these candidates are counted as incumbents, then the numbers change to 31 incumbents in 1996, 35 in 1998 and 33 in 2000, still showing a decline but a smaller one…”
I would point out the writers of the report forget that the reform in Maine didn’t even bring 2000 incumbency levels to what they were in 1996, so it’s arguable that the effect on incumbency is actually less than negligible.
February 19th, 2009 at 2:28 pm
Hmm… this is the second of two posts on the subject, and a healthy comment thread at the first post and at least some comments here, and still no one has mentioned Fiorina’s “The Case of the Vanishing Marginals”? That paper doesn’t quite rise to the level of “No one should be allowed to talk about X without first reading Y,” but it’s pretty seminal. It is indeed nice that there’s additional quantitative work disproving the gerrymandering link, but it should be noted that the scholarship reflected this knowledge three decades ago.
February 19th, 2009 at 2:34 pm
it should be noted that the scholarship reflected this knowledge three decades ago.
It should be noted that the premise of the current anti-gerrymender concerns is (largely) that computer modeling of districts and voters is far more sophisticated than it was 3 decades ago. 3 decades ago, computers had not added any productivity to the American white collar workforce*. I would not presume to rely on research from that time to prove that they still don’t.
* You could look it up.
February 19th, 2009 at 2:35 pm
While I’m sympathetic to the idea of public financing, I think I would argue that a challenger’s only real hope of unseating an incumbent is by outspending her.
The advantage of incumbency is the most significant factor in re-election for house members. Built in name ID, constituent services, and ability to bring in pork to the district all provide structural advantages beyond simply cash.
If the challenger cannot outspend the incumbent, she will have a hard time overcoming these advantages.
February 19th, 2009 at 2:38 pm
Yes, but given ideologically divided parties, one party will have an inherent advantage, depending on which party’s national platform (broadly considered) is closer to that median voter.
There’s a reason, after all, that many communities in the country have been governed by one party continuously for decades.
February 19th, 2009 at 2:39 pm
Any constituency, no matter how you draw it, is going to have a median voter to whom one can appeal. The shape of the district ought to alter what kind of candidates are viable, but never make it impossible to field viable candidates.
That may be the case, but there can certainly exist districts where the median voter is sufficiently far to the left (right) such that no candidate could simultaneously appeal to that voter and comfortably exist in the Republican (Democratic) party.
February 19th, 2009 at 2:44 pm
At any rate, it’s not clear to me what the gerrymandering-is-harmless claim is, exactly. Is it merely that gerrymandering won’t permanently guarantee a seat for a given party? That’s a pretty weak claim*, and you’re welcome to it.
Is the claim that it doesn’t create new incumbents, who are returned at 98% rates? That would be surprising, but since one term incumbents aren’t super-strong, there’s no guarantee that a beneficiary of a gerrymandered seat will be invincible 2 years hence.
Is the claim that gerrymandering can’t affect the ideological makeup of a district? After all, just because a Dem can win a TX district that was created to return Rs doesn’t mean that the Dem will be ideologically similar to the Dem whose district was reshaped so as to make him uncompetitive. IOW, to go back to the Austin example from yesterday, the liberal city of Austin used to be represented by 1 person, who could reliably win election by appealing primarily to his liberal constituents. Now that Austin is split among 3 districts, no one can win election by appealing primarily to those folks. Even if all 3 districts end up being Dem, none of those Dems is likely to be as liberal as the one-Austin Rep was.
Or is that wrong? Is that, specifically, what the research disproves? Or does the research merely prove that, in most places, it’s impossible to create a district that is immune to one party or the other?
* Did you know that the South used to largely vote for Dems? True fact.
February 19th, 2009 at 2:49 pm
MY is viewing this on party lines, when it needs to be viewed on ideological lines.
Heath Shuler’s election does progressives no good. If an area is drawn to overrepresent conservative voters, then that area will always elect conservatives.
Also, Ohio. Also, Texas. These states have had huge growth in statewide Democratic voting from 2002-2008, with very marginal growth in Democratic representation.
If 5% more people vote Democrat, I want 5% more Democratic reps.
February 19th, 2009 at 2:54 pm
Exactly. Most gerrymanderings are designed to corral the other party’s supporters into a few districts, while spreading one’s own supporters among a larger number of districts that ought to support one’s party normally.
Again, look at Michigan where the Reps controlled redistricting in ‘01 (Rep legislature and governor), and created 15 districts among which Kerry in ‘04 carried five and Bush won in ten, despite Kerry winning the state 51/48. The narrowest Kerry in in any of his districts was 59/40, and the other districts had Kerry margins of 21 to 64 points. By contrast, Bush won six districts with margins anywhere from one to ten points, and only two districts had Bush margins as wide as the closest Kerry district.
Now, this works fine as long as times are normal, and the advantages of incumbency can carry the members in the marginal districts. Doesn’t work so well when things are shaken up. The marginal districts can fall quickly. Best example I can think of is Indiana in the 70’s. The Reps tried to create districts that would elect 9 Reps and 2 Dems. By the middle of the decade, I think Dems had 8 seats and Reps 3. For another example, the present districting of the State House in Michigan was intended to return a permanent Rep majority, but a combination of bad times and term limits means the Dems now have 67 out of 110 seats.
The only time when gerrymandering creates lots of blow-out seats for both parties is when, as in California this decade, one party decides not to press things too far and creates mostly safe seats for both parties. My guess is this mostly happens when one party (e.g., the Dems in California) normally has a safe majority and just wants to ensure it stays safe barring an electoral tsunami, or else when there’s divided control (e.g., California in the 70’s when the Dems had the legislature and Reagan was governor), and neither party wants to play chicken and throw the decision into the courts. This last situation resulted in districts (brokered by the late Phil Burton) that made all of the incumbents safer and divided the new seats equitably between the parties.
February 19th, 2009 at 3:04 pm
And regarding MY’s point that in principle the minority party could nominate an ideologically compatible candidate for any district, yes that’s technically true, but unlikely these days because candidates aren’t chosen by party bosses but in primaries. In the three safe Dem districts in Michigan that don’t have African-American majorities, the remaining Reps who would vote in a primary are conservative enough to nominate a candidate who has no chance absent the indictment of the Dem. Dems seem to do this better (e.g., Heath Shuler) than Reps. And yes, Shuler is pretty damned conservative, but he is a vote for Pelosi as Speaker, and Dem representation on committees is determined by the number of seats they hold, so Shuler helps to maximize Dem representation on every committee in the House.
February 19th, 2009 at 3:12 pm
Matt, I think you’re missing the point. Gerrymandering is supposed to turn an even state into one with several more seats for one party, by crowding their supporters into a few districts and making all the others slightly advantageous. And it works. The most egregious states currently are Florida and Michigan:
FL: R+19, R+2, D+14, R+14, R+5, R+8, R+4, R+3, R+4, D+1, D+11, R+5, R+4, R+10, R+4, R+2, D+35, R+4, D+21, D+18, R+6, D+4, D+20, R+3, R+4
Overall: 15 of 25 reps Republican for a tossup state.
MI: R+2, R+9, R+9, R+4, D+12, R+2, R+2, R+2, R+0, R+4, R+1, D+13, D+32, D+33, D+13
Overall: 7 of 15 reps Republican (in a down year) for a moderately blue state.
You see all those R+1 to R+5 districts? That’s gerrymandering.
February 19th, 2009 at 3:24 pm
The Democrats recently – barely – won control of the Ohio House of Representatives. This was noteworthy because it is the first time in modern history that a party won control when it did not draw the district lines.
Columbus, Ohio is now overwhelmingly Democratic in statewide and local races, and if you add up the votes democrats win comparable margins in congressional and legislative races as well. They have drastically diluted representation in the statehouse and congress (2 of 3 representatives are republican, and the third freshman democrat barely won) because ultra-conservative rural and suburban areas have been sprinkled into the mix just enough to give republicans a decisive edge. The GOP has only been losing seats recently when their statewide collapse in support became severe enough that 55-45 districts became 50-50 districts.
Given our obvious local example, the clear recent examples in the Pennsylvania, Michigan, and Texas congressional distrcits, and so on I have to wonder what the point here is. Angling for a spot at Slate?
February 19th, 2009 at 3:26 pm
I think Gelman may be assuming to much when he says gerrymandering makes races less competitive when he writes: “Legislative elections have been gradually becoming less competitive, but they are typically more competitive after redistricting.”
All else being equal, elections would naturally be more competitive right after redistricting whether or not they gerrymandered since incumbent advantage is reduced by jumbling up the district lines, and taking away or creating new districts. Once incumbents are established, however, gerrymandering tends to lead to high re-election rates of incumbents, unless there is a major shift in the partisan balance, which can then lead to a phase shift — e.g., the GOP tried to gerrymander districts that are 55% GOP while using the Voting Rights Act to cram all the black and Hispanics into legally-mandated majority-minority districts that are, say, 75% Democratic. This worked great for Republicans until their support slipped more than 5 percentage points, and then Democrats flooded in.
February 19th, 2009 at 3:27 pm
Another note: would you really expect a 1989 article, which missed the previous 2 redistricting cycles with modern computers, to have captured the recent dynamic?
February 19th, 2009 at 3:30 pm
“a median voter”
You write about politics, but you don’t know about the Rule of Four?
February 19th, 2009 at 3:58 pm
If gerrymandering isn’t the cause, how does one get the 95% plus incumbent re-election rate that is so prevailent in the US? No other country in the world re-elects incumbents at anything close to that rate.
The ‘turnover’ in US elections is microscopic when compared with other nations – especially Canada and UK where half the seats in a given election might change hands. Such a result is unheard of in US elections – it just isn’t possible.
Btw, when one studies gerrymandering, merely comparing pre and post any given period of redistricting is worthless given that the district was already gerrymandered before the redistricting.
In other words, finding that a gerrymandered district that gets redistricted (re-gerrymandered) doesn’t make the seat less competitive is meaningless data given that the district likely wasn’t competitive to begin with.
When you want to ’seriously’ look at gerrymandering, start doing some comparative analysis with places that don’t permit that game. Comparing US gerrymandered districts with other US gerrymandered districts in comparison with earlier gerrymandered districts isn’t going to produce viable data. It just produces the data that the authors want.
February 19th, 2009 at 3:59 pm
I don’t even know how to respond. You must be joking. Of course, gerrymandering makes seats safer and makes them more polorized.
We have a control group to compare to the House. It is called the Senate.
435 house members and 100 senators. not a big sample but not too small either.
There are things called moderate senators. Collins and Snowe and Nelson make 3. Lieberman and Spector?? That makes 5?
Now, how many moderates are in the House? Basically, right now you only have a very few Republicans and a handful of bluedogs. There is no way you have 3% of the house as moderate as the 3 senators above.
Now, look at re-election rates. It is harder to win re-election in the Senate because they can’t gerrymander a state border.
Honestly, what am I missing????
February 19th, 2009 at 4:03 pm
I always thought the idea was to pack as many of your opponents into one district, thereby creating a number of other districts that are likely to vote for your side.
For example, if there are 100 voters that need divided among 4 districts and their party affiliation is evenly split between Party A and B, Party A would do well to create the following districts:
D1 – 5 Party A, 20 Party B
D2 – 15 Party A, 10 Party B
D3 – 15 Party A, 10 Party B
D4 – 15 Party A, 10 Party B
Despite each party having 50 registered voters, Party A is likely to capture 3 of the 4 seats due to the gerrymandered District 1.
February 19th, 2009 at 4:03 pm
One other point about gerrymandering. The ‘polarization’ comes from the primary process for gerrymandered seats. If a district is safe (and over 90% of them are ’safe’ in Congress) then the only challenge to that district is from a primary challenger.
Thus, in safe districts, one tends to see more extremist positions (far left and far right).
One only has to look at the ‘ideology’ curve for Congress – it looks like a big “U” – very high at the extremes and virtually nothing in the middle. That is the product of gerrymandering in combination with the primary system.
February 19th, 2009 at 4:28 pm
Effective gerrymandering certainly makes seats safer. The famous Delay plan, for example, actually made Delay’s district more competitive by taking a handful of Republicans out of his district and roling them into others. So, roughly, you went from having 1 very safe Republican district, one swing district, and 1 fairly safe Democratic district to having 2 fairly safe Republican districts and 1 swing district.
The sort of incumbent protecting hyper partisan gerrymandering doesn’t really exist in all that many places outside of California state legislative districts.
February 19th, 2009 at 4:32 pm
Gerrymandering might not affect the competitiveness of most races, but will affect the competitiveness of a few races — i.e., even if 70% of the races are unaffected, the 30% that are can produce a significant change in the makeup of the legislature. The “average” is deceptive.
February 19th, 2009 at 4:32 pm
Gerrymandering might not affect the competitiveness of most races, but will affect the competitiveness of a few races — i.e., even if 70% of the races are unaffected, the 30% that are can produce a significant change in the makeup of the legislature. The “average” is deceptive.
February 19th, 2009 at 6:22 pm
I don’t see how Gelman’s counter-argument holds up against the reality of racial, ethnic and ideological gerrymandering.
When black voters are ghettoized into a congressional district, the adjacent, whiter districts elect more conservatives and/or Republicans. Much the same thing happens in the Southwest with the creation of predominantly Latino districts. And in most larger states, you can find one or more examples of the party in power diluting concentrations of opposition voters by drawing them into several districts.
This effect may be obscured by playing with statistics from the last two election cycles, which were Democratic wave years — with a number of upset victors in unlikely places, most of them destined for short tenures — and which occurred toward the end of the decade, when districts drawn in 2001 have changed demographically.
Gerrymandering has a real effect, but outside ghetto districts it wears off as time passes.
February 21st, 2009 at 12:13 am
There can and there do. The Republicans could in principle nominate someone who would appeal to the median voter in Berkeley, but no such person could ever be accepted as a legitimate member of the Republican party. Nobody that far left of center would ever want to be a Republican and real-world Republican primary voters wouldn’t be interested in voting for such a person. Same goes for a Democrat in Lynchburg.
March 1st, 2009 at 5:43 am
viagra
Very interesting site. Hope it will always be alive!
March 2nd, 2009 at 5:10 am
levitraGreat site. Good info
March 11th, 2009 at 4:40 am
Incredible site!
March 22nd, 2009 at 6:14 am
tramadol
Very interesting site. Hope it will always be alive!
March 22nd, 2009 at 10:29 am
buy viagra online
Incredible site!
April 2nd, 2009 at 5:18 am
Very interesting site. Hope it will always be alive!
buy cheap viagra
April 9th, 2009 at 6:02 am
thanks !! very helpful post! viagra
April 19th, 2009 at 8:51 am
N8Bc6j kmqyjewxmlfy, [url=http://yoykpxejjkex.com/]yoykpxejjkex[/url], [link=http://qphfsxbzvxef.com/]qphfsxbzvxef[/link], http://bnperbfqjocs.com/