Bell Curve The Law Talking Guy Raised by Republicans U.S. West
Well, he's kind of had it in for me ever since I accidentally ran over his dog. Actually, replace "accidentally" with "repeatedly," and replace "dog" with "son."

Wednesday, January 03, 2007

More on Redistricting Reform

From today's "state of the State" speech by NY Governor Eliot Spitzer (D)
"Still, we must do more. We will submit legislation that reforms our
elections – specifically legislation that establishes an independent,
non-partisan redistricting commission. Until this happens, I will veto
any proposal that reflects partisan gerrymandering. More competitive
elections will lead to a more responsive government."

FYI, Republicans control a close NY Senate, Democrats control the NY house by more than 2/3.

Dr. S is expected to be excited. In CA, Schwarzenegger is also going to take another stab at this. The big question now is whether only blue states will go for "non-partisan" schemes, or whether red states will do as well.


Anonymous said...

Dr S is expected to be excited. RBR is expected to be grumpy. LTG is expected to be defensive. US West is expected to be conciliatory. And Bell Curve is expected to be silent on the matter. 

// posted by Quokka of ..uh

Anonymous said...

Well, far be it from me to disappoint my fans! Woo-hoo!!

It seems to me that there are two basic strategies for partisan redistricting: minimize or maximize risk. Protect or grow.

In the protection strategy, you minimize the likelihood that representatives of your own party will lose their seats by shoring up their districts to include favorable voters. Naturally, this also helps opposition representatives keep their seats as well by draining their districts of dissenters. California did this famously in 2000. This is useful when the party in power already has a strong lead, or when the concerns of local politicians trump national party considerations.

In the growth strategy, you try to spread out your favorable voters thinly so as to maximize the number of seats your party can gain and control. Texas did this famously under Delay in 2003. This is useful when the party in power is only weakly in the lead, or when the national party can impose its will on the state officials (as Delay so famously twisted arms).

If one assumes that "non-partisan" schemes would be more likely to fall somewhere in the middle of those two extremes, it would seem that national party leaders might push for such schemes in states that have chosen the protective route, like California. My question for the devoted constitutional scholars amongst us... is what I just wrote here about right?

Anonymous said...

And Quokka is expected to be a smartass...

Just kidding. No flaming please. Couldn't resist.


// posted by LTG

Anonymous said...

(flashback--no such thing as "non-partisan" districting).

States that seem to have reputations for "fair" districts have BI-partisan districting boards in which the partisan incentives are balanced off against each other - hopefully cancelling each other out. If NY goes that direction, I don't have a huge problem with it.

But if Blue states like NY and CA try for the mythical "non-partisan" solutions such as handing this over to judges (notoriously more conservative than the rest of us), it will be a huge boost to Republicans.

Frankly, I don't believe for a minute that Republican states are even considering redistricting reform. If I'm right about that, I'd advocate states like California holding GOP districts there hostage to protect districting fairness in places like Texas. The idea would be to impose a kind of national bi-partisan districting approach. 

// posted by RBR

Anonymous said...

I don't get it. Is that just defensive or are you moving to passive-aggressive? :-) (How could I resist?)

I think the nub of the disagreement between RbR and Dr S is in the parentheses. RbR, quite rightly, does not believe people can be non-partisan. In his mind's eye the fairest system would be to have the two parties having equal say and fighting it out in some board meeting.

Dr S believes it is possible to obtain as-fair-as-possible redistricting by using statistical methods to form regions based simply on population numbers and geographical distance, given appropriate rules. This would be non-partisan in that it is based on measures that have no political bias. That is not to say that there is no political implication: obviously wealthy people group together and will vote in certain ways.

What am I trying to say? Well I obviously side with Dr S because I'm science-trained and through that you work towards removing biases, and we may see things in ways to remove bias. RbR's statement 'no such thing as "non-partisan" districting' we will obviously not agree with because it implicitly relies on RbR's political science assumptions of truth, as opposed to Dr S's assumptions of statistical/mathematical truth.

Does that explanation ring true to anyone? 

// posted by Quokka with Intelligent Buttocks

Anonymous said...

I agree with 95% of what Quokka said. (No surprise there.) The 5% that I don't agree with the very last bit, regarding political science assumptions of truth vs. statistical/mathematical assumptions of truth. I understand (I think) where Quokka is coming from and I see where he has gone, but I don't think it is necessary to go that far.

I am sure RbR can tell that some redistricting plans show heavy lopsided bias favoring one party, while other plans are more of a mixed bag without any clear intentional bias in the design. Indeed, political scientists are often employed by lawmakers to ensure that the plan produced is horrendously biased in their favor. I also believe (though I am less sure of this) that RbR agrees that a clear set of rules could be constructed which, if faithfully and honestly applied, would be very likely to generate redistricting schemes that are not heavily biased toward one party.

To my mind, the chief difference between us is that RbR is unwilling to trust any committee to honestly and faithfully apply those rules--no matter how cleverly the members are selected or how carefully the process is designed... whereas I am willing to trust an appropriately crafted process to produce significantly less biased redistricting schemes. As RbR sees bias to be unavoidable, he naturally prefers that bias to be out in the open rather than hidden behind a veil of supposed non-partisanship. As I believe (through good process design) that bias can be significantly mitigated, I am willing to accept a (mostly) non-partisan process as a worthwhile improvement over the current free-for-all.

There is, however, another difference between RbR and I that we've discussed in the past at length... but still bears repeating. (This is what I think Quokka was talking about mostly.) RbR and I have different concepts of fairness, I think. When a neutral outcome can be postulated--and I think both RbR and I agree that it can be postulated here--I do not believe that, "two parties with equal say fighting it out," is a fair process when the near-certain outcome is winner-takes-all... in other words, a process that always leads to an unfair outcome is not, in my view, a particularly fair process. In these situations, I much prefer processes that are designed to lead to compromises, that split the difference down the middle. RbR is content with the coin flip; I want to try as hard as possible to balance it on its edge.

Anonymous said...

We've got a contest of two ideas of governance here, not a contest between different kinds of truth. It's a little obnoxious for Quokka to state that his views are scientific, as opposed to others that are unscientific. The science of government is based on long experience and observation. With respect to those who are trained in the "hard" sciences, it is a mistake to ignore the accumulated wisdom of the science of government in favor of your own independent reasoning. Aristotle's attempt to reason through the hard sciences should be instructive here. Empirical knowledge matters. Learning matters. Thought experiments cannot substitute for actual experiments. As a lawyer, I constantly deal with people who think that if they're smart enough, they can understand the law without going to law school. They make enormous mistakes. Doctors experience the same instinct of laypeople that they should be able to use their reason (and WebMD) as a substitute for their expertise. Were it not for the need to get a doctor's signature on a prescription, these people would medicate themselves to death. I know Dr.S despises the constant commentary almost everywhere using "quantum physics" as the justification for all manner of unscientific hunches. Democracy does not mean that every citizen is equally able to understand government.

The Madisonian theory of government is that bias ("faction") is best handled by checks and balances, not by trying (in the words of Learned Hand) to select a "bevy of platonic guardians." Dr.S and Quokka seem to prefer the latter - designing a system to select the right people and give them the right set of rules to follow. RBR believes that this idea only leads to pervasive biases unchecked, and that the only way to control bias is to allow democratic contest.

Dr.S and Quokka are right in one sense, that "hard" scientists tend to the "guardian" approach. Scientists and engineers also, I think, paradoxically favor libertarianism for the same reason - a desire to design a perfect, efficient system. A belief in deductive logic from basic postulates about human behavior. The implicit analogy is government as an engine or a system, and the desire is for government to behave like a meritocracy, perhaps as an idealized university. Madisonian government, with its checks and balances, puts little emphasis on efficiency. Some take the marketplace analogy, where we expect openness and competition will result in good policy, or at least the best policy possible. I think a battlefield is more accurate.

Madison's theories, as elaborated 200 years later, are based not just on philosophy about human beings and reason, but on studies of government, on the experience of government. The Federalist Papers are noteworthy as an exposition the failures of past forms of government, and an attempt to justify the "noble experiment." When RBR says that non-partisan redistricting does not eliminate bias, this is not a plain philosophical proposition, but the result of scientific observation of human institutions that deserves serious consideration. The question is really whether you prefer a subtle, pervasive, hidden (likely right-wing) bias of a non-partisan commission, or results of the contest between the majority's desire to gerrymander for open political gain, and the willingness of the courts and the minority to tolerate that behavior. There is no easy way out, no easy "middle ground" intellectually. I suspect we do not have enough data yet to evaluate which policy produces better results, or even an agreement about how to measure that.

There's another interesting irony here. Dr.S and Quokka believe in expertise in government, but dismiss expertise about government. RBR and I take the reverse position. I think we can have this debate without insulting each other's intelligence, faculties of reason, or education.

// posted by LTG

Anonymous said...

In my "conciliatory" role, I mind add just a touch of practicality here. I am the woman, after all. ;-)

Redistricting is based often on census figures taken every 10 years (unless you are Texas where you think it should be done more often.). Considering how often and how far Americans move and considering the constant influx of immigrants to this lovely country, the composition of any district will change-perhaps dramatically-over time. The Census is meant to address this fact. However, there is great debate (on this blog as well as many others) about the accuracy of census data and whether sampling is adequate for our purposes. So if we can’t even settle on the proper method of accounting for people, and on a time frame for how often we account for them then I don’t see how we can deal with redistricting.

Add to this the changes that are taking place in the parties, and the flip flopping on issues, and you basically have a moving target. Just take the idea of "big government" and how the parties have rearranged themselves on that issue alone.

To touch on what LTG said, and perhaps I am rephrasing him . . . in our conception of government, democracy is the people, not just a method for managing the madness. As such, it is organic and will change over time- you know . . . EVOLVE. This evolution will be influenced by many factors, some unanticipated. You can build a system for redistricting, and you might even control it for awhile, and it may work for a few election cycles. But then, eveything will change and suddenly it doesn't work anymore. No system is perfect (Sorry Dr. S), nor are they static. Witness our own government. We have had to amend the Constitution 27 times. The best laid plans are often usurped. So I guess what I am getting at is the idea that maybe we don’t, in the end have much to worry about (I realize this is a departure from my normal take on things. I am trying to adopt a more broad, long-term view in 2007.)

In the social sciences, we talk about systemic analysis or organizational theory. Perhaps in the sciences it is Chaos Theory? How about just plain and simple Biology?

So, maybe rather than getting into discussions over "hard" science vs. everything else, which is usually code for quantifiable data over qualitative, we might look at the place where social sciences and life sciences overlap? They both have so much to offer in terms of understanding how systems work, and what kinds of adjustments can be made to influence how they work.

// posted by USWest

Anonymous said...

I agree with just about everything else LTG said. His description of the experience and scientific study of government was succinct and eloquent--especially his description of Madisonian democracy, checks and balances, etc. And I accept that RbR's position is not merely a "lack of faith" in committee processes--not merely a philosophical point of view (as I believe I unfairly characterized it)--but rather an expert opinion based on a great deal of historical evidence. There are, however, a few things LTG and USWest said that I want to respond to.

First: LTG wrongly accused me of bashing political science. I am tired of that charge, especially now that I often go out of my way to show respect (in part as penance for past transgressions). Despite what LTG implied, I clearly wrote that I do not think our disagreement is about what constitutes science or truth, and I said nothing dismissive of anyone's expertise in the study of governance. If my attempt to contrast my position with RbR's was seen as belittling to the evidence and expertise behind his opinions, I apologize.

Second: I do not believe in Platonic guardians. I do not advocate finding the "right" people and letting them govern "rightly." I wish you would listen to me more closely. I really do believe in the vital necessity of checks and balances! (And I am happy to add that the discussions on this blog have really helped persuade me of that.) But I also think it is possible to establish a successful "non-partisan" committee-type rules-based process WITHIN that system--like a bubble kept afloat on a raging sea--that has been designed to produce a "neutral" solution to a specific problem to which a set of solutions generally recognized as reasonably "neutral" can be proposed. There are a lot of caveats in there, but it's still a fairly strong statement--and I think there is good evidence to support it (it's not just "independent reasoning"). I believe redistricting is precisely such a problem that should be approached in this manner.

Third: I fully agree that no system or process is perfect and that it will always evolve--usually deteriorate--over time. (For god's sake, my doctoral field was statistical physics. I grok entropy.) I'm really not some pie-in-the-sky, naive scientist who thinks he knows everything and thinks a little math is the solution to all problems on Earth. I am tired of being painted with that brush. When it comes to respecting each others' intelligence and education, it goes both ways.

Anonymous said...

I want to agree with USWest about the evolutionary nature of government, and to remind us all that the current US constitution (with its wildly unequal representation in the Senate, the filibuster, and the electoral college) would never be adopted today. Some of the great innovations of our system, judicial review, minority rights in the Senate, to name just two, could only, I suspect, come about through gradual evolution. I think that is one of the lessons of political science that needs to be better understood: democratic institutions cannot descend from the sky like a New Jersualem. They must be given space to change and grow. This is part of the problem in Iraq today.  

// posted by LTG

Anonymous said...

Dr. S, I agree that we can develop a better mouse trap, and along side of that, we have to have 2 things: Evaluative mechanisms in place to ensure that these systems remain as neutral as is humanly possible (and we know how weak we as humans are), and a means to adjust the system when it no longer suits us. I think we are lacking these two things in regards to redistricting.

I think what happens is that we cruise along with these systems in place and then we forget about them. Suddenly, we see it isn't working and we think our hands are tied. And indeed, they are. So long as the benefit of the nation is secondary to the interests of those who benefit from a broken system, progress will be impossible. I remind us that we have been locked in a tyranny of a twisted majority. And as has been pointed out, it the new majority will soon find itself in the minorty if it can't move forward. What worries me is that the Democrats, who always have to play the adult, will do well and still loose to the dirty tricks machine.

I respect what the sciences have to offer us, and Dr. S and I agree more than not. I never intended to leave the impression that I think scientists nor Dr. S are naive. In fact, what I'd like to figure out is how, when an illness has entered a system, it can be eradicated? How can we innoculate our government from the viruses that have over taken it as late? Or maybe we need to ask the pyschologist how your cure mental instability once it has taken hold? Ah, now these are the question that have plagued all types of scientsits for centuries.

I think Dr. S and I share a common idealism and belief in the better angels of our nature. We would like to believe that when people work together towards a common goal, that a problem can be solved, even if the resolution is temporary. And our only fault is in asuming that we all have a common goal. And thank goodness someone does posses this fault, for without that belief in the better angels of our nature, progress wouldn't be possible. That, my blog mates, I think is something Madison and Hamilton would both agree upon.

// posted by USWest

Anonymous said...

I've been away from the blog for a while and missed most of this exchange. I see that my blogmates have been anticipating my angry response to Quokka (a new pseudonym for an old visitor?). And they are right to think I would angry at Quokka's arrogant and unprovoked attack on my profession. But I will attempt to restrain myself as I have made my feelings clear about such attitudes by "hard" scientists in previous posts and I think LTG sufficiently addressed this problem for the time being.

On the more substantive aspect of this little debate: I agree with much of what US and LTG said but I have one quibble. I don't agree with US West that Hamilton and Madison would agree with her beliefs in the "better Angles of our nature."

Here is what they said in Federalist 51 (authorship is attributed sometimes to Madison and sometimes to Hamilton).

"It may be a reflection on human nature, that such devices should be necessary to control the abuses of government. But what is government itself, but the greatest of all reflections on human nature? If men were angels, no government would be necessary. If angels were to govern men, neither external nor internal controls on government would be necessary. In framing a government which is to be administered by men over men, the great difficulty lies in this: you must first enable the government to control the governed; and in the next place oblige it to control itself. A dependence on the people is, no doubt, the primary control on the government; but experience has taught mankind the necessity of auxiliary precautions."

And here is their proposed solution (note the absence of any reference to non-partisan guardians!)

"This policy of supplying, by opposite and rival interests, the defect of better motives, might be traced through the whole system of human affairs, private as well as public. We see it particularly displayed in all the subordinate distributions of power, where the constant aim is to divide and arrange the several offices in such a manner as that each may be a check on the other that the private interest of every individual may be a sentinel over the public rights. These inventions of prudence cannot be less requisite in the distribution of the supreme powers of the State."

The arguments in the Federalist Papers have since been put to more rigorous tests by 60 years of modern political science research.

I echo US West's view that something like evolution takes place in the development of political institutions. And LTG is right that understanding this should be (and is) a major focuse of political science going back to the Federalist Papers.

If we are to tackle this redistricting issue we have to first acknowlege that any and all redistricting arrangements that we could possibly adopt have serious policy biases built into them. And so chosing one process over the other is an act of de facto bias regardless of the rhetoric one uses when making the choice.

For example, I've said before that when people talk about fairly designed districts they are implicitly advocating districts with centrist biases (I can explain why again if people so request). Centrists would love that but voters on the left and right would not. And the voters on the left and right have preferences that are no less legitimate than those in the center. Why should centrists' views be given special status merely because they are served by districts with more appealing shapes on a map?  

// posted by RBR

Anonymous said...

What RBR points out is that the very nature of choice is that you are placin gone set of priorities over another. I work in the field of educational testing. And I know that the very nature of a test is disciminatory and exclusionary. By setting standards, you are excluding all those who fall outside the parameters of what you are testing and, depending on the type of test you are developing or administering, you are attempting to label or categorize the examinees for a givien reason, usually determined by policy makers.

As I have explained before, if the policy makers don't like the results, they will bear down on the developer to change the parameters or to alter the nature of the test in order to get results more pleasing to the eye, even if such a thing is counterproductive for their inital goal. (Wheww, that was a long sentence!) In a sense, I live redistricting every day. I wage a limited battle where I can to hold my parameters. But I won't win any wars.

RBR is correct to point out my use of the "better angels". Point taken.Yes, government does help us to hear our better angels. We have both better angels and worse angels. Some of us like to listen to our better ones while being aware that the worse ones exist.

// posted by USWest

Anonymous said...

USWest writes, "I think Dr. S and I share a common idealism and belief in the better angels of our nature." Yes. And I believe there are many Americans who hunger for the opportunity to serve in a meaningful way--and if given half a chance would do so. That "half a chance" may, I think, be provided by one of the "better mousetraps."

RbR writes that those who advocate "non-partisan" redistricting are "implicitly advocating districts with centrist biases." Could he explain more? In my earlier comment, my biggest caveat for a "non-partisan" process was that the problem under review had to have a set of solutions that were generally viewed as not unfairly favoring either party. I believe such solutions exist for redistricting, so in that case, isn't calling a lack of bias a "centrist bias" like calling atheism a religion?

Anonymous said...

"To me, calling a lack of bias a "centrist bias" is something like calling atheism a religion."

OK, I'll take that a request for explanation.

Here we come to the heart of Dr. S's (and Numbat/Quokka's) misunstanding. Dr. S seems to think that "centrist bias" is some kind of oxymoron. In this statement (and others) both he and Numbat/Quokka demonstrate their narrow, statistical understanding of the word bias. Bias has several definitions and the one used in statistics is not the one most appropriately applied to this situation.

In statistics, bias suggests that observations are being consistently distorted from the "true" value. It implies "error." By imposing that definition on normative political institutions Dr. S and Quokka/Numbat are implying (repeatedly and without fully realizing it) that some policies are objectively more correct than others.

Another subtext in most of what Dr S argues in these kind of debates is that there is some "true" policy out there that is best for everyone. Dr. S. denies that he thinks this but his misunderstanding of the relationship between bias and centrism exposes the root of the misundertanding. He's arguing that policies favoring centrists can't be biased by definition. He is confusing political centrism with the statistical concept of the mean.

Perhaps "predisposition" would be an easier term to understand "hard" scientists and others with lots of statistical skills but little experience in studying political behavior/institutions.

Electoral districts are going to produce representatives that have views similar to those of the median voter of that district. The problem Dr. S and others point to is that our current districts are designed in such a way that the local median point of view is significantly different from the national median point of view.

If electoral districts are all set up to be "normal" looking to appease Dr. S's athetic criteria for district shape. That would presumably be independent of ideological distributions within each district's population. Such a scheme would be more likely to set up more districts in which the most likely winner would be close in outlook to the median voters nation wide.

But this is not to say that the national median voter's point of view is objectively better than or more legitimate than anyone one else's point of view. If we set up institutions to ensure that only points of view tha approximate those of the national median voter we have introduced a predisposition (or bias) in favor of those policies.

I might be convinced that a bias in favor of centrist policies is preferable to a bias in favor of the far right or far left. But that's because I'm closer to the center than the far left or far right. It's not because centrist policies equal unbiased policies.  

// posted by RBR

Anonymous said...

RbR: my "request for explanation" was more transparent than you portrayed it to be. (I wrote, "Could [you] explain more?")

I wish you had not laced your explanation with little digs. I also wish you had spent more time explaining what you meant rather than trying to tell me what you thought I meant... since apparently you did not really understand me. Let me try again.

In an earlier comment, I noted that I only advocated "non-partisan" type processes when there was a specific problem, "to which a set of solutions generally recognized as reasonably 'neutral' can be proposed."

Then, in a subsequent comment, I referred to this again, to add emphasis: "my biggest caveat for a 'non-partisan' process was that the problem under review had to have a set of solutions that were generally viewed as not unfairly favoring either party."

Let me attempt to rephrase that more clearly, using some of the (better) language you provided in your last comment. For the purposes of my comments here and previous, I have defined an "unbiased" or "neutral" redistricting plan as one which, ceteris paribus, is neither designed to nor would be expected (statistically speaking) to elect a body whose composition (by party) was notably different from the underlying party preferences of the electorate. (Note we are talking about favoring political parties rather than ideologies.)

Individual districts would often favor one party or another but when considered together, such a "neutral" or "unbiased" set of districts (as I am using the terms) would be basically a wash as far as party dominance in the legislature was concerned. This is what I meant as unbiased, and I honestly think I was fairly clear about it, RbR. I am not saying that this is the "best" or "truest" scheme possible--but it is a decent definition of what we usually think of as a "level playing field" for political contests. It is a decent description of the kind of bias most people think of when they think of biased redistricting plans.

Is such a plan "centrist" biased? It depends... You wrote that, "Electoral districts are going to produce representatives that have views similar to those of the median voter of that district." But I thought that depended far more on the details of the primary election process. If the primary election process tends to force candidates to appeal to their base rather than reach across party lines, wouldn't we end up with ideologues on the left and right running against each other?

Anonymous said...

I should clarify. I am and was well aware that there are different ways of defining bias. That is why I was so careful to explain what I meant by neutral. RbR has explained now that a redistricting scheme not designed to favor the election of members of either party might be called "centrist," but I am not convinced that such a scheme actually favors the election of ideological centrists.

Anonymous said...

Primary methods matter and while there are variations in the local rules they tend to be fairly simliar in most districts. The biggest difference would be between those with open primaries and those without. But if these rules were more important that ideological distrubtion, the shape of the districts would not be so controversial.

So what about the distribution of ideological preferences?

Suppose we have a district with a single mode - a nice normal distrubtion or bell curve distribution. The median voter will be the modal voter. In that sitution both parties will have incentive to put up candidates close to that modal/median voter in order to maximize their vote share. But in some regions that median will be left of the national median and in other it will be right of the national median.

Now suppose we have a district with a bi-modal ideological distribution. But both modes are evenly matched in numbers. In that situation we would still have a gravitation to the median voter between the two modes.

Now suppose that we have a uniform distrubtion within a district. In this situation as well both parties will put up candiates that gravitate towards the median voter in the district.

When we have heavily gerrymandered districts, we often see bimodal distributions in which one mode has a substantial population advantage over the other. We can see this distribution within non-gerrymandered districts too (and in Senate seats). In these districts there will be little incentive to gravitate to the middle ground between the two modes because that middle ground may not contain the median voter any more. Indeed, the party representing the weaker of the two modes may have an incentive to put up candidates quite distant from their own intraparty median position in an attempt to be more competitive locally - the rise of pro-life, pro-gun Democrats in the South and the persistence of "liberal" Republicans in New England may be examples of this phenomenon.

OK, so we have all these local distributions. Suppose we have a bi-modal national distrubtion but most local areas have single modes or unbalanced bimodal distributions. Would it be "neutral" to design districts with uniform distributions? Should the shape of the district on the map be the primary consideration?

Of course there are numerous variables that can constrain this gravitation towards the median voter. But this general principle figures prominently in political science (and economics - see Hotelling). For example, this tendency to gravitate to the median voter would be especially strong if barriers to entry for new parties were high - as they are in this country (for social scientific work discussing this and related assertions see the works of Hotelling, Downs, Durverger, Riker etc).

My main point is that this is an extremely complicated problem. It is one with enormous policy and distributive implications regardless of the rules we choose. What's most important though is that it is far from obvious that one set of districting rules is ineherently more fair, neutral, unbiased, or legitimate than another.  

// posted by RBR

Anonymous said...

To me, the idea of "non-partisan" redistricting is to try to create a reasonably "level playing field" (when averaging over the whole map) for electoral contests between the two main parties. This is not how everyone sees it. Another widely-held idea is that the purpose of "non-partisan" redistricting is to create more compact, contiguous districts--or districts that align better with urban/county boundaries. That is a different notion than mine. I have tried to be clear about that, but it bears repeating.

If I understand RbR correctly--and please correct me if I am wrong--he writes that there is no obvious way to define such a thing as this "level playing field." He might be going so far as to say there is no way to define such a thing at all that would be generally (if grudgingly) accepted as such... but I think he stopped short of that. On the other hand, RbR might just be reminding me that no redistricting map or method is inherently more true, more fair, or more legitimate than any other... but I would hope by now RbR realizes I already understand that and require no refresher on relativism. RbR: do I understand you correctly here?

Now I know I am moving a little off topic here, but I wonder if RbR could explain a little more why something he asserted was true. He wrote that in a single district with a balanced bimodal distribution--or even with a uniform distribution--political parties would tend to put up candidates that gravitate toward the median voter's preferences. RbR: could you explain in more detail why that would be expected--or just point me to a web reference? Please assume I have given the question some thought already, and I don't yet see it. Is this one of those cases where history and experience serve to prove a point that is not easily amenable to what LTG calls, "independent reasoning"?

Anonymous said...

I would go so far as to say that there is no disctricting rule that would be generally accepted (if by "generally" we mean by some reasonable approximation of everyone in the electorate). But Dr. S. is on record as objecting to my justification for that expection: namely that everyone has policy preferences that influence their institutional preferences.

Now, as for gravitation to the median:

Imagine that every person in a district is a data point. Imagine that they arrange themselves on along a road from left to right. Imagine that their place on the road represents their policy preference on some important issue (or bundle of issues, say the tax rate or the level of funding for education). Where two people have the same position, they stand one in front of the other lining up perpendicular to the side of the road.

In both the balanced bi-modal case and the uniform distribution case if we have two parties (which we have in this country for all intents and purposes), they will strategically place themselves along that road such that they attract the maximum number of voters.

If the Left party adopts the position just left of the median voter, they will attract all the votes at that point they adopt and to their left. If the Right party adopts the position just right of the median voter they will attract all the votes at that point and to their right. But suppose the Right party adopts a position far to the right of the median voter. The Left party could increase their vote share by moving to a point between the median voter and the point adopted by the Right party. The Left party would then attract all the votes from their new adopted position and to their left, including the median. That group is by definition a substantial majority of the votes. The same arguement obviously applies if the Left party adopts a position far to the left of the median voter.

That applies in both the balanced bi-modal case and the uniform distribution case. It also applies in the normal distribution case.

If we weren't dealing with party organizations it would probably also apply in the unbalanced bi-modal case. However, because of the primary processes mentioned by Dr. S, parties may prefer to elect candidates that are closer to their intra-party modal voter than to the median voter. That is what may be a winning strategy in the disctrict may not be a winning strategy within the party. Which party wins in these cases depends on which mode is closest to the median (assuming a single political dimension, there is much justification for this assumption in the US case). In practice is most likely that the mode with the highest population will be closest to the median voter. In this situation, a party can afford to be constrained to it's own intra-party mode AND still win the district.

If we have a national distribution that is bi-modal in nature, and we impose districts that maximize the incentives to gravitate towards the national median we may be disenfrancishing two enormous blocs of voters and giving prefered status to voters with moderate opinions despite the fact these moderate opinions are neither the majority nor even the plurality opinion in the population as a whole.

That's about as good as I do off the top of my head. Here are some links (some are from wikipedia but I checked them out and they seem reasonable and are far easier to get at and read than the jstor articles I could have linked to).

Hotelling's Law  
Anthony Downs
Median Voter Theory
Durverger's Law
William H. Riker (not the one from Star Trek)


// posted by RBR

Anonymous said...

Thanks for the outline of the Median Voter Theory, RbR! I know that's just the tip of the iceberg--variations on the basic model could lead to different equilibria, where the two candidates are pushed toward polarization rather than centrism. In your experience with all this, RbR, do you believe that a real world district, with a fairly well (but not perfectly well) balanced bimodal distribution, would produce centrist candidates, even under the U.S. primary/general election system?

You wrote that there is, "no districting rule that would be generally accepted." Probably not, I suppose... at the very least, a substantial minority would probably object to any proposal. But are you saying we cannot even define what a "level playing field" would be, in terms of a district map? I mean, if I can show you a map that most political scientists would agree is distorted to favor the Republicans, and if I could also show you another map that most would agree has been skewed to favor Democrats... then surely there must be some map, or some set of maps, that most would agree fall somewhere in the middle ground between? And could we not then find some algorithm for generating or characterizing that set of maps? And could we not, through some careful process involving vigilance and oversight, ensure that this algorithm is followed when redistricting?

Whether such a "level playing field" is desirable is a separate argument. I have--and I accept RbR's chastisement for this--often assumed that such a thing must be desirable. But I realize now that is not a given. But if we can at least agree that such a thing exists, we can move on to the discussion of whether we want it, or whether we can have confidence that any given political process would actually achieve it.

Anonymous said...

"But are you saying we cannot even define what a "level playing field" would be, in terms of a district map? "

The point I have always tried to make is a response to this question. If we design a "playing field" by the rules that apply on it, then all rules have implications that favor one group over another.

When you say "level playing field" you keep drifting back to proposals that disfavor the positions on the ends of the spectrum and favor positions in the middle. That is not "level" or "neutral." It is however, moderate. It is an approach that seeks to constrain outcomes to a narrower, more moderate range. That's not a level playing field from the point of view of anyone but a moderate.

"In your experience with all this, RbR, do you believe that a real world district, with a fairly well (but not perfectly well) balanced bimodal distribution, would produce centrist candidates, even under the U.S. primary/general election system?"

It is possible and likely. But it would depend on the strategic situation within each party. It would depend particularly on the party members who chose the candidate prefering winning the election over any internal concerns. Often though, activists wish to retain control of the party even if it means losing the election. We heard rhetoric such as this from the far left of the Democratic party in 2000 and to a lesser extent in 2004. We hear it now from the far right of the Republican party.

However, once these internal struggles are settled somehow, winning the election becomes the main goal and candidates - even those than ran to the extremes during primaries - run as moderates in the general. If they don't their opponent may do so and beat them.

It was this kind of interal party strugle trumping general election concerns that enabled the Democrats in 2006 to run as centrists in so many Republican districts and win - even in gerrymandered districts.

In my own district there is a balanced bi-modal distribution. We have a couple of small cities with professionals and intellectuals who tend to vote for liberal candidates. We also have a large swath of rural America in which voters tend to the far right. Our candidates tend to run as moderates and make a big deal of their independence from party leaders (this claim is often false but it's not because of the district shape/make up).

I hope that clarifies my point of view. Thanks for asking.

// posted by RBR

Anonymous said...

In an earlier comment, Dr.S. stated:
"I have defined an "unbiased" or "neutral" redistricting plan as one which, ceteris paribus, is neither designed to nor would be expected (statistically speaking) to elect a body whose composition (by party) was notably different from the underlying party preferences of the electorate. (Note we are talking about favoring political parties rather than ideologies.)

The obvious way to do that is to have a perfectly proportional legislature with a single nationwide district. Thus, if the voters gave 47% to the Dems, then 47 seats to them. Of course, this highlights the fact that our system seeks geographically-based representation. While this is partly a function of history, it is also a function of preference: we want voters and representatives to have a more intimate and direct connection to one another. This means that we have a policy preference for districts that are "compact and contiguous," because such districts will increase voter-representative identification and connection. Drawing districts concerns more than just trying to, well, fudge proportional representation.

Also we prefer more ideological variation among representatives than PR would give us. For this reason, we allow various candidates to be chosen at the local, not national or even statewide level. As RBR notes, gerrymandering may have its biggest effect on determining who the median voter is in a given district. Sometimes it means a tradeoff between three ultra-right-wingers or five center-right moderates.

It is instructive that the Senate is wildly gerrymandered in favor of rural interests, but its party makeup is similar to the House.

What goals would we set for a nonpartisan redistricting board? What if they aimed for each district to have a median voter equivalent to the median voter of the whole state? That might produce a uniform  slate of moderate Dems or Republicans, rather than a proportional split.  

// posted by LTG

Anonymous said...

If I understand RbR correctly, he is saying that any redistricting map will favor some group of people. This is an interesting statement. I will mull it over.

LTG says my notion of a "level playing field" is just an attempt to minimize disproportionate representation within the single-member district system--and he reminds me there are other things to consider when drawing districts. Fair enough. Although I can't prove it, I believe it would be possible to find maps that minimize disproportionate representation while also meeting those other desirable criteria (e.g. compact and contiguous). In other words, these goals need not conflict. Does that make sense?

Anonymous said...

Actually the disproportionality comes from the plurality electoral system combined with single member districts. If you increase the number of representatives per district you can reduce but not eliminate disproportionality.

Even most PR systems have some disproportionality built into them. First, because very few use the single national electoral district scheme that LTG alludes to. Second, because most also impose minimum vote thresholds for representation to prevent the proliferation of tiny parties.

Finally, all PR systems rely very heavily on constitutional roles for political parties. Most are organized based on party lists. You don't vote for a representative you vote for the list of candidates put forward by your favorite party. The candiates are awarded seats according to the share of the vote the party gets and their rank on the list compiled by the party (some systems allow people to manipulate that ordering on their ballot, these are called "open list" PR systems).

It is worth pointing out that these systems involve a great deal of centralized power by the party leadership. The leadership makes up the list and determines which candidates get seats when votes are scarce. Loyal candidates go to the top of the list, mavericks are near the bottom.

So even in PR we have some of same problems in practice that motivate the cries for districting reform in the first place. 

// posted by RBR

Anonymous said...

Just because geography is traditional does not mean we should be wedded to it. Imagine an electoral system where we voted by age cohort, or by occupation, rather than by zip code. So, all nurses would vote for one rep, or all 25-year-olds. We could also just do it by letters of the alphabet (e.g., all persons with surnames ending in R get to vote for up to 3 reps...). Sound nuts? Is it more nuts than doing it by zip code or census tract, as we do now?

Consider this possibility: We could also just randomly divide the entire US population into 500 equal "districts" and have each elect a member. Ponder what that would accomplish.

// posted by LTG

Anonymous said...

I'm not sure but I think I remember something about the Irish upper house having seats allocated to professions (or people with professional status having weighted votes or something). I have a friend who specializes in Irish electoral politics, I'll ask him what I'm thinking of. 

// posted by RBR

Anonymous said...

RBR is onto something. Ireland has an electoral system for the upper house that will make your head hurt . The basics are that there are five industry/sector panels that nominate about 40 senators, voted on by existing Senators, lower house members, and city/county councils. All university graduates also vote on 6 senators.  

// posted by LTG

Anonymous said...

It is interesting that Ireland seems to have thought that the House of Lords was a good idea from an elite representation point of view but objected to its aristocratic character. 

// posted by RBR