Why is American Politics Uni-dimensional?
Paul Krugman briefly mentions a somewhat intriguing puzzle. Most people, if you ask them about it, would say that political beliefs are “multidimensional.” We often think of a simple 2-dimensional models like the Nolan Chart in which people should be sorted along both a left-right axis about economics, and then along a second axis about social/cultural issues like gay rights. But as Krugman observes, Congress doesn’t work this way:
That’s what I would have thought a few years ago. But then I became familiar with the Poole-Rosenthal work on Congressional voting. They use a clever algorithm to jointly map bills and members of Congress in a hypothetical issues space. The number of dimensions in that space is arbitrary — but they found that historically just two dimensions accounted for the great bulk of voting. One dimension corresponded to left-right on economic issues; the other was basically race/segregation.
And since the 1960s, with the great Southern realignment, the race dimension has collapsed. So Congressional politics is left versus right — end of story. Oh, and polarization along that dimension has increased hugely: the center did not hold, and there really isn’t any middle ground.
To offer some qualitative examples, Susan Collins and Olympia Snowe are pro-choice Republican Senators. But they’re also the two senators who seem like they might possibly vote for a national health care bill. Rather than representing some kind of ideal type of upscale northeasterner who’s socially liberal but economically conservative, they’re less conservative across-the-board than their colleagues from the South and the Mountains. Conversely, when you stroll down to Arkansas’ Democratic Senators, you don’t see cultural conservatives with populist economics, they’re just more conservative across-the-board than their coastal colleagues.
Something to consider along these lines is the map of states with anti-union “right to work” laws that substantially prevent union organizing:
This is about as purely an economic issue as you can think of, but it’s the liberal coastal elites who have union-friendly laws. My guess is that this winds up creating a higher degree of unidimensionality among practical politicians than exists in the population. The presence of a meaningful labor movement on the ground has huge implications for how politicians will vote on economic issues. And unions tend to be strong in the same places where people have relatively left-wing views on cultural issues.
49 Responses to “Why is American Politics Uni-dimensional?”
Register
RSSLogin
Topic Cloud
ArchivesBlog RollAbout Matt YglesiasContact Me
Donate
August 3rd, 2009 at 8:47 am
Right to work laws don’t jsut have an effect on the political incentives faced by politicians. They also affect the kind of messages that people come in contact with. Having lived in Idaho and now moved to Ohio, its really hard to describe the difference in local media and local ads. Also Walmarts in Ohio have cleaner floors.
August 3rd, 2009 at 8:57 am
And unions tend to be strong in the same places where people have relatively left-wing views on cultural issues.
Or alternatively, the blue collar jobs (held by cultural conservatives) have been moving out of the strong union states, leaving only the cultural liberals.
August 3rd, 2009 at 9:03 am
All this is true, but why it is true isn’t exactly a puzzle. The “Southern Strategy” was deliberately designed to bring about this alignment of “conservative” economic interests with resentment-based politics.
And there have been barrels of ink (now megacoulombs of electrons) spilt regarding why certain U.S. populations are willing to vote contra their economic interests in favor of their resentments. But suffice it to say that is an empirical fact, and the GOP made use of it to craft their strategy for the last 40 years.
August 3rd, 2009 at 9:04 am
I suspect the right-to-work laws are an epiphenomenon of tribalism/partisanship. People strive to discover and adopt the values of their party. In the U.S. there are, and really can only be, two parties, so all political divisions are squashed into a single dimension. This seems like a testable hypothesis but I’m no political scientist.
August 3rd, 2009 at 9:23 am
Re: And unions tend to be strong in the same places where people have relatively left-wing views on cultural issues.
Huh? What about the Rust Belt states?
Though it’s perhaps fair to say that unions are _no longer_ terribly strong in much of the Rust Belt.
Re: Conversely, when you stroll down to Arkansas’ Democratic Senators, you don’t see cultural conservatives with populist economics, they’re just more conservative across-the-board than their coastal colleagues
There are a few culturally conservative Democrats in the coastal states- Bob Casey, James Langevin and Stephen Lynch (from, believe it or not, Boston) come to mind.
August 3rd, 2009 at 9:24 am
I don’t understand the term, “coastal elite.” It clearly doesn’t include Virginia, Texas or any state in between. But there’s a lot of coast there!
August 3rd, 2009 at 9:37 am
“Right to Work” anti-union how, exactly? When I was a teacher in NY State, I would have loved to have the right not to join the union. It’s not as if they were doing anything for me anyway; pay was strictly seniority based, with no real input from peers or management, and I was nearly caught up in a strike action I disagreed with (but that would have been physically unsafe for me not to join).
Closed shop laws are at least as, if not more coercive than “right to work” laws.
August 3rd, 2009 at 9:46 am
In the U.S. there are, and really can only be, two parties, so all political divisions are squashed into a single dimension. This seems like a testable hypothesis but I’m no political scientist.
This issue is what people are referring to when they cite Poole-Rosenthal. The basic setup is that for a long period prior to Goldwater and Nixon, the parties were basically aligned around economic issues, but there was also a South versus non-South split organized around race and other resentment-based issues. This split was mostly internal to the Democratic Party, since they had a virtual lock on the South.
So it actually wasn’t the case the two-party system caused a complete squashing down to one dimension. Rather, that took a deliberate realignment strategy in which the GOP elites agreed to back certain populations on their resentment-based policies (at least on the state and local level), provided those populations supported the elites on their economic issues.
August 3rd, 2009 at 9:57 am
Huh? What about the Rust Belt states?
This isn’t really a counterexample because the “Rust Belt” region isn’t particularly “culturally conservative”–more so maybe than the upper coasts, but much less so than the South.
There are a few culturally conservative Democrats in the coastal states- Bob Casey . . . .
Pennsylvania is only part coastal in a political sense. Moreover, while Casey is pro-life and pro-gun, he otherwise isn’t particularly “culturally conservative”.
August 3rd, 2009 at 10:06 am
James Robertson,
I think you answered your own question as to how RTW laws are anti-union. They are anti-union because they make it harder for unions to form and then to function. Matt said RTW laws were anti-union, not anti-whatever-James-Robertson-happens-to-want.
August 3rd, 2009 at 10:07 am
@James Robertson:
“Right to Work” means right to freeload. In other words it allows workers to enjoy the fruits of collective bargaining without paying for its cost. Depending on whether your local was affiliated with AFT or NEA your dues paid for things like labor relations consultants who advised your local on contract negotiations, research departments that aided your local in research with respect to comparing your contract to others as well as an analysis of local school finances. Additionally the contract can be enforced through a grievance process that is guided by labor lawyers paid for by dues money. All of these services ensure the integrity of collective bargaining and cost money.
As for the contract you worked under, surely your peers and management had input at the bargaining table as to how pay is structured. The fact that you disagreed individually was outweighed by a majority vote of your peers. That is the essence of collective bargaining; an individual affiliates with a group and is more powerful from it.
Try teaching in the non-union South and see how well bargaining your own pay goes.
August 3rd, 2009 at 10:09 am
Well, considering that the “war on drugs” has actually been a war on black people (and the evidence on that point is overwhelming), saying that “the race dimension has collapsed” is pretty much to say that Congress has acted in total unison, or, IOW, our so-called liberals have actually been segregation-by-other-means enablers.
And we’re paying the price.
August 3rd, 2009 at 10:13 am
In America, racial politics are deeply infused in economic and distributional politics. I’m not sure how one goes about trying to disaggregate these variables.
August 3rd, 2009 at 10:17 am
Isn’t this just pushing the question back further? Why are those states the ones with right to work laws? Since 1947, all states have been able to pass right to work laws, but only some states did. Why those states?
August 3rd, 2009 at 10:18 am
Sluggo – I’ve had to negotiate for my own pay ever since I left teaching, and believe me, it’s gone way better. Collective bargaining tends to reduce everyone to the mean – it neither punishes nor rewards.
August 3rd, 2009 at 10:28 am
@DTM
Again, my political science is what I’ve acquired by reading blogs such as this one, but isn’t the period of bi-partisan co-operation you’re talking about generally considered anomalous? If that isn’t the case, there certainly are a lot of folks using this anomaly theory to critique the Villagers’ bipartisan fixation.
The test I have in mind would be to compare the dimensionality of variation in the aggregate of two-party systems to multi-party systems. You’d need a large enough sample to dampen the effects of historical accidents such as the Civil War.
August 3rd, 2009 at 10:32 am
@James Robertson:
I’m sure you’ve done better because you have demonstrable talent and accomplishment in whatever field you’ve entered.
Teaching is a much broader based labor market where it is easier for school boards to use their power to dictate terms of agreement. The only tangible way of demonstrating worth is through relative scarcity in the labor market.
Your point about reducing to the mean has merit; arguably math, science, foreign language and special ed are subsidizing K-3 elementary teachers because of the huge differential in supply of both specialties. On the whole, though, collective bargaining is the tide that lifts all boats. Comparing the salaries on the map in MY’s post will confirm that.
August 3rd, 2009 at 10:52 am
@John :
I recommend Albion’s Seed, by David Hackett Fischer, for an insightful view of cultural evolution in American regions.
August 3rd, 2009 at 11:12 am
If I remember my John Aldrich correctly, parties are theorized to play just this role in Why Parties. The problem lies in this: in a multidimensional system, you cannot find a single position to defeat all others. This is problematic for the type of party theory which states that they will move to the position that will win them the most votes; when you have multiple dimensions you can’t do that. That’s the kind of theoretical answer and I may be mis-remembering it slightly.
August 3rd, 2009 at 11:12 am
American politics is unidimensional because we have a two-party system, and the parties organize the conflict on a number of dimensions into one dimension. The Poole-Rosenthal scores CANNOT DISTINGUISH between ideological and non-ideological votes. So the scores cannot distinguish between party politics and “left-right” disagreements.
It’s sad to see that even Krugman is under the sway of Poole and Rosenthal.
August 3rd, 2009 at 11:13 am
The final line to that should read: And this is why parties attempt to restrict things to a single dimension.
August 3rd, 2009 at 11:15 am
@John, @kme: I think there’s also some reversal of cause-and-effect here. Yes, unions are stronger in non-RTW states. However, RTW states are largely those that did not have significant manufacturing bases prior to Taft-Hartley, and hence did not have strong existing unions to beat back RTW laws when they were proposed and passed. I suspect the causality goes something like: weak unions -> RTW laws -> no unions. RTW laws largely calcified the union structure that already existed in the 40s.
August 3rd, 2009 at 11:22 am
Again, my political science is what I’ve acquired by reading blogs such as this one, but isn’t the period of bi-partisan co-operation you’re talking about generally considered anomalous?
My understanding is that for most of U.S. history, there has tended to be some sort of major split in at least one of the two major parties at any given time, very often involving race in some way.
The test I have in mind would be to compare the dimensionality of variation in the aggregate of two-party systems to multi-party systems.
My guess is that two-party systems would indeed average fewer dimensions than multi-party systems. So I would think the question is whether it actually gets as far down as averaging close to one dimension in two-party systems.
August 3rd, 2009 at 11:25 am
Krugman is right. There is no middle ground. The middle ground is now a bottomless chasm. Head for the middle ground and you are heading toward an abyss. Toward your doom.
So Obama should think twice about this bipartisan thingy he feels so strongly about. I don’t think Obama realizes he is trying to bridge the unbridgeable abyss. Reaching out to the other side, he could easily tumble over the edge, and if he does, he takes me with him. Which I am not happy about. I have no desire for doom. I am still young. At heart.
August 3rd, 2009 at 11:32 am
Looks to me like the mudsill theory is at least a partial explanation.
The past is never dead, it isn’t even past.
You’d think wherever these nebulous ‘elites’ live would be exactly where laws predicated on a belief that the mass of mankind has been born with saddles on their backs, and a favored few booted and spurred had been born ready to ride would be stronger, not weaker.
Unless ‘egalitarian’ is somehow synonymous with ‘elite’. Which would be odd.
But then, the Republican party in this country isn’t very ‘republican’ in a dictionary sense.
August 3rd, 2009 at 11:33 am
I’d say it’s mainly a consequence of the two-party system, which is itself mainly a consequence of every election being winner-take-all.
August 3rd, 2009 at 11:48 am
A few points here on an interesting topic.
1. It’s probably more accurate to say that in states where the Northern Democratic Party was strong in the 1940s – which eerily maps onto the 2008 electoral map – given that unions were a major constitutive part of the New Deal Coalition, states did not pass so-called right to work laws against them. Historically speaking, such laws were an attempt by the South to punish the CIO for Operation Dixie and its support of civil rights.
2. BP Beckley – those states above still have higher unionization rates regardless of what happened to industrial employment. New York City, for example, which has lost virtually all of its industrial base, has a unionization rate of 25%.
3. I think Matt is ignoring the extent to which his map might indicate places where people have left of center ECONOMIC views. Not everything’s the culture war.
4. James Robertson – you happen to be the exception that proves the rule. If we look at this statistically versus anecdotally, the union membership premium tends to run about 21-32%. And (anecdotally) speaking as someone working under a union contract in a workplace that was unionized about 8 years ago, the difference is enormous, as literally every person who used to work in the non-union days has told me.
August 3rd, 2009 at 11:49 am
As for # of parties–my understanding is that the Poole-Rosenthal method doesn’t work for non-partisan legislatures–like the Confederate legislature–or for periods in US history when, for whatever reason, there was not two-party competition–the Era of Good Feelings being the prime example, IIRC.
August 3rd, 2009 at 12:03 pm
The conventional wisdom on the Southern Realignment is that the New Deal created a split along economic lines, and it wasn’t until Nixon’s explicit embrace of the “southern strategy” that the issue of race resolved along similar partisan divisions.
Recently published research (subscription required for full text access) argues that this conventional wisdom is a bit off. Basically, Feinstein and Schickler argue that the current partisan coalitions (racial and economic liberals vs. conservatives) were implicit in the New Deal economic ideological cleavage, and developed over time through the various state party platforms. Before the full-blown civil rights movement emerged, there were two main areas of racial liberalization in public policy: equality of employment opportunities and fair housing legislation. Both of these sets of policies empowered the state to override the preferences of property holders (capital in employment and real estate in housing). This was anathema to Republican economic ideology, and thus we get a coalition aligned against racial equality policy by virtue of its support for economic conservatism.
Labor unions come into the picture more explicitly as follows: one of the main drivers of racism, many scholars argue, derives from the “split labor market” theory. In order to weaken the political power of labor, management promoted the division of the workforce along racial lines. By treating blacks and whites as separate (but not necessarily equal), employers were able to sow resentment within the labor class and prevent the coalescence of a unified movement. The strategy is divide and conquer (you can read a heavy dose of neo-Marxianism into it, but that’s not entirely necessary).
Granted, the heights of both of these processes are about a generation in the past, so one might not expect the divisions to remain prominent (and indeed, it appears as though the polarization continues). But these dynamics have a way of perpetuating themselves through various forms of path dependence.
August 3rd, 2009 at 12:04 pm
Our “first past the post” election system so fossilizes the two party system that its impossible to imagine a peaceful transition of powert to a new political party (the last time a new major party rose to power, the immediate result was Civil War).
If we had a multiparty election system, many Catholic politicians like Bob Casey and Protestants like Jim Webb would run as “Christian Democrats” (certainly a US version of this European idea would have a more religiously inclusive name).
http://en.wikipedia.org/wiki/Christian_democracy
August 3rd, 2009 at 12:08 pm
On another note: the Poole and Rosenthal stuff is interesting in understanding how Congressional voting operates with respect to party, but I’m not completely satisfied with considering it a perfect representation of ideological dimensionality. A lot of what gets picked up is party discipline, and the outcome of intra-party log-rolling instead of real ideology. I suspect, for example, that you will see Specter (D-PA) move leftward subsequent to his party-switch. If it were really measuring ideology, whatever leftward movement would have occurred before his switch.
August 3rd, 2009 at 12:18 pm
So Obama should think twice about this bipartisan thingy he feels so strongly about.
Yes, he feels so strongly about bipartisanship that his congressional allies have regularly been passing the important items on his legislative agenda with little or no Republican support.
The truth is that Obama has wisely branded himself as someone who is not hyperpartisan and is willing to work with the Republicans. That is just good politics. However, in truth he and his congressional allies have been unwilling to give up anything more than what is strictly necessary to get his legislation passed.
And by the way, this combination is actually driving Republicans crazy.
August 3rd, 2009 at 12:21 pm
@30
We don’t, in fact, have a first-past-the-post system. I know a lot of political scientists who say that we do, and they should know better. Such a system would have a threshold vote-share as a victory condition. What we have is a plurality system, which is different.
What makes the two party system so strong is the winner-take-all single-member district system for all offices besides the Presidency (which is winner take all and single-member, but not district based). Duverger’s Law argues that these kinds of elections favor two-party systems (at least within the district) because supporters of third place candidates will strategically switch their votes to one of the first two candidates in order to more likely influence the outcome. For a long time in this country, we had a much more diverse collection of local and regional political parties, but each district could still be characterized as largely two-party. The emergence of unified national parties is primarily a story of political professionalization and reaping the benefits of inter-regional intra-party coordination.
August 3rd, 2009 at 1:25 pm
DTM,
When I said ‘conservative’, that does require some qualification. The state of Michigan for example has a fair amount of anti-abortion sentiment (which I disagree with) and anti-gay marriage sentiment (which I disagree with). Sadly it also has its share of racial conflict. Certainly a Democrat like Bob Casey, Bart Stupak or James Langevin is not ‘conservative’ in a general or absolute sense, particularly compared to the South. However, they are too conservative (on abortion, basically) to win the Democratic nomination, which counts for a lot.
August 3rd, 2009 at 1:26 pm
Correct that: I _agree_ with anti-choice sentiment and _disagree_ with anti-gay marriage sentiment.
August 3rd, 2009 at 1:27 pm
[...] was musing yesterday about why it is that a single crude left-right axis does such a good job of modeling [...]
August 3rd, 2009 at 2:02 pm
Hector,
I think you could plausibly claim Casey is somewhat “culturally conservative” for a Democrat, and similarly that the Great Lakes states are somewhat “culturally conservative” among reliable Democratic states. But that still puts them on the “relatively left-wing” side of the broader American divide, at least enough so as to confirm the correlation with the RTW map Matt presented.
August 3rd, 2009 at 2:42 pm
DTM: …and similarly that the Great Lakes states are somewhat “culturally conservative” among reliable Democratic states..
Ohio, at least, is not reliably Democratic. I don’t know about any of the other Great Lakes states.
August 3rd, 2009 at 3:11 pm
Ohio, at least, is not reliably Democratic.
Admittedly I’m speaking in a sort of political shorthand: Ohio, Indiana, and Illinois all have portions which are not part of the Great Lakes region, and therefore are not reliably Democratic to that extent.
August 3rd, 2009 at 4:32 pm
“it’s the liberal coastal elites who have union-friendly laws”
Uh Matt? What do you call states that border the Gulf of Mexico, or the Atlantic Ocean below the Mason-Dixon line?
August 3rd, 2009 at 4:36 pm
I did scim a paper with to many equatations to fully understand it last weak that made a socialpsychological case why nationalism and non redistributive attitudes are linked.
Anyway, a voteing system like the US one simply just supports two parties. So idenpendent of links in peoples attitudes towards nationalism and social redistribution or any of those other scales, the politicanas simply have to act according to their artificial alliance.
August 3rd, 2009 at 6:50 pm
Re: Huh? What about the Rust Belt states?
Hector, have you been to the Rust Belt states? At least in Michigan and Ohio a lot of blue collar folk are pretty liberal about social issues, even if they would not use the word “liberal”. They drink like fish (and like their pot too), cuss the air blue, cohabit without benefit of clergy, and generally believe in
“Live and let live”. It’s usually the middle class where you find the cultural conservatives– people who have enough time and money to worry about what their neighbors are doing behind drawn blinds, rather than where their next rent check is coming from.
Re: Ohio, at least, is not reliably Democratic.
Northeast Ohio (Rust Belt Ohio) is very Democratic, with extensions west to Toledo and southwest to Columbus. A lot of the rural areas of the state, and the Cincinnati metro area, are Republican. Dixie really begins at I-70 in Ohio.
August 3rd, 2009 at 6:55 pm
Uh Matt? What do you call states that border the Gulf of Mexico, or the Atlantic Ocean below the Mason-Dixon line?
Easy, you call it Dixie.
As Faulkner pointed out in Absalom, Absalom, “you can’t understand it. You have to have been born there.”
For as Ambrose Bierce described, there is a certain issue that is the greased pig of American politics. You ought to substitute Republican for Democrat these days in the Devil’s Dictionary entry, but the South still holds as an article of faith that setting n = the White Man yields an unsatisfactory conclusion.
This, of course, results in a, shall we say, unique viewpoint.
August 3rd, 2009 at 9:04 pm
Re: Hector, have you been to the Rust Belt states? At least in Michigan and Ohio a lot of blue collar folk are pretty liberal about social issues, even if they would not use the word “liberal”.
JonF,
I live in one of those states now, actually. I confess that most of the people I know here are middle-class, and I tend to hang out with a lot of fairly religious people, so that may color my opinion somewhat.
Again, I wouldn’t say this is a conservative place compared to the South, but it is more conservative than Massachusetts.
August 3rd, 2009 at 9:49 pm
What came first? The chicken or the egg?
August 3rd, 2009 at 11:24 pm
[...] highlight the fact that ideological conflict in the US maps fairly clearly onto a uni-dimensional left-right spectrum (although it is true that Congressional leaders keep off the agenda issues that might challenge [...]
August 3rd, 2009 at 11:39 pm
I’m surprised by the repeated reference to “America” here, as if its current position is unusual in some way.
Politics in all developed countries is unidimensional, with the economic left-right axis dominant, socially liberal attitudes highly correlated and racial attitudes highly correlated with both. The only difference is that, for historical reasons, the US racist right was in the “wrong” party until Nixon sorted things out.
August 4th, 2009 at 11:34 am
The notion of unidimensionality seems right on to me, although I can think of a very interesting exception to the rule.
Maryland’s first congressional district had a GOP representative who defied the axis from ‘91 until ‘09. He was a GOP orthodox fiscal conservative and pro-gun but took a very stance towards environmentalism (which is important to MD-1) and was socially liberal in a socially conservative district, co-sponsoring repeal of don’t ask don’t tell for example. While he supported the invasion of Iraq, he had the same concerns and agenda later on as many orthodox democrats like Clinton. He was judged to be in the middle of the political axis but he wasn’t really a moderate, he was quite liberal in some regards and quite conservative in others.
What became of this guy? Well, in 2009, Wayne Gilchrest (R-MD1) lost a primary challenge to Andy Harris, a completely conventional conservative who fits onto a one dimensional axis just fine. And here, I think, is the reason for the one dimensionality. Primaries favor candidates who fit onto the one dimensional axis. And without winning the primary, it’s darn hard to win the general election.
August 5th, 2009 at 5:56 pm
Uni-dimensionality is interesting in the context of Arrows Impossibility Theorem (see wikipedia). The important parts of which are:
Voting systems in general are unfair.
Voting systems that select from a one dimensional quantity are fair under majority rule.
Thus, while the election of US politicians is limited to a one dimensional scale (right vs left), it is a fair selection. (Majority rule produces a candidate that optimally satisfies voters as a whole)