Princeton Study Concludes US is an Oligarchy not a Democracy

[GostHacked]

http://en.wikipedia.org/wiki/Oligarchy

Oligarchy (from Greek ὀλιγαρχία (oligarkhía); from ὀλίγος (olígos), meaning "few", and ἄρχω (arkho), meaning "to rule or to command")[1][2][3] is a form of power structure in which power effectively rests with a small number of people. These people could be distinguished by royalty, wealth, family ties, education, corporate, or military control. Such states are often controlled by a few prominent families who typically pass their influence from one generation to the next, but inheritance is not a necessary condition for the application of this term.

The Rockefeller family fit the description of 'Oligarch'. Even the Bush family can be considered as Oligarch'.

[Michael Hardner]

In the meantime your own link underscores the most obvious effect of oligarchy.

Just because I offer criticism, it doesn't mean I disagree with the basic premise. This seems to me to be a hole that many people on here fall into.

[Michael Hardner]

Time-series is not exactly relevant when they're looking at today's government and how it flies in the face of what is conceived as traditional American values.

Well, exactly - is it actually a traditional American value or not ? That's a question I didn't think of, and proof of time as an indicator of relevance. BC will likely argue (maybe he already has) that 'twas always so but the question is - does the data bear that idea out or not ?

[cybercoma]

That's not a question they were exploring. They were looking at today, not change over time.

[eyeball]

Not sure where you get that from.

There was more than enough resources and natural capital to go around and new technology to boot. The latter will probably go towards making the world safe for oligarchy when the other things start getting scarcer.

[Michael Hardner]

Technology is very good for the investor class.

[GostHacked]

Technology is very good for the investor class.

Not good for the rest of us. High speed trading computers makes sure Wall Street has an advantage in every way possible.

[Michael Hardner]

Not good for the rest of us. High speed trading computers makes sure Wall Street has an advantage in every way possible.

It also destroys the livelihood of individual owners too. Press Baron isn't something I'd want to be today.

[eyeball]

Technology could also destroy the close intimacy that scumbag politicians and lobbyists enjoy in their ornately-panelled opulently-furnished smoke-filled back-rooms. But..."Oh the humanity!"...apparently.

[Michael Hardner]

But..."Oh the humanity!"...apparently.

I wish that I had an example of somebody calling this intimacy out, with an actual plan to address it. Do you ?

[GostHacked]

It also destroys the livelihood of individual owners too. Press Baron isn't something I'd want to be today.

Press Baron? Like Rupert Murdoch?

[Michael Hardner]

Press Baron? Like Rupert Murdoch?

Well, he's more of a media baron but maybe yes... Digital is killing whole industries now.

[eyeball]

I wish that I had an example of somebody calling this intimacy out, with an actual plan to address it. Do you ?

Yes I do.

[Michael Hardner]

Yes I do.

Ok, good to know. :|

[eyeball]

You should know by now.

[eyeball]

Just try imagining yourself how it would work...monitoring, auditing and validating...is it really that hard to get your head around it?

[Michael Hardner]

Just try imagining yourself how it would work...monitoring, auditing and validating...is it really that hard to get your head around it?

I thought you said it's happened already ? Why do I need to imagine it then ?

[eyeball]

To help yourself get your head around the concept. You have to sip the water into your mouth yourself.

[-1=e^ipi]

Okay, I shall comment on the paper now. Didn't do it earlier because I was lazy. There are a number of comments I could make about the paper, but I will just point out the important ones.

1. The way they 'test' the 'Majoritarian Electoral Democracy' theory is rather questionable. They claim that the 'Majoritarian Electoral Democracy' implies 'median voter theorem' for the sake of the study. 'Median voter theorem' means that the preferences of the median voter dictate policy decisions. But they admit:

"If citizens' preference orderings are not unidimensional and are sufficiently diverse,

majority rule – hence also two-party electoral competition – might not lead to any equilibrium outcome at all."

Political positions aren't close to unidimensional. Furthermore, the US democratic system (with things like the electoral college system) encourages a stagnant political system that makes it difficult for views contrary to the democrat-republican dichotomy to be represented. Nether multidimensionality nor political stagnancy where given serious consideration in the 'study'. With these issues, why should the median voter theorem hold?

The paper even goes a step further and assumes that median voter = voter at the 50th percentile in income. While this assumption is not that unreasonable, the correlation between income and political preference is not necessarily strong, let alone clear (see the two graphs below). They could have simply computed the median voter by saying that the median voter is the one that agrees with the majority for every single issue... the 50th percentile definition seems unnecessary.

http://test.causeweb.org/wiki/chance/images/7/72/Income-vs-Republican.png

http://test.causeweb.org/wiki/chance/images/5/50/Republican-Vote-Share.jpg

2. For the 1779 'political issues' from 1981-2002 that the study analysed, there were no tests made to examine if these issues were representative of the concerns of the population. For all we know, all the 'political issues' could have involved the legalization of marijuana. The authors even admit in their appendix that 387 of the 1779 questions asked 'essentially the same policy issue'. They could have classified issues into different categories (Economic, Cultural, Environment, Defence, etc.) and see how represented these categories are relative to how the public values these different categories (they could have used public opinion polls, issues discussed during elections, etc.). It is also worth mentioning that the data used is on average 20 years old, so any conclusions that are made are outdated by ~20 years.

3. Overall, I applaud the paper's attempt to test between various theories. Their methodology is simplistic, yet effective at testing the 4 hypotheses and easy to understand. It is clear that doing such tests is difficult and the data is difficult to obtain/scarce, but at least the authors of the paper put different theories to the test.

4. The results on table 2 indicate strong issues of multicollinearity in the data. In particular, the the opinions of the 50th percentile and the 90th percentile have a correlation coefficient of 0.78 (note that this is the 'corrected' R-squared value, the original R-squared value was 0.94, see appendix) and the correlation coefficient between all interest groups and business interest groups is 0.96. Admittedly, because all interest groups includes business interest groups, and because business interest groups and mass public interest groups is only -0.5, the second correlation might not be a big issue. But the multicollinearity between 'economic elites' and 'the median voter' indicate that any conclusions that distinguish between the preferences of the two groups will be rather questionable. Perhaps the study should have used something better than the preferences of the 90th percentile as a proxy for the preferences of political elites.

5. On table 3, the R-squared values are extremely low (between 0.028 and 0.074). I understand that the data quality might not be that good and that low-ish R-squared values are not unexpected in political science studies, but these R-squared values are really low... The models tested do not adequately describe the political preferences of society.

The most important result on table 3, according to the paper, is the lack of statistical significance of 'the preferences of average citizens' in Model 4. They use this lack of statistical significance to make their conclusions to reject 'Majoritarian Electoral Democracy'. But with the issues of multicollinearity, multidimensional political preferences, stagnant political system, etc. is this result unexpected or unreasonable? I don't think the results of the paper are sufficiently strong to reject 'Majoritarian Electoral Democracy' as the authors do.

6. No diagnostic tests were performed on the regression analysis to see if the various assumptions used on ordinary least squared regression hold (such as normality, homoskedasticity, etc.). Also, the model used clearly has specification error and lacks explanatory variables.

7. The error analysis was poorly done. The authors claim that "The failure of theories of Majoritarian Electoral Democracy is all the more striking because it goes against the likely effects of the limitations of our data.". But I just listed some reasons why the statistical significance of the regression coefficient of the median voter could be rejected without having to reject Majoritarian Electoral Democracy (points 1, 4, 5, 6). The political bias of the authors prevent them from seeing any source of error or issue with their study that doesn't help their conclusion (see point 8). Admittedly, using the 90th percentile of voters to represent economic elites might understate the importance of economic elites, and their omission of interest groups might understate the importance of interest groups.

8. There were a few cases of unjustified political opinion in the papers that make it unprofessional. While most of the paper was good, they constantly mentioned 'labour unions' as representative of public opinion and about half way through page 23 they go completely overboard with opinion. They use the paragraph beginning with 'we tend to doubt it' to dismiss the criticism that the results could be due to having an uninformed voting public.

They could have just wrote the paper, tested the model, and then made their conclusions about a rejection of Majoritarian Electoral Democracy (and it would have been a good paper). But for some reason the authors feel the need to put unsupported opinion at the end in order to claim that the rejection of 'Majoritarian Electoral Democracy' is necessarily problematic for society. It's completely unprofessional! Especially unsupported statements like "All in all, we believe that the public is likely to be a more certain guardian of its own interests than any feasible alternative".

Overall, I don't think the results of the paper are anywhere close to strong enough to reject 'Majoritarian Electoral Democracy'. The authors made their conclusions out of their clear political bias (which was unprofessional). If they did their job properly they should have concluded that the results, while interesting and useful for encouraging future studies, are insufficiently strong to reject any of the 4 models.

Many people in this thread seem to be taking the conclusions in the paper and running with (much further than the authors did) them to support their dogma regarding 'the economic elites control everything'. I don't give a damn that the paper was written by an ivy league university, that does not matter. Many aspects of the paper were not done as well as they could of and the evidence isn't strong enough to make the conclusions people are making in this thread.

[cybercoma]

To your points,

1) So you've identified an even bigger criticism of US politics that is actually worsened by the outcome of the paper.

4) They mentioned earlier in the paper when they talked about the wealthy elites that the 90th percentile was too generous and it would have similarities to the median group. They also identify correctly that this would mean any effects would be under stated when it comes to the political power of economic elites. If anything, this is tempering the results, not exaggerating them.

5)The R-squared show that there is unexplained variability, but that does not have anything to do with the significance of the results nor does it mean that there is no relationship between the independent variables and the outcomes.

6) Normality and homoskedasticity are assumptions. Most published papers that are not in econometrics do not include discussions of these assumptions and the analysis because it's a given with this type of modelling.

7) Now you're introducing political bias, which assumes you have access to the beliefs of the authors. You don't and it's not demonstrated in any meaningful way here. This study wasn't looking at a comprehensive model to predict the outcome. It was looking at the relationship between specific predictors and that outcome. Is there some other predictors that co-vary with political will and policy, most likely. So there is potentially an argument that something is in the error term that would erase the effect of political wills, but that would need to be demonstrated beyond what we're going to get into on a message board.

8) Of course the paper talks about opinion. It's about public opinions. Of course they're going to look at labour unions, since they stand in contrast to corporate/business interest groups. I don't get this objection at all. It's like you don't actually understand what it was that the paper set out to do.

You say many aspects of this paper weren't done as well as they could have been, but you're criticizing it against things that the paper did not set out to do. It sought to determine how well government policy aligned with the interests of various, competing interests in society. It wasn't looking for a comprehensive model to predict how government policy is made. It was testing the hypothesis that government policy aligns more closely with the interests of the wealthy, even when those interests are directly opposed to the interests of the majority of voters. The results are clear. The political will of the average voter has no statistically significant relationship with what policies are actually made, whilst the economic elite does. There are several limitations to the study that they've outlined. And you're absolutely right to wonder about some other co-varying influence on both public opinions and policy outcomes. However, the question is about competing public interests and which ones the government appeases with its policies. It's clearly not the majority of voters that win out.

[-1=e^ipi]

Cybercoma

@ 1) Maybe, but the point was mostly that the way they define and test 'Majoritarian Electoral Democracy' is extremely questionable. Median voter theorem is not necessarily true under multidimensional politics. Stagnant political systems can prevent median voter theorem even under unidimensional politics. Median voter isn't the same as the voter at the 50% percentile of income.

@ 4) Your 'rebuttal' to 4 has nothing to do with multicollinearity...

Maybe this will help you understand: http://en.wikipedia.org/wiki/Multicollinearity

@ 5) I never made any claim about the results not being statistically significant. But if you have a very low R-squared, it is a good indicator that either to model is bad or the data is bad (or both in this case). Also... why didn't they do the fisher test? That is pretty standard.

@ 6) It isn't given... If the assumptions are strongly violated then the results of the tests lose credibility. Yet they do no tests.

@ 7) Their bias is clear from the text and choice of words. It is clearly demonstrated and I gave examples. Their statement "it goes against the likely effects of the limitations of our data." is false/unjustified, period.

@ 8) It isn't just public opinion. They use the word 'we' several times. It's one thing to talk about public opinion in a neutral manner. It is another thing to state political opinion without support. I'm surprised the referee didn't ask them to remove some of the paragraphs in the conclusion. It is unprofessional.

For your last paragraph, as I previous explained, I do not think that the results of the paper are sufficiently strong to justify your conclusions.

Edited April 23, 2014 by -1=e^ipi

[cybercoma]

Again, you're getting hung up on arguing that their model doesn't explain enough of the variation in the outcome, when that's not what this particular research was designed to do. You're coming at this from an economics mindset and that's not what this research is.

[-1=e^ipi]

Again, you're getting hung up on arguing that their model doesn't explain enough of the variation in the outcome, when that's not what this particular research was designed to do. You're coming at this from an economics mindset and that's not what this research is.

Low R-squared value is only one of the 8 points I made.

I'm also not sure why one's approach to regressions should greatly change when looking at a different field of study.

[cybercoma]

Low R-squared value is only one of the 8 points I made.

I'm also not sure why one's approach to regressions should greatly change when looking at a different field of study.

Because it depends on what the study sets out to do. If you're looking at predicting a particular outcome, then probably want to explain as much variability in that outcome as possible. If your goal is to explore whether there is a relationship between things or what the difference between predictors is, then it's not necessary to explain all of the variability in the outcome. There may, as I mentioned earlier and agreeing with some of your statements, may be an issue with endogeneity of one of the regressors. You also mentioned multicolinearity, but the authors addressed that when they discussed the limitations of the research. The economic elites are the top 10%, but a portion of that could have more in common with the middle income group. You're right that this could introduce multicolinearity, but the effect would be understating the effect that the top income group has on the outcome. In other words, the top income group may actually have a larger impact than is being reported here, but almost certainly not a smaller impact. Anyway, the report is not without limitations, but I think they've done well to outline what those are. Your criticisms mostly seem to be holding the research to things that they did not set out to accomplish and in any case do not necessarily contradict the numbers presented. In fact, the multicolinearity issue, if addressed by changing it from top 10% to top 5% or top 1%, would have actually shown are larger marginal effect, rather than smaller.

[eyeball]

Your criticisms mostly seem to be holding the research to things that they did not set out to accomplish and in any case do not necessarily contradict the numbers presented.

The old rubeum allec trick.

Edited April 24, 2014 by Charles Anthony
Image removed - not part of the argument.