Monday, May 03, 2010

Want democracy? Get a figurehead king

Fuzzy Queen of Couchland by Nibby Nebbulous from flickr (CC-BY)

Followup to: How much democracy in your democracy?.

I find the very concept of monarchy obscene - its premise is that some woman has a magical vagina, and whatever first comes out of said vagina is blessed with magic powers that somehow make it most well suited to rule a country. Alternatively, some man has a magical penis - but monarchies are older than DNA testing so while the theory often emphasized penises, the practice was mostly vaginal.

In either case, this is just so ridiculous, I cannot imagine how fucked up must someone brain be before they start treating it seriously, and yet they did! Some still do. But as Karl Marx said, being determines consciousness - and nobody ever came up with an idea so ridiculous, that some privileged people couldn't use it as an excuse for existence of their privileges. Cosmic Jewish Zombie? Magical Vaginas? Social Darwinism? Efficient Market Hypothesis? Anything goes, as long as it implies the privileged stay privileged of course.


Anyway, in spite of my repulsion with the concept of monarchy, I decided to run some data analysis on effects different systems of governments have on levels of freedom and democracy.

Data comes from Wikipedia - government systems, Democracy Index (by The Economist), and Freedom in the World Index (Freedom House, essentially CIA).

Monarchies are more free than republics

It doesn't take a sophisticated statistical analysis to see that monarchies are more free than republics. Here's Democracy Index, higher is better:
  • All countries: 5.6 +- 2.22 [n=163]
  • Monarchies: 6.37 +- 2.63 [n=28]
  • Republics: 5.44 +- 2.08 [n=135]
That's over 1/3 of standard deviation - about as much as the difference between Canada and USA - decent different, but maybe not worth writing a blog post about yet.

Just to verify, let's check Freedom in the World Civil Liberties and Political Right indexes (for both lower is better). Here's Civil Liberties:
  • All countries: 3.19 +- 1.81 [n=192]
  • Monarchies: 2.56 +- 1.7 [n=43]
  • Republics: 3.38 +- 1.8 [n=149]
And Political Rights:

  • All countries: 3.35 +- 2.13 [n=192]
  • Monarchies: 2.88 +- 2.15 [n=43]
  • Republics: 3.49 +- 2.1 [n=149]
So CIA and The Economist agree - that's again about 1/3 of standard deviation. Magical vaginas are not that bad.

IMG_5513 by Dave Malkoff from flickr (CC-NC-SA)

It gets better

The results so far were not terribly interesting, so I tried something else - how important is having ceremonial head of state versus one with actual executive power? It's rather shocking (higher better):
  • Ceremonial: 7.47 +- 1.42 [n=54]
  • Executive: 4.67 +- 1.93 [n=109]
Civil Liberties (lower better):
  • Ceremonial: 2.01 +- 1.28 [n=70]
  • Executive: 3.87 +- 1.72 [n=122]
Political Rights (lower better):
  • Ceremonial: 1.96 +- 1.37 [n=70]
  • Executive: 4.16 +- 2.07 [n=122]

That's 1.25 σ for democracy, and a bit over 1 σ for the other two indices!
 
Combining two criteria gets expected results - figurehead kings > figurehead presidents > executive presidents > executive kings. Democracy Index (higher better):
  • Ceremonial Monarchies: 8.12 +- 1.41 [n=18]
  • Ceremonial Republics: 7.15 +- 1.32 [n=36]
  • Executive Republics: 4.82 +- 1.96 [n=99]
  • Executive Monarchies: 3.23 +- 0.67 [n=10]
Civil Liberties (lower better):
  • Ceremonial Monarchies: 1.76 +- 1.13 [n=29]
  • Ceremonial Republics: 2.2 +- 1.35 [n=41]
  • Executive Republics: 3.82 +- 1.75 [n=108]
  • Executive Monarchies: 4.21 +- 1.47 [n=14]
Political Rights (lower better):
  • Ceremonial Monarchies: 1.83 +- 1.42 [n=29]
  • Ceremonial Republics: 2.05 +- 1.32 [n=41]
  • Executive Republics: 4.04 +- 2.09 [n=108]
  • Executive Monarchies: 5.07 +- 1.71 [n=14]
So turning a king into a figurehead gets you 2.25 σ improvement (or 1.5 on CIA's indices), and even switching from American style executive president to German style figurehead president + prime minister / chancellor with real power scores you a free 1 σ or so.

Correlation and Causation


What does it mean? If correlation equaled causation then replacement of Prime Minister Gordon Brown with President Gordon Brown would turn UK into Russia or Pakistan, on average. Now we all know that correlation doesn't equal causation, but do you really want to risk that? So how about not fixing what's not broken, and fix what is broken - first past the post election system - first?

Cats would work as such kings and queens just fine - data clearly shows that the head of state should be a figurehead, so why not a cat? Is Cat King much sillier than Magical Vaginas, really? And even if it is, silliness is no excuse not to do science.

If someone thinks this post is a serious argument for monarchy, they might be interested in this kind of science too, just saying...

9 comments:

  1. The science could be better, certainly. Especially when considering the sigma differences aren't as much as you'd really like to see for firm evidence.

    Still, it might be worth looking in to this difference between giving the heads of state executive power or not. Some research in it might turn up some interesting ideas. Well assuming any of the relevant researchers are willing to go for it. (Some times I wonder about that)

    ReplyDelete
  2. Quickshot: The difference between best and worst performing country is 3.8/3.3/2.8σ - the difference between figurehead king and powerful king is over half the scale, it can hardly be greater.

    If a real academic scientists did this analysis, they'd throw dummy regional variables and try to calculate statistical significance on that, but this would be so wrong from Bayesian perspective I'm not going to get anywhere close to it.

    ReplyDelete
  3. What about sizes of countries? Surely any sort of government would go down based on the size of the population. Which monarchies are being tallied here? Current or prior?

    44 current monarchies exist (http://en.wikipedia.org/wiki/Monarchy), 28 of which aren't "ruled" by Britain's queen so let's say there are 28. Of those 28, I don't see any particularly large countries or any military powers. Unless I've missed something they appear to have no real natural resources (so they're pretty safe, i'm guessing) either. Though, some are in Africa so who knows there.

    Just commenting that the science on this may not have taken everything into account. And if you were to take it all into account, what is "it all"?

    ReplyDelete
  4. nowthatsamatt: These are current monarchies only. Indexes are about internal freedom and democracy, not about how these countries behave internationally so I'm not sure if things you mention are terribly relevant. Democracy Index skips many tiny countries, but Freedom indexes do not (notice different sample sizes), if that's what you're asking.

    If you look at IMF list of countries by economy size, of top 25 - #3 (Japan), #6 (UK), #13, #14, #17, #21, #22, #24 are monarchies, so it's not only tiny insignificant countries.

    Many are former imperial powers.

    Many monarchies are rich in natural resources - most oil states are monarchies for example.

    ReplyDelete
  5. Sabremesh14:54

    The statistics are fascinating but there are practical reasons why splitting the head-of-state's ceremonial and executive roles might be advantageous.

    Firstly, being the political/executive leader of a country (like the British Prime Minister) is surely a full-time job on its own. The ceremonial duties of the Queen are also a full-time job (albeit with some obvious benefits!) If you consider that President of the USA has to carry out both these roles, it puts a huge stress on the incumbent.

    In addition, I believe it fosters a cult of personality around the President which is not altogether healthy. The ironic reality is that the dual roles occupied by the President of the United States make him more like a mediaeval monarch than the politically powerless British Monarch.

    ReplyDelete
  6. Obviously this is a quick shot statistics but:

    a) eyeballing the raw data, it looks like most of the effect would disappear if you control for population & GDP (A bunch of African "republics" are dragging the score down).
    b) P-values would be more convincing that sd differences
    c) Correlation does not imply causation... duh!

    If you like, I can dust off my rusty R skills and run a full ANOVA this weekend and see what I can squeeze out of the data.

    email me at hang[at]bumblebeelabs.com to remind me.

    ReplyDelete
  7. Xianhang Zhang: P-values, ANOVA and such would be standard science, and also a total nonsense, as samples are nowhere near being independent.

    We have absolutely no way of getting useful data out of cross-country comparisons like that. With enough statistical shenanigans you can make anything be or not be significant - but it would be of extremely limited relevance to the real world.

    Try writing system and democracy (clearly statistical results will say that Latin script is far more democratic than Arabic script).

    The problem is that all variables spread with culture and economy - and none of it fits into "dependent/independent variables" framework of classical statistics, in spite of academic economists' best effort to look the other way and not see the problem.

    ReplyDelete
  8. taw: I think the problem is in the interpretation, not the analysis. I could say the same thing about your analysis s/Arabic script/African Countries/.

    ReplyDelete
  9. This reminds me that they are waiting to see if the 'democracies fight less with each other' statistic will reach 2 sigma any time soonish. Basically democracies are kind of new and as such we lack the time frames required to get as good a statistics as we would like.

    Or to put it another way, we've got a fair amount of data on freedom in monarchies who have executive power, if we really felt like collecting it all. ^_-

    ReplyDelete