February 06, 2006

Lies, Damned Lies, Statistics, and Islam

Dean Esmay has a pretty good post about whether or not Islam compatible with democracy or not. Go read it.

Also, after you have read it you should probably read my essay Marx, Communism, Totalitarianism; Muhammed, Islam, Terrorism, in which I argue that Islam is a socio-political ideology every bit as much as it is a religion. It is much like Marxism in that the core ideology itself leads is incompatible with Western liberalism.

Now, let's return to Dean's post. Unfortunately, the data he provides is a classic example of what statisticians call selection bias. That is, you use data which supports your hypothesis and then exclude data which nullifies it.

Another problem is that the data do show that Muslim nations are becoming more free. A good sign, no doubt. But becoming more free does not make one free. One might argue that China is becoming more free, or that the Soviet Union in the late 1980s was becoming less authoritarian, yet there is no doubt in my mind that Communism is somehow intertwined with the fact that freedom was so scarce. Becoming more or less of anything is kind of irrelevant to the discussion.

Further, Indonesia has been a terrible ally in the war on terror. See this, this, this, this, and this. They have not cracked down on radical Muslims any more than has suited their own national interests. It is in no way a liberal democracy. Freedom House is simply wrong in this respect. I would point out that it is a crime for a Christian to give a Bible to a Muslin in Indonesia. And much of the Indonesian culture itself is not tolerant in any liberal sense of the word. And is a country truly a liberal democracy if a pictureless version of Playboy is banned?

Last, he gives the example of Senegal and Mali as the only nations on earth that have a Muslim majority and which are listed by Freedom House as liberal democracies (excluding Indonesia). Given that Freedom House gets it dead wrong on Indonesia, I suspect that there may be problems with their coding of Senegal and Mali as well. But I could be wrong. I don't know enough about these countries to speak on whether or not Freedom House correctly codes them.

However, statistically speaking two examples do not a falsification make. There are always exceptions to rules. My original point was that the Left loved to use Yugoslavia as an exemple of a 'free' Communist country. Even if we were to grant them that Yugoslavia was 'free' during the Cold War, to claim that there was not a causal relationship between Communism and totalitarianism based on a single outlier is an example of poor analysis (or poor operationalizing skills).

If one were really interested in seeing whether or not there is a relationship between Islam and liberalism, I would suggest the following. In fact, I dare any one to run the following analysis.

Hypothesis: there is a strong correlation between the percent of a nation's population that is Muslim and the extent to which that country's population is free in the liberal sense of the word.

Null Hypothesis: there is no relationship between the percent of a nation's population that is Muslim and the extent to which that country's population is free in the liberal sense of the word.

Plot a simple OLS regression model with the two variables. The first variable would simply be % Muslim. The second variable would be the Freedom House numbers. Since the Freedom House Numbers are coded negatively the following results should be found.

If Dean is right, and there is no relationship between Islam and freedom, then obviously the plots should be completely randomly distributed.

If I am right, and there is a relationship between Islam and freedom, then a positively sloping line should emerge. That is to say, as the percentage of Muslims in a country goes up, the Freedom House numbers should also go up.

The third alternative, of course, is that there is a positive correlation between Islam and liberal democracy, in which case one would find a negatively sloping line.

If we are agreed that the above is a moderately fair way of empircally testing the relationship between Islam and tyranny, then the gauntlet has been thrown. I personally do not have the time to run the numbers, but perhaps some enterprising blogger with moderate experience using SPSS would like to give it a go?

PS-Dean offers a spreadsheet, but since I don't know what exactly it represents, I'm not sure how to interpret those data.

Also, Dean and I are on very good personal terms. Please do not misunderstand my criticisms of his analysis as anything more than friendly. I consider Dean and Rosemary Esmay personal friends (in the blogging sense of the word) and will not tolerate any talk of Dean being a 'dhimmi' or other such nonsense.

By Rusty Shackleford, Ph.D. at 11:00 AM | Comments |