October 30, 2009

Islam by the numbers

Posted in International development at 1:46 pm by Eamon Aghdasi

I have a good grad school friend named Adam (not his real name) who’s from a predominantly Muslim country but is harshly critical of Islam. Like many countries in the Middle East and Central and Southwest Asia, Adam’s country is developing but moving slowly in certain regards, and threatened by the possibility of catastrophic collapse. It is a burgeoning democracy, but (by most measures) suffers from poor governance and widespread corruption. It has millions of moderate, peaceful Muslim believers who care more about their families’ wellbeing than politics, yet violent islamists are growing in strength within the country.

Adam and I have a usual ironic pattern to our arguments. It usually involves him (Muslim, at least by family) arguing how terrible Islam is while I (not Muslim but very respectful of Islam) defend it. Usually the arguments are about some aspect of Islamic history or doctrine (for instance, where Adam will raise the example of some hadith where the Prophet Muhammad said such-and-such ridiculous thing, and she sheer ridiculousness allegedly proves Islam’s falsehood).

To Adam, though, Islam is not just a false religion, but it represents a movement and belief system that is actually harmful to his country’s progress. Recently, we argued over email from 8,000 miles apart about this, when Adam pointed out how poor female labor force participation is in many Muslim countries. This isn’t an opinion, but a fact; in reality, Muslim countries perform very poorly on a wide range of social indicators. Below is a scatter plot, just to provide a visual example, with World Bank Voice & Accountability scores on the y-axis, and the proportion of countries’ populations being Muslim on the x-axis. The downward sloping relationship is pretty visible here.


However, making the jump to the conclusion that Islam somehow causes these poor social outcomes (or anything else) is something that no sensible person, including Adam, would do. Adam’s argument was based both on the data and on anecdotal evidence from his own life. On the contrary, rather, it is possible (likely) that the prevalence of Islam happens to be correlated with some other factor that is driving these outcomes.

Is it poverty? In other words, do Muslim countries seem to be different simply because they happen to be poor (prevalence of Islam is definitely correlated with poverty)? I was curious about this, so I decided to run some regressions, with the dependent variables being a series of social indicators, and the explanatory variables being 1) the percentage of the country’s populace being Muslim, and 2) per capita income. If there’s a connection between Islam and the social indicators other than through the correlation with lower income, then the coefficient of the “percent Muslim” variable should be statistically significant.

The results were pretty remarkable. The prevalence of Islam, controlling for income, is associated with poorer scores on all six World Bank governance indicators; lower levels of education; higher fertility rates; less frequent use of contraception; less use of the internet; lower ratios of female-to-male school enrollment; a higher ratio of female-to-male unemployment; fewer women in parliament; and a lower ratio of female-to-male income. The one bright spot was that it was also associated with less inequality. (See here for a table of results.)

It occurred to me that oil might be playing a part in the results. The reason is as follows: a) poor results on the social indicators mentioned above are strongly correlated with lower incomes; b) the prevalence of Islam in a country is correlated both with lower incomes and with having oil; and c) having oil can be seen as a “short cut” in some cases to higher incomes, where the link that the same endogenous process of rising incomes and improving social factors doesn’t take exist. Is it possible that a handful of wealthy, oil-rich Gulf countries which happen to be predominantly Muslim were skewing the results?

I ran a second regression, this time adding another control variable, the number of barrels of oil a country produces per thousand people. If the link between Islam and poor social outcomes is explained by oil, then adding this control variable should render the Islam coefficient statistically insignificant.

It turned out there is some evidence for the oil hypothesis above. The coefficients for most of the factors examined get smaller, meaning the effects of the “percent Muslim” variable is weaker when you introduce the oil variable. For instance, the first regression on contraceptive prevalence predicts that in an all-Muslim society, 17 percent fewer women would use contraception than in a no-Muslim society; in the second regression, this difference is just 12 percent. There are also some social factors for which the statistical significance of the “percent Muslim” variable decreases. For regressions on rule of law and internet use, the coefficient no longer passes a 5% test (as it before adding the oil control), and four other factors see the coefficient no longer pass a 1% test. Of course, you have to take this with a grain of salt, because the additional multicollinearity when you introduce the oil variable should drive up standard errors, so this might just be a purely statistical phenomenon, not anything meaningful. (See here for a table of results for the second batch of regressions.)

Nonetheless, you can’t deny the obvious here, which is “how Muslim” a society is does have a statistically significant link to different social outcomes, even after controlling for income and oil. The next question, and one I think is much harder to explain, is why.

It is naive to look at a group of people who embrace a particular set of beliefs, identify their characteristics, and conclude that the beliefs are the cause of those characteristics. This blind leap from correlation to causality is as naive as the first-semester statistics student who, upon learning that people who eat caviar tend to live longer than those who don’t, concludes that the consumption of slimy fish eggs is the key to long life. It takes common sense, and a will to resist one’s intellectual pride, to wonder about what the real reasons might be, even if we have eliminated the possibilities of income and oil.

What my own common sense tells me is that, though many of Islam’s social teachings (having emerged in the 7th century) seem antiquated, this is not unique to Islam. In every other religion that precedes Islam, we can find teachings that are seemingly barbaric by today’s standards. And yet, I have never read a news story about some Israeli adulterer getting stoned to death, even though the Torah says so. Neither have I read such a story coming from Turkey either, by the way.

So what’s really going on here? Add your comments below.


1 Comment »

  1. Adam said,

    Left by Adam, via email:
    You have rightly pointed out the antiquated beliefs enshrined in other religions of the world, but the difference is that not many of these religions get declared as “State Religions”, where as Islam does get put up on that pedestal in almost every other democratically approved constitution in every other muslim country. For instance, how many christian majority countries have Blasphemy laws? this is despite the fact that in medieval europe people were burnt alive for being blasphemous. But in most islamic countries blasphemy is punishable in the court of law. So all in all, I think muslims in general might be a tad bit closer to their religious beliefs than the rest. Given this reverence it is very difficult for a muslim to reject a hadith that would appear “ridiculous” on modern moral standards. Because as principle, religion is supposed to shape his modern moral values, not the other way round. And to turn the causality, these needs to be a consensus among a sizebale portion to make the new stand “acceptable”, and for that you need time.

    And this brings me to my next point, which is that when you are comparing Islam to other religions you have to consider the age of Islam as a belief system i.e. the number of years that Islamic beliefs have been in public scrutiny. Because, it is this scrutiny that has made the modern Christians and Jews very different from their medivel predecessors, this current evolved form lets them pick and choose, which is a choice that the ones before them didnt have; imagine being an openly gay priset at the time when they were burning witches alive.

    In that respect, Islam still has around 600 years of evolution to go through to be comparable to modern Christianity and even more to be compared with modern Judaism or Hinduism.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: