Nov 28, 2008

A non-causative, but suggestive correlation

In a cross-section of 40 countries in 1995, religion (measured as the percentage of people attending religious service at least once a week, from World Value Survey) is positively correlated with corruption (measured from the index developed by Transparency International). There seem to be three clusters of countries within the scatterplot reported below: countries with low levels of religiosity and low levels of corruption (e.g., New Zealand); countries with intermediate levels of religiosity and medium-to-high levels of corruption (e.g., Brazil); godless and corrupt countries (e.g., China). One should remark that all very religious countries are also very corrupt. However, high levels of corruption are associated both with low and high religiosity, indicating that the impact of religion on corruption is fragile. One may think that in some of the countries in the NE region of the graph attendance of religious services is low because it was/is actively discouraged by the government (e.g., China, Russia). The results are however not different when I use a definition of religiosity based on personal beliefs rather than active behavior. In any case, the link disappears after controlling for GDP per capita.
In conclusion, I fail to add a statistical significant impact on corruption to the list of bads brought by religion, but it is quite safe to claim that the level of religiosity in a country does not incoculate it against high levels of corruption. Italy being a point in case.


Pierre-Louis said...

i guess education is another driving force of both religiosity and corruption. In my memoire I had found that one more year of education is associated with a 0,02% (not much) increase in the probability of intergrity (think its very wrong to accept a bribe)...but also that believing in heaven was increasing integrity by 3,3%!
the fact that this does not hold at the macro level is that even though religious people say they have stronger moral values, it just isnt true in practise...

cosimo said...

Yeah man I also anted to control for education and the empowerment of women. For the former, I couldn't find good data in the WDI and I was lazy enough not to look elsewhere, the second is not as important after controlling for gdp. But these are very preliminary results at much aggregated level for a few countries. When I have time I'll look at the WVS micro data and I'll do other stuff. A +