Friday, February 1, 2008

Murder Plots

Last entry I ranted about the hapless New York Times' reports on the murder rates among Iraq veterans, and about the overall murder rate in the US.

All of which brought me to the NationMaster website, which collects national statistics on everything from population density to levels of happiness. On the murder front, they rank the US at #24, far and away the leader among western democracies. They also show what correlates with the national murder rates of all countries, implying that abortion, oil wealth, and early mortality are all associated with murder. Now pointing out that early mortality is linked to murder might be the same as pointing out that a lack of food is related to starvation, but the rest is worth review.

Correlation is not causation, of course. To say that murder and oil wealth are positively correlated, simply means that they vary together: as overall oil wealth rises, so often does the murder rate. There’s no way to know if either one causes the other to occur, or if there is actually no causal relationship. For example, it’s entirely possible that some underlying third variable is ‘causing’ both these variables to change. And there’s no guarantee that the correlation infers a linear relationship between the two variables – they could easily have a much more complex relationship. Isolating two things and seeing how they work together is interesting but usually artificial, and ignores the complexity of real life.

Then there’s the whole problem of measurement. How do you get a uniform reporting of “murder” across nations as diverse as India, Columbia and France? And Abortion? Looking at these correlations, you begin to wonder how much is masked, and how much measures of abortion, education, life expectancy, and female literacy are surrogates for a measure of oppression, poverty, and division, things far harder to quantify.

No comments: