Presumably similar techniques could be used to explain national variations in oil consumption. I probably would not understand the methodology but presumably someone trained in the mathematics of statistics could.
For instance, if you look at Nuclear energy consumption per capita, posted above by Starvid, you see Three categories of countries (as there are two clear breaks in the data).
The first category comprises Sweden and France, which are wealthy social democracies with a strong state commitment to nuclear power. The rest of the countries could be divided into three categories: developed nuclear power and undeveloped nuclear power. The "developed nuclear power programmes" category includes industrialised countries (including former socialist), from Finland to Russia. The "undeveloped nuclear power programmes" includes countries that for one reason or another have been unwilling (Netherlands?) or unable (developing countries: Romania, Argentina, Mexico, Brazil) to develop a nuclear power programme, and those whose main purpose is military (South Africa—now filly civilian—, China, India, Pakistan). I suppose in this analysis Sweden and France should be called "overdeveloped nuclear power programmes", and this is probably the case: France actually exports electricity if I am not mistaken.
The statistical technique used is called analysis of variance (link to a long but not very informative Wikipedia article, I was actually planning on rewriting it), but in most cases 1) the important part of the work is the enumeration of the "candidate factors" for explaining the ofserved differences; and 2) if you infer a correlation from looking at data, and then apply a statistical test to the same data to "verify/falsify" your inference, the standard statistical tests are biased in favour of accepting your inference. Nothing is 'mere'. — Richard P. Feynman
