Studies of studies are always dangerous
In any statistical study of statisical studies, 5% of the subsidiary studies will lie outside the 95% confidence level. That means that of the more than 100 countries studied here, a number of the countries' figures are going to be just plain flat out wrong. So seizing on a particular country, eg Iceland, to demonstrate something is highly dangerous. The subsidiary results you think are the most interesting could well be those where the numbers are simply not true.
Also, these sorts of studies are often riddled with the individual economist's own biases, and especially their overwhelming desire (true of most scientists in general) to see their name on published artlcles. On a subject like this one, I'll only be convinced one way or the other once scientists have figured out which bit of the brain is responsible for mathematical proficiency, and then studied an unbiased sample of 100's (at least) of brains to see what those people had in that part of their brain.
Biting the hand that feeds IT
