The article fails to provide important details. Imagine the following two situations: - In every country, there are as many boys and girls taking the science e

The article fails to provide important details. Imagine the following two situations: - In every country, there are as many boys and girls taking the science exam, and the girls get a better average. But the top n students are all boys. - In every country, there are three times as many boys taking the exam, and the girls get a better average. In both cases, we could conclude that the girls outperform the boys if we use the average as metric. But the first scenario would echo Larry Summer's comment on large standard deviation, and the second would be an instance of selection bias. Another point that would be interesting to know is wether gender relative performances at age 15 carrie on to later years.