I also see that you have edited your piece to include another source of data, but that study (“Culture, Gender and Math”, available at www.kellogg.northwestern.edu/faculty/sapienza/htm/science.pdf) does not include your graph. It also does not deal with the same issue: It treats the average PISA test results of boys and girl…
As stated in the article, it refers to the data, which was worth adding.
Plus, the relevant points indicate that the mathematical skill gaps exist and are narrowed in the case of gender equal societies. It is fully accounted for by them. There is no unknown variables in that gap. So education is it.