One might also argue that testing a 14 year old is rather useless since educational standards might've changed, i.e., decisions might've been brought about having kids start school later or perhaps their education was a bit lightened up in the years up to 14 or so and hardened after 14 or so. These are of course huge assumptions but such findings are entirely possible, and especially when you have the data pool of the entire world, meaning about 200 countries.
This is why I personally find the PISA scores to be a rather bad standard for comparing education in various countries. For example, in some countries you start school at 7 and in some at 5. This automatically means the former countries' results will be lower assuming the educational quality is the same. I'm seriously baffled as to why they don't test adults who're done with their education and judge the education of a given country based on that.