I've decided to cut and paste a post I made a while ago in a thread about academies about the 'outperforming' and league tables.
"As soon as anything is put in a table schools start to game the system to move up that table. So if you have a table that ranks by average points per pupil, you make your pupils sit more exams - 4 A2s for instance rather than 3, 12 GCSEs rather than 10. This isn't useful to the pupils or necessary for university entrance, but it pushes the school up the tables. A more interesting ranking, for many parents, would be average points per exam entry - how well did the pupils do on average in each A2 they took? If you rank the 2014 table in this way, the results change dramatically. In 1st, 2nd and 3rd place you have three independent schools, St Paul's Girls' School, Wycombe Abbey School and St Paul's School, followed in 4th place by Westminster. The pupils at these schools take fewer exams each but get better results in them. More useful for the pupils and their university entrance, but not shown by the headline way that the table is ordered."
This doesn't diminish the achievement of the highly performing state schools, though most are also selective schools, as are the independent schools. It's fantastic that they are improving and good for everyone. But it's top line rhetoric to say they are outperforming top independent schools. On one measure, they are. But not on the measure of what is most useful for their pupils.
For example, I know a local school who made their pupils take two English GCSEs by different boards and then entered the highest result in their data. No use at all to the children who had to sit two exams but possibly mildly helpful to the crucial 5 A-Cs including Maths and English data that the school had to supply.