Ok, I’ll try to be more clear.
It’s a game, so the felling should matter (and a lot), then you can back them with numbers.
Also you can just take those number as they are, there is a way of reading statistics.
For example: let take a pop of 1000 games, a sample of 10 games played with the turks, 9 out of 10 are wins, so turks have a 90% win rate, but that it’s not rappresentative of the population.
For those data to be more significant, they should consider only game with random civ, that way you would more useful datas, you wouldn’t have the problem of the pick rate and you can compare better the win rate, it would be a population with less interferences.