Can you please verify if there's any truth in that?
The problem is, just a few days of testing gives you a very small sample. Unless there are singular radical changes (like about everybody stopping to fly the pony), things like type of terrain being up or a single player having the run of his life for a week in his favourite plane will have too large of an impact.

"% of fighter kills" means all kills made by fighters.
With an experiment like you are currently experiencing, I'd like to see at least a full tour completed so that I could also check the scores for any significant statistical changes in gameplay behaviour.