Edit: On second thought, I don't think my processes, even as strictly methodical as they may be, would not represent two full strings of perfectly duplicated cartridges to fairly determine variation precisely due to shooter only. The controls would need to be extremely extensive.