I'm impressed by your diligent methodology. Too many people draw conclusions from a dozen pot-shots. This is the only way to get results you can put any faith in. And even here, in terms of the groupings, there's clearly a lot of randomness involved.