4 ms·

So the best way to run a controlled trial is to maximize the external noise and confounding variables?