From my understanding, you'd have to repeat the experiment isolating each variable to see what difference each one actually makes, no?