Imagine you have an experiment testing more: let's say you measure the influence of temperature and light on the growth of plants. Your result show an influence, but you don't know whether it was temperature, light or a combination of both that caused this effect!
If you test only one variable at the time, you won't run into this problem
