Sometimes we cannot design and perform our own experiment to test an idea and need to use an experiment that was already performed by somebody else. Why is it still important to make a prediction based on the idea being tested before we look at the data from that experiment?