You keep making different tests until they show what you want to see

You look at your hypotheses and choose which one you like best

You compare the data from your experimental results to the prediction they tested

You guess what the results should be