This is important to ensure that one set of results is not a "fluke". Ie, if you did an experiment once, the results may have been caused by an extraneous variable, this is a condition which occurs that is beyond the scientists control. Therefore, to ensure the results are reliable, the test must be repeated, usually with a few things changed in the method each time. The more results that match from different tests, the more accurate the results are likely to be.
