You create a hypothesis and a testing plan on how to test the hypothesis. You collect data and analyze results. If the data does not support the hypothesis, you either need to figure out if your test parameters aren't correct, the analysis isn't correct, or the hypothesis isn't correct. If the data does support the hypothesis, you need to verify that the test parameters, analysis, and hypothesis are correct and take further data. If, in the end, everything pans out you write a paper and submit it to one or more journals. The journals will select reviewers who are experts in the subject matter you're writing about, and the reviewers will then attempt to completely destroy everything you've done for the past few years. If, after many rounds of questions and answers, you can satiate them, they will sign off on your paper. The journal may at that point then publish your paper if they feel it's appropriate for their journal.