P hacking is easy. Get test and control groups. Measure 40 health attributes. Give test group your special compound. Measure 40 health attributes again. Keep on

P hacking is easy. Get test and control groups. Measure 40 health attributes. Give test group your special compound. Measure 40 health attributes again. Keep only results that show positive benefit (avg. around 2 by random chance). Publish, profit. Data would be clean and does not need to be manipulated. The overall scientific process is rife to be corrupted as long as tester and measurer are the same party, especially when there is financial incentive.