Sometimes, those big professors do not have enough time and energy to do some experiments by themselves. Data come from their students and PDFs. To prevent fake data, they have to be smart and find some way to verify the data, but that can be difficult.
Just think it over the motivation for those big professors to make fake data: they have been famous and they have many good papers in Nature or Science. Why do they still need to produce fake data for a Nature paper? Thus, the problem often comes from those young researchers like students and PDF.