Independent replications really are the ultimate arbiters of what is real in science. Any one study can be a fluke or the product of biased research. Only phenomena that are real, however, should show up in the data no matter who is doing the research.
