One of the first things a professor taught me on a class on the design of experiments is "Never doom yourself to success." The experiment shouldn't inherently

One of the first things a professor taught me on a class on the design of experiments is "Never doom yourself to success." The experiment shouldn't inherently be designed to confirm your biases. I've carried that with me ever since. I agree with you - I suspect bias, judgement calls that favor what you want, looking for errors intensively when you get weird results and not when you get expected ones, etc. are far more common than outright fraud.