Data and expectations

Today at Scientopia, I'm writing about how the experiments that I'm currently doing should fit previous data in order to go into a paper. Is this ethical and how can we prevent bias in such cases?