As a complement to pure Bayesian reasoning, I recommend Mayo's "Error and the Growth of Experimental Knowledge". http://www.amazon.com/Experimental-Knowledge-Science-Concept...