Yup... uh... this is great advice for life in general. Really what this guy is saying boils down to... pay as much if not more to model selection / validity than you do to model analysis