Using credibility ceilings to explore skepticism about observational evidence.