Measuring case mix and quality of care: Validity studies for the graduate medical education study