You're explaining a student's response to the exam structure, but you're not explaining why the institution wants to fail some students every year. If you care

You're explaining a student's response to the exam structure, but you're not explaining why the institution wants to fail some students every year. If you care about student learning, you do a criterion-referenced assessment. If you want to compare students to each other, you do a norm-referenced assessment. Why is a grad institution more interested in comparing students to each other than in assessing whether people learned what they were supposed to learn?