You got between 75-100% on all assignments and problem sets yet not sure you can get above 50% on the finals. Doesn't add up. Sounds like you're just stressed out and not really in danger of failing.
The MCQs are normalized according to some normal distribution. Therefore, the worst you can get is 0. However, if you score > 0, but you get the lowest mark in the class, you receive 0 marks. To make matters worse, much of the questions are written in such a way that, if you have in-depth knowledge of the subject area, the questions become ambigious, because the examiner might not have fully considered all the ways in which the true/false statement can be interpreted or reasoned (I think this would require some kind of perfect human being?). In effect: mess up on the MCQ's and you're toast.
Hope that makes sense!
On the one hand it's fascating. Yet on the other, it's absolutely terrifying.
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?