Someone has asked about exam question weightings

I've had questions on that.  On two matters: relative importance of correct answer over good deduction, and on weighting of problem "numbers" vs "letters" vs "roman-numbered items".  To address the first one, it is possible to get a very generous score with wrong answer if the error is an insignificant calculations mistake - it happens quite often that an answer gets a score worth an A or close to it despite there being a glitch of some kind. And you are required to justify your answers; simply writing down an answer without any sort of justification has limited value.  (For example in the spring 2014 problem there is a "does ... have ...?" question.  Simply answering "yes" or "no" gives you a fifty/fifty chance of being right, but is that answer really much more worth than "a coinflip says that ..."?)

Then it is the weighting of questions.  In particular, someone put forth the question on whether problem 4 of the spring 2014 exam counts 25 percent.  The short answer is "no, think 1/12 not 1/4".

The short and imprecise story (read disclaimer below) is that I try to facilitate equal weight over letter-enumerated items (a), (b) etc., and that a "Problem 4" with no letter is intended to be as one "4 (a)". For spring 2014, think 12 problems: 1a, 1b, 1c, 2a, 2b, 2c, 3a, 3b, 3c, 3d, 3e, 4.

Now here is the long story. With disclaimers all over:

• Unless the problem set prescribes anything - which it rarely does - I want to leave weighting up to the grading committee.  And in case of an appeal on the grade: up to the new committee.
• What I mean by "try to facilitate" and "intended to be" above, is that I split problems up in such a manner that I think the committee could reasonably attach equal weight to "letters".  That is one reason why you frequently see sub-items like (a) i) and (a) ii) (another is precision - it hopefully makes it harder to overlook a question).
• The committees sometimes find equal weighting unreasonable (maybe by itself, maybe after having seen that it has unreasonable consequences for the grading) and choose to deviate. They could e.g. decide that "(d) i)" and "(d) ii)" should count as if they were "(d)" and "(e)".  That is up to them.
• Sometimes there is a lot more work on, say, "(c)" than on "(a)". On one hand, that may appear strange.  Then on the other hand, the long work on part "(c)" may involve a lot of routine calculations where it is easy to make mistakes that should not be penalized too harshly - and part "(a)" may involve theory so crucial that it is worth quite a bit (one may argue that one should be penalized for being clueless about even a "calculate the partial derivatives of orders one and two" question).  Ultimately the committee decides.

- Nils