Friday, February 10, 2012

The mustard seed approach to grading

Today I learned a new thing about grading.

What do you do if a student presents an algorithm in two ways, at a high-level and with pseudo-code, and if the two versions are different, one correct and one incorrect? Should one count the answer as correct or incorrect?

The "mustard seed" response: there's some good in there, so focus on what's good and count it as correct.


  1. I like the mustard seed approach that you take, but I have my worries. A student might make it a habit to present two versions for each problem where he/she is a bit unclear. What then?

  2. The other day, in an exam, I had a situation similar to the following scenario.
    Question: how much is 2+2?
    Student's answer: 4 because the Earth is flat.
    How do you evaluate such things? They know the answer to the question but it is obvious that something is deeply wrong...

  3. If the space of solutions is compact, then a student can (in theory) densely cover it and get the full mark without knowing the right answer himself/herself. Some complexity control is needed.


Note: Only a member of this blog may post a comment.