We're being encouraged to use multiple answer tests at my school. In case you're not familiar with the concept, each question may have multiple correct answers; students are encouraged to mark all valid answers.

There's at least one correct each for each question. Each question has either 4 or 5 answers to select from. It is possible that all answers are correct.

I would like to grade this so that, even when there are multiple correct answers, if the student only selected one answer and it is correct, they receive partial credit.

What I'm not sure about is how to grade problems in which:

• all answers have been selected, but not all answers are correct.

• some of the selected answers are correct and others are incorrect.

We've been given no guidelines beyond what I've mentioned above. When asked about how we should grade these tests, we've been told no more than to try to award partial credit if at least one correct answer has been selected.

Does anyone out there have experience grading tests like this? If so, how might you design a grading rubric.

One simple way to design a grading rubric for this situation is to consider the question as though it were a set of true-false questions. So if there are 5 answer and the question is worth five points, then there is 1 point for marking each true, and 1 for not marking each false answer.

Thus, for example, consider the following question:

Which of these are mammals? (5 points)

1. Cat
2. Turtle
3. Lizard
4. Fish
5. Dog

If the student marked "Cat, Turtle", then they would receive 3 points: they lose 1 point for marking "Turtle" and 1 point for failing to mark "Dog."

• This is almost what I've come up with (I'm not happy with my system hence the posting here). That problem I have with it is that if students leave the question blank, I'm not sure I would want to say (as in your example) they get a score of 3/5. It seems that there should be some kind of penalty for leaving question blank. Similarly, it seems there should be some kind of penalty for excessively marking all answers. Any ideas there? Dec 13, 2014 at 1:05
• @A.Ellett It depends if you are grading on a curve... and is it really worse to have them leave a question blank than assert an incorrect answer? Dec 13, 2014 at 1:14
• It isn't really worse. I've taken a bit of the philosophy that it's the fault of the author of the exam were it possible for a student to score well with such strategies. Nevertheless,.... I was just wondering what feed back others might provide. Dec 13, 2014 at 1:21
• Make sure (and tell them) overall exactly half of the answers are correct, and scale so 50% doesn't pass? Dec 13, 2014 at 8:23
• If exactly half of the answers are correct, leaving everything blank gains just as many points on average as marking answers at random. Since students don't need any knowledge to use these strategies, I think this is a good property of the grading system. Dec 13, 2014 at 12:54