In the quiz statistics report, there is a section 'Response analysis' which gives you details of how many students selected each possible answer. For many question types like multiple choice, this is simple:

For other question types, it can be less clear what we mean by "possible answers" but most question types manage to come up with something sensible.

Currently, the ordering question type does not support this sort of analysis, and I am trying to think of a way to add it. However, I am stuck trying to think how to classify reponses.

The conceptually simplest approach is to just count each possible order. However, this rapidly gets rediculous. With 6 items, there are 720 possible orders. That is too much.

So, what else could we do. We could could "How many students put Item X in Position Y" but I don't think that is very helpful. If you created an ordering question, and then later get told that out of 100 students who attempted the question, 40 put "Frog" in third place, does that really held you see whether they understand the subject. Also, this is still quite a lot. With 6 items it is 36 possible classes.

The only idea I have had so far that I am not too unhappy with is, what if we consider each pair of options, and they count "How many students correctly put X in front of Y" (you only need to count it one way around, because everyone else will have got them the wrong way around). It might acutally be useful to know that out of 100 students, only 40 of them knew that you should check the patient is breathing before checking their pulse*. Also with 6 items, there are only 15 pairs, so this grows at a manageable rate.

Does anyone else have any better ideas? Or thoughts on these suggestions? Thanks.

* Acutally, it seems that first aid is no longer that simple: https://en.wikipedia.org/wiki/ABC_(medicine)