The current scoring doesn't work as I would expect and having discussed this with a few team members I would question whether this should be the default scoring behaviour. For example with the current develop branch of adapt-contrib-mcq, you are given a score based on how well you answered the question. This means that you could answer every question incorrectly in an assessment but still end up with a positive score, which depending upon configurations could actually result in a passing score.

As default, would it be best to only give a user a postive score for correctly answered questions? If you wanted to allow postive scores for partially correct answers then this could be a configrable attribute on the model.

There looks to be an issue with the way partially correct scores are currently calculated too. It bases the percentage of the _questionWeight on how mnay items there are rather than how many correct answers there are. This means that a partially correct question with more options will generally provide a lower score than one with fewer options. For example:
10 options - 2 correct options - _questionWeight:10 - partially correct answer score = 1

3 options - 2 correct options - _questionWeight:10 - partially correct answer score = 3.33

Surely in both of these cases the score should actually be 5?

I will raise an issue on github for this also, but wanted to see what the general opinion was regarding this default functionality. I haven't yet checked other question types to see how they work. I assume they also reward partially coprrect answers?