Well, it is true that the chosen behaviour gets applied to all question types. But it is a bit more subtle than that.
You can probably ignore the rest of this post unless you happen to be interested, but here is the explanation:
A) First, in the quiz settings, you choose "How questions behave". This is really only a polite request ...
B) A key moment is when a quiz attempt starts. At this moment, the quiz goes to each question and says something like "We want to attempt you as part of a quiz using 'Deferred feedback behaviour', what behaviour should we use?" (In the code this is the question_definition::make_behaviour method.)
- For most question types, they politely acceed to the request.
- But, some question types have a good reason to do something different. For example Essay questions alway reply "Well, actually, I can only work with the Manually graded behaviour." Descriptions alwasy use the "Information item" behaviour.
- And, sometimes, there are special cases. For example if you tell a matching question you want "Interactive with multiple tries" behaviour, then it acutually uses "Interactive with multiple tries (countback-scoring)" which does a more complex calculation for partially correct responses, but otherwise is just like the standard Interactive behaviour.
C) Ever after that, the behaviour is in charge of what goes on as the student attempts the question. That is, it tells the question type what to do, for example "is this response different from the last one we saw?", "please grade this response now", ...
D) However, the question type is still in charge of how to do those tasks it is told to do. So, with multiple choice questions:
- Single response questions (radio buttons) will return a negative score if the student selects a choice with a negative score.
- Multiple response questions (check boxes) never return a negative score. The plusses and minuses for all the selected choices are added up, but the final score is not allowed to go below zero.