This post only applies to the "One answer only" option of the Multiple choice question type.
When a quiz is set to Deferred feedback, negative grades are taken into account directly when calculating the total score of the quiz. For example, if the student obtained 100% for the first question and -60% for the second question, the total score for the quiz is (100-60)/200 = 20%.
On the other hand, when the same quiz is set to Adaptative mode or Interactive with multiple tries, the negative grades are replaced by 0 before calculating the total score. For example, with the same grades as above for both questions, the student's total score for the quiz is now (100+0)/200 = 50%.
It seems inconsistent to have different marking schemes for the same multiple choice questions when the quiz is set to Deferred feedback on the one hand or to Adaptative mode and Interactive with multiple tries on the other hand.
Can anyone give me the reasons why this behaviour would be desired? Should this behaviour be documented as is, or should it be the object of a tracker?
Think about the way penalties work in adaptive mode. The rule is adaptive behaviour is: each try you make, your score is reduced by the penalty, but it never goes below zero.
So, this is about adaptive behaviour behaving consistently, not about multi-choice behaving inconsistently.
Also, adaptive behaviour tends to be used for quizzes where the purpose is for students to learn from their mistakes by getting feedback. Deferred feedback is use when the purpose is to measure the student. Therefore, negative grading is only really important with Deferred feedback.