Generating Exams and Formative Feedback

Abstract: Creating an exam that ensures an even coverage over all the Intended Learning Outcomes and at the same time ensures that to pass, the student should have passed all the Intended Learning Outcomes can be a difficult task. After an exam have been graded, formative feedback should be given to the students, especially in those cases where they did not pass the exam. This can be a time consuming process, and because of this, it is not always possible to give as good feedback as one would otherwise like. In this project an exam tool was created, for assisting in creating exams and give individual formative feedback to the students after an exam. By storing all the questions in a database, where each question is connected to an Intended Learning Outcome, along with writing the questions in a standardised format, with regards to point representations, grading rubrics and references it becomes possible to automatically generate exams. Generated individual feedback is created by combining the information in the question together with the students result. The feedback provides motivation of given grade, which topics a student should read up on, along with recommended reading instructions. In those cases a student got a full mark, it is also possible to provide further study instructions, this can be anything from recommended articles and book chapter, to courses. Differential privacy have been used to anonymize the grade distribution to make it possible to show the students how the exam went, without risk exposing what grade a certain student got. The exam tool created achieves all of these goals, however there are still much room for improvement. The anonymisation function needs further development, since differential privacy is not suitable for small datasets. The usability of the interface and feedback reports needs more work. However in its current state it is fully functional and have been used in numerous courses.