Part 3: Oral Exams

I have used Oral Exams in several courses in the past (STAT 430/830: Experimental Design, ACTSC 455/855: Advanced Life Insurance Practice, and ACTSC 613: Probability and Statistics for Actuaries) based on research I had done while completing my Certificate in University Teaching. So I was eager to use them in STAT 334 as well. I always use them in addition to, rather than instead of, a traditional written final exam.

If you are interested in further information, I spoke about my use of Oral Exams at the Actuarial Research Conference in August 2015 in Toronto, and my slides from that presentation are attached.

Oral Exams have a number of advantages, including:

• Easier to assess the depth of knowledge by asking follow-up questions
• Can clarify questions for students if they get off track
• Students get immediate feedback
• Strengthens communication skills
• Preparation for Oral Exam requires deep thinking about relationships between concepts
• Can help prepare for technical interview questions (unintentional but great effect!)

They are not without concerns however, including nervousness on the part of the students, cheating or bias/subjectivity in grading, and the time it takes to administer. I made several choice in the way I run my Oral Exams, which alleviate most of the concerns.

First is the way the exams are structured. Students are told they will be asked 5 questions (one definition, one advantages/disadvantages, one compare/contrast, one describe a process, and one discuss the impact of a change) and given the opportunity to practice one or two questions, so there are no surprises. The exam is open book so they can bring in absolutely anything (including the textbook, although I recommend making summary notes to be able to find things quickly if they need to look something up). And the exam itself is 15 minutes and only worth 5% of their final mark.

Secondly, in terms of fairness, I have a bank of questions (it is EXTREMELY easy to create questions following the 5 categories above) and pseudo-randomly generate a set for each student so that all the main topics in the course are covered for each person, but no two people have the exact same questions. For the grading, a clear marking scheme helps cut down on subjectivity. For example, it could be that each key point they get without any prompting/help, they get 1 mark, and for each point with prompting they get 0.5 marks, until they have 5 marks or time is up and we move on to the next question. They can also skip or go back whenever they like within the 15 minutes, or ask questions about what I mean.

Finally, the time it takes to run the exams can be prohibitive if you have a large class (I've done it in classes ranging from 18 students up to 60, which was too much. 40 students is 10 hours of exams, for example.) But with a reasonable class size, it can give you a much clearer picture of the knowledge your students have gained, and make them review the material at a much deeper level, which will help them on their written exam too.

I asked some of my former students about their experience with these exams and I'll end with a quote from one of them: "This helped me understand the concepts better when studying for the final exam and in my future courses because it did not just consist of memorizing information and forgetting it immediately after the written exam. I actually had to learn to understand the concepts in order to communicate them effectively."