Oral Exams in a Virtual Classroom

This past summer, I taught a six-week online undergraduate course on algorithms in computer science. The class is usually somewhat large -- for instance, last spring, more than 200 students enrolled. But the summer session only had 21 students, so I figured it would be feasible to experiment with an oral option for the final exam. Each student could choose to take a three-hour written final or a 30-minute oral final conducted privately with me over Zoom.

Since this was an unfamiliar format for everyone, including me, I also gave mock oral exams at the end of the second and fourth weeks. The material in the first mock oral covered the first two weeks, and the material in the second mock oral corresponded to the middle two weeks. I offered a grading incentive for taking a mock oral, which was applied regardless of how well they did on the mock oral. In particular, anyone who attempted the mock oral received a slight boost in their quiz grade for those weeks, without exceeding full credit. So from that perspective, the mock orals also acted as a second opportunity for students to demonstrate their mastery of the material without requiring my doing the additional work of writing, delivering and grading quizzes.

In the end, the experiment went well, and all but four of the students -- each of whom attended at least one mock oral -- chose the oral final option. The oral final, as well as the two mock orals, offered many benefits, including ones that I had not anticipated.

Before the course began, my primary reason to give an oral exam was to minimize cheating, as I know unauthorized collaboration is common for written exams administered online. And as far as I can tell, however, no collaboration occurred among anyone who took the oral exam. It’s difficult, if not impossible, to search online for solutions during an oral exam without raising suspicion.

Another motivation for offering oral exams was to reduce grade grubbing. Although it takes less than 30 minutes to grade a written exam, that doesn’t account for the headache of regrade requests. Many students believe that they have the right to ask for extra points, especially if they are on the border between two letter grades and especially on assignments worth a lot of points. In theory, that belief is justified, but in practice, it results in a disproportionately large amount of my time being spent on haggling over a small number of points. Furthermore, it only benefits students who are willing to make regrade requests, which is unfair for those who are less inclined to do so.

Oral exams have helped with this problem, because I could grade every student immediately after they finished the exam. For each student, I asked if they would like to know their final grade on the spot or if they’d like to see it later by themselves. Every student chose to know their grade immediately, and nearly all of them felt that their grade was reasonable, so we finished the course on a pleasant note. For the few who weren’t as satisfied, I emphasized that grades probably aren’t as important as they think, and that my grading distribution was comparable to that of previous semesters.

That said, the grading process for oral exams isn’t as straightforward as I thought it would be. Over all, the students and I would like the exam to imitate a regular conversation, but unlike a regular conversation, this one has a fixed duration (30 minutes) and agenda (a sequence of questions). The student is incentivized to correctly solve each problem as quickly as possible, but sometimes I can see that they require a bit of nudging. In such situations, I can nudge them a little (giving them greater opportunity to demonstrate their understanding) or nudge them a lot (giving them more time for later questions). This has resulted in an individualized judgment call on my part for each student. Yet, fortunately, things have turned out fine: by giving them their grade immediately, I’m not concerned that, sometime later in the future, they will be unhappy with what had happened during the exam.

For computer science students, one of the benefits of oral exams is quite tangible: it serves as practice for a technical interview. Many such students aspire to become software engineers, and one notorious obstacle for many such jobs is that technical interview. It usually involves solving a problem on a whiteboard, in real time, in front of an interviewer. Since many of those problems overlap with the course material, this benefit was natural and apparent to the students.

It was also where the mock orals really shined. Since I wasn’t actually interviewing them, and the stakes were essentially zero, the students didn’t feel any pressure to fumble their way through every question. On the contrary, many of the mock orals organically evolved into a private session of office hours. It is well-known that private instruction is perhaps the most effective pedagogical tool, and my experience with every student confirmed that. In fact, as they reflected on the various aspects of the course, one student wrote that the mock orals were “most helpful,” because they “made [him] think actively and participate in a more active manner.”

I want to emphasize that this benefit applied to every student, not just the ones who struggled to answer the questions. For students who were struggling in a mock oral, I could clarify their understanding of a confusing topic or fill in a gap in their knowledge. With students who weren’t struggling, I explored more difficult follow-up questions, and they enjoyed the challenge.

We also had stimulating conversations about more advanced topics, which I found refreshing. Furthermore, I took the opportunity to encourage every student to seek outside resources in different ways, such as reviewing the course material in other textbooks and learning about topics beyond the scope of the class. Although I don’t have data to support this claim, I believe that encouragement is more effective when delivered privately than to the entire class at once.

So over all, offering mock oral exams was beneficial to both me and the students in multiple ways. Other recommendations I’d have for instructors who are considering oral exams are the following:

• Lean toward many short questions that have specific answers, rather a few long questions that have open-ended answers. This makes the exam more conversational and increases consistency in grading.
• Start with easy questions -- ones that any student who has paid attention in class should be able to answer. That decreases the possibility that the student perceives the exam as an interrogation, which reduces their nervousness and allows them to fully demonstrate their knowledge.
• Ease in and out of the exam with casual conversation. Most students were too nervous to chat before the exam, but afterward, they were happy to discuss how they felt about the process. These conversations naturally became small, enjoyable teaching moments.

In the end, the purpose of any decision an instructor makes in the classroom is to maximize student learning. How did that go? Just before the final exam, I asked the students to rate how helpful they found various aspects of the course on a four-point scale from “not helpful” to “very helpful.” For the mock orals, 14 students gave a four out of four rating, and six students gave a three out of four rating. One student didn’t do any mock orals, but they did extremely well on the oral final. So over all, this experiment with oral exams was quite successful, and I hope that in the future, more instructors -- and students -- can experience the benefits, too.