Teaching Undergraduate Artificial Intelligence Classes: An Experiment with an Attendance Requirement

Sven Koenig, Tansel Uras, Liron Cohen

Abstract

In undergraduate education, instructors often face a steady decline in student attendance as the semester progresses. This study investigates the impact of a mandatory attendance policy in an "Introduction to Artificial Intelligence" course. By comparing a section with a grade-based attendance requirement to a control section, the researchers evaluated effects on physical presence, academic performance, and student satisfaction. While the requirement successfully boosted attendance, it yielded no significant improvement in grades and resulted in a substantial decline in course evaluation scores.

Introduction

A common challenge in higher education is the "mid-semester slump," where lecture attendance frequently drops to 50–60%. Educators are divided on how to address this: some favor mandatory attendance to ensure exposure to material, while others argue that university students should manage their own learning. This paper presents a controlled experiment designed to quantify the trade-offs of imposing an attendance requirement in a technical computer science course.

Methodology

The study was conducted during the Fall 2018 semester at the University of Southern California (USC) within the CS360 "Introduction to Artificial Intelligence" course. Two sections were compared:

  • Section 1 (Mandatory): 74 students. Attendance was required, with unexcused absences beyond a threshold resulting in a grade penalty.
  • Section 2 (Control): 62 students. Attendance was optional with no grade penalty.

Both sections were taught by the same instructor using identical materials, assignments, and exams. Attendance was tracked via a location-based mobile app (Arkaive) and verified with manual headcounts.

Impact on Attendance

The mandatory policy was effective at increasing physical presence. In Section 1, attendance averaged approximately 77%, compared to roughly 56% in the control section. However, the data revealed that students in the required section often "gamed" the system by checking in via the app from outside the room or attending only to work on unrelated tasks.

Attendance trends for lectures and exercises show consistently higher presence in the mandatory section (Section 1) compared to the control (Section 2).

Academic Performance vs. Student Satisfaction

Surprisingly, the increased attendance did not translate into higher grades. There was no statistically significant difference in performance across three programming projects, a midterm, and a final exam. In fact, the control section (Section 2) performed slightly better on average (80.49% total score vs. 78.85% for Section 1).

The most dramatic finding was the impact on course evaluations. Students in the mandatory section rated both the instructor and the course significantly lower than those in the control section.

Course evaluation ratings across multiple categories show a consistent and significant drop in satisfaction for the section with mandatory attendance.

The researchers attribute this "evaluation trap" to selection bias. In the control section, students who were unhappy likely stopped attending and were absent during the final lecture when evaluations were administered. In the mandatory section, these frustrated students were present and voiced their dissatisfaction, leading to a nearly 1-point drop on a 5-point scale for the instructor's rating.

Conclusion and Limitations

The study suggests that while attendance requirements can put "bodies in seats," they do not necessarily foster engagement or improve learning outcomes in an undergraduate AI context. Instead, they may create a hostile environment that negatively impacts instructor evaluations.

Limitations included "section mixing," where students occasionally attended the other section's lecture, and environmental factors like poor classroom acoustics. Future research might explore applying requirements only to specific session types or using more neutral evaluation methods to avoid the selection bias observed here.

Summary updated

Abstract

We report on an experiment that we performed when we taught the undergraduate artificial intelligence class at the University of Southern California. We taught it – under very similar conditions – once with and once without an attendance requirement. The attendance requirement substantially increased the attendance of the students. It did not substantially affect their performance but decreased their course ratings across all categories in the official course evaluation, whose results happened to be biased toward the opinions of the students attending the lectures. For example, the overall rating of the instructor was 0.89 lower (on a 1-5 scale) with the attendance requirement and the overall rating of the class was 0.85 lower. Thus, the attendance requirement, combined with the policy for administering the course evaluation, had a large impact on the course ratings, which is a problem if the course ratings influence decisions on promotions, tenure, and salary increments for the instructors but also demonstrates the potential for the manipulation of course ratings.
BibTeX
@inproceedings{koenig2020teaching,
  author = {Sven Koenig and Tansel Uras and Liron Cohen},
  title = {{Teaching Undergraduate Artificial Intelligence Classes: An Experiment with an Attendance Requirement}},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year = {2020},
  doi = {10.1609/aaai.v34i09.7060},
}

Loading related content...

Figure