Unrealistic Expectations Bedevil Computer Science Students

Novice computer science students often believe they are doing poorly when they restart a problem, ask questions, or take time to plan their work.

But these practices are normal—even for the most experienced professionals—a message students need to hear, according to a new award-winning research paper co-authored by Northwestern University’s Eleanor O’Rourke, associate professor of learning sciences in the School of Education and Social Policy and of computer science in the McCormick School of Engineering.

The paper, Understanding the Reasoning Behind Students’ Self-Assessments of Ability in Introductory Computer Science Courses, won the Best Paper Award at the 2024 Association for Computing Machinery’s Conference on International Computing Education Research. It offers new insights into why students become self-critical.

In addition to O’Rourke, the study was co-authored by Northwestern graduate students Melissa Chen, a third-year doctoral student in computer science, and Yinmaio Li, a PhD candidate pursuing a joint degree in computer science and learning sciences, the first program of its kind in the US.

While enrollment in computer science courses continues to grow, retention rates remain low, with high dropout and failure rates. Previous research suggests that misperceptions about the programming process contribute to the problem.

“Our data revealed the power that instructors have to set course policies aligned with accurate expectations about the learning process and to model practices like debugging,” O’Rourke said. “Being more intentional could go a long way toward reducing negative self-assessments, which lead students to drop the course.”

The researchers interviewed seven beginning computer science students, presenting a vignette about a hypothetical student, for example, who didn’t understand an assignment. Students were asked to think aloud about whether they would be performing poorly in the class if they experienced a similar moment, and to explain their reasoning while filling out a survey.

The analysis revealed that students with low confidence had high expectations of what was required and felt they lacked the skills or resources to overcome setbacks. Conversely, asking questions or taking time to plan led students to feel more doubtful about their abilities.

“Students often rely on their perceptions of their capabilities, their expectations, and their observations of ‘normal’ behavior in the computing community to self-assess their performance,” the authors wrote. “We also identified unprompted sources of these beliefs, such as the behaviors and policies of instructors and the actions of peers and professionals.”

O’Rourke, whose research focuses on how student beliefs and expectations about programming shape their motivation and learning, co-directs Northwestern’s Delta Lab with Liz Gerber, Matt Easterday, and Haoqi Zhang. She is also a founding faculty member of the university’s joint PhD program in computer science and learning sciences (CS+LS).

Her work blends human-computer interaction, artificial intelligence, and learning sciences. Other projects include building AI models to detect moments when students may feel badly about their programming skills and designing tools to help novice web developers learn from authentic professional websites.

“Part of what I really appreciate about the School of Education and Social Policy is the way colleagues in other disciplines have pushed my thinking,” O’Rourke said. “Building one little tool—which is what I learned in my computer science-oriented graduate training—isn’t going to solve big structural problems. What we need are the different pieces: the social community, the curriculum, and the tools students are using.”