Grant expands active learning in computer science

Tough questions, group discussion, and on-the-fly lecture revisions are improving computer science instruction.

November 7, 2012

An interactive teaching method that has proven its worth in college physics and is entering other fields will be developed in computer science, thanks to a two-year, $200,000 grant from the National Science Foundation. Leo Porter, assistant professor of computer science, is Skidmore’s principal investigator in the project; the other two lead investigators are at the University of California at San Diego, where Porter was a graduate student, and at the University of San Diego. One of 125 winners out of 1,050 proposals, the project is part of the NSF’s Transforming Undergraduate Education in Science, Technology, Engineering, and Mathematics Program.

The project is based on UCSD research demonstrating that the “peer instruction” pedagogy yields better learning outcomes than traditional lecture formats in a range of computer science courses. Tracking several courses and instructors, Porter and colleagues have shown that PI’s advantages include a 50 percent reduction in course failures (grades of D or F or withdrawals). The new grant project seeks to disseminate PI throughout the computer science community by providing teaching materials, faculty mentoring, and measures for evaluating success.

The goal of PI is to help students not just learn facts or formulas but reach and retain a deeper understanding of ideas. In PI classrooms, instructors break up their lecture by asking multiple-choice questions carefully designed to engage students with the course’s core concepts. When a question is asked, students answer it on their own and then talk in groups before answering it again. The student responses are immediately used by the instructors to reshape classwide discussions and lecture material to address misunderstandings or knowledge gaps. With these in-class questions, Porter says, “I don’t grade for correctness. This allows me to pose difficult questions without penalizing students. Exams are for assessing correctness; class is for analyzing, discussing, and learning as a group.”

As Pierre-François Wolfe ’15 remarks, “It’s easy to follow a lecturer’s logic only to realize later that you are unable to reproduce it.” In Porter’s PI course, he says, “testing comprehension at every step of the way helped me better understand the material. Those questions that stumped the entire class were addressed in more detail by Professor Porter.” Rebecca McCourt ’14 says, “In the groups, if you really understood the topic, you would get to be the teacher and explain it to your peers, or if the concept was eluding you, your peers would explain their thinking and help you understand.” For her, “The interactive teaching style provided a perfect environment for actual in-class learning, instead of scribbling notes and cramming for exams.”

Porter, who teaches only with PI methods, reports that the vast majority of PI students say they’d like more professors to use it. That’s not always easy. “Dynamically adjusting lecture content based on student responses can be challenging,” Porter admits. “Also, developing really good questions is a lot of work,” which is why the new project supports preparation of questions and other course materials so that new PI instructors can adopt some ready-made, proven best practices.

“We hope to offer new PI courses nationwide by next fall,” Porter says. “We’re eager to see this practice become widespread, because the US needs more students in math, science, and engineering disciplines.”