Improving Undergraduate Computing Education through Peer Instruction and Interactive Ebook
Undergraduate computing education has several serious problems: skyrocketing enrollment which is reducing student support and access, high failure rates, and low diversity. How can failure rates be reduced with evidence-based methods that scale, especially when lectures are online? Decades of research on Peer Instruction, as defined by Mazur, have shown that it improves student engagement, learning, and retention; especially for underrepresented students. However, Peer Instruction has only been tested face-to-face. Research is needed on its effectiveness in online settings. In addition, most instructors who are aware of Peer Instruction do not use it due to time and cost concerns. However, many computing instructors plan to adopt ebooks, which provides an opportunity to leverage interactive ebooks to increase the adoption of Peer Instruction. This approach has the added benefit that interactive ebooks that contain executable code and practice problems with immediate feedback improve student outcomes over static ebooks.
In Peer Instruction, as defined by Mazur, students read before lecture and complete an assessment on the reading either before or at the beginning of lecture. During lecture the instructor displays a difficult multiple-choice question which students answer individually, then discuss with peers, and answer individually again. The instructor then presents the results of the two answers and discusses misconceptions. Most computing instructors create their own questions which takes a great deal of time, especially to create good questions. A good question is one that about 40% to 60% of the students answer correctly the first time. If good questions were easy to find, adoption rates should increase. A recent poll of computing instructors found that 25% of the respondents would adopt Peer Instruction if public resources were available.
This proposal will design, build, and test a system, Peer+, to identify and serve Peer Instruction questions in interactive ebooks. It will leverage free ebooks for the pre-lecture reading and reading assessment. It will also mine existing ebooks and public question banks to identify good questions, serve the questions in ebooks (in both in-person and online modes) to reduce the cost, and integrate the results from the questions in an innovative spaced-practice tool to test medium-term retention of knowledge. Workshops for undergraduate computing instructors will improve awareness of Peer Instruction, increase knowledge of best practices, and build a community of practice.
This research will increase knowledge about effective STEM education. It will investigate 1) the effect of online (both synchronous and asynchronous) peer discussion on learning and student satisfaction, 2) the medium-term effect of Peer Instruction on learning, 3) the effect of Peer+ on student retention, especially for underrepresented students, and 4) how instructor attitudes and knowledge change due to a workshop and use of Peer+. A design-based research approach will be used, based on theory, (social constructivism and expectancy-value) and contributing to theory, but also testing the system in real educational settings. Both qualitative and quantitative measures will be used to evaluate the research.
At least 2,300 students in two computing courses at the University of Michigan will use Peer+ over two years. Peer+ will be freely available on Runestone, an open-source ebook platform, with a creative commons license. Runestone offers 20 free ebooks for CS0, CS1, CS2, and more. It serves 25,000 users a day. The tool and research will be disseminated through blogs, articles, and talks at SIGCSE, CHI, LAK, and ICER. Underrepresented students are most at risk in computing because they are less likely to have prior programming experience. Improving student engagement and learning through interactive ebooks and Peer Instruction should reduce dropout and failure rates, especially students from underrepresented groups.
The grant was funded by the National Science Foundation. The amount of the award is
$297,257 over the project period.