NSF Award
2434362
The University of Chicago, University of Illinois, and Utah University seek to transform online computer science theory education by developing exercises delivered by online tools to provide more equitable, inclusive learning experiences for a broad set of students. Theory courses teach critical skills that help software engineers write efficient code, allowing them to optimize for saving energy, speed, reliability, etc. Undergraduate computer science instruction is utilizing an increasing amount of online learning in a variety of ways, including online homework activities only, online lectures with access to in-person office hours, and fully online courses. While introductory coding instruction has made great strides in developing exercise types amenable to online instruction, theory education lags behind. Algorithms and discrete math courses have long depended on hand-graded, large start-to-finish homework exercises, hindering the quality of its online instruction. If successfully developed and integrated into computer science instruction, such innovative solutions will increase student success in obtaining computer science degrees, especially students who are less confident in their abilities. This Computing in Undergraduate Education Transformation project will improve career outcomes for students from underserved populations as well as build a stronger computing workforce.<br/><br/>This project will explore the design and use of online homework problem types for theory instruction integrating several attributes: instant, automated feedback, isolated skills, adaptive complexity, and culturally responsive contexts. First, the project team will explore isolated skill exercises, inspired by Parsons problems in coding and recent Proof Blocks in proof-building. These problem types will focus on individual skills rather than the entire problem-solving process. The purpose is to improve both student confidence and skills. Second, the project will explore adaptively scaffolded sequences, responding to student mistakes through gradually easier problems that scaffold their learning of an isolated skill. The purpose is to improve both student confidence and skills. Third, the project team will develop culturally competent variations, moving away from symbols and mathematics to real-world problems. After developing a robust set of base problems, a Large Language Model (LLM) will be used to produce equivalent variations of those base problems, all with different contexts. The purpose of these problems is to improve student belonging, engagement, and skills. When analyzing challenges and strategies, the project will recruit diverse participants and analyze data intersectionally, providing case studies on the experiences of students from multiple populations historically marginalized in computer science to augment traditional quantitative methods across the entire population. Through the three phases of the project -development, piloting, and evaluation, the project team will employ several techniques. During development, they will utilize think-alouds and focus group member checks. During pilot sessions, they will collect detailed automated data on student behavior, surveys on student reactions, and student submissions for student performance. During evaluation, they will run a quasi-experimental comparison study between classrooms using the new exercise types and classrooms that do not. These strategies will improve career outcomes for students from underserved populations as well as build a stronger computing workforce.<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
