Pencil Puzzles as an Inclusive Domain for Learning Computer Science

Zack Butler
Rochester Inst of Tech

Need:It has been shown that introductory computing assignments with a framing context are generally beneficial to students. However, it is also important to expand the types and number of contexts for faculty to use, and to study these in universities with different student demographics and course delivery models. In particular, in Computer Science, faculty opt to teach with a variety of different approaches and languages, and students come with widely varying backgrounds, both personally and academically, and we wish to learn what type of contexts are appropriate across this variability.Guiding Question:Our goal is to study the use of one particular context, namely pencil puzzles, at a variety of institutions with different student demographics. While pencil puzzles were shown to have effects largely independent of student gender and prior experience, these results were based on implementation at a single university. Expanding this study to other institutions using the same assignment would be a study of the assignment rather than the entire contextual domain. Therefore, we recruited instructors at eight institutions and worked with them to design a pencil-puzzle-based assignment suitable for their particular introductory-level course. The assignments shared the context that we aim to study: pencil puzzles as an instructional domain encouraging computational thinking. We evaluated the assignments using student grades and survey responses regarding student perceptions of the assignments including self-assessed learning. Outcomes:Our initial analysis focused on two statements in the survey: “I appreciated this assignment as a learning experience.” and “I felt that this assignment helped me to learn this week’s material.” We examined these results against student demographic variables including gender, ethnicity, interest in puzzles, prior programming experience, and more. Since different assignments were used in different courses, we chose to use a mixed-effects linear regression, which allows us to use all of the data in a single analysis, using the demographic variables as fixed-effect variables and the course ID as a random grouping intercept. This factors out the course-to-course variations and illuminates any demographic dependencies that are present across the entire data set. Our initial analysis surprisingly showed no significant effect of the course ID, implying that different assignments did not produce significant variability for these questions. while there was some dependency of student responses on their prior programming experience, and female students’ feedback were more positive about one aspect, overall these types of assignments do not appear to put particular groups of students at a strong (dis)advantage.Broader Impacts:As Computer Science grows and continues to try to broaden its student body across gender and ethnicity, it is important to understand the implications of the ways that we teach it. Our work seeks to address these questions through an abstract, though approachable, context. So far we have seen very little impact of demographics on the effectiveness of or student appreciation for this type of assignment, and we are also looking to study how easily it can be adopted by diverse faculty going forward.


Zack Butler, Rochester Institute of Technology, Rochester, NY; Ivona Bezakova, Rochester Institute of Technology, Rochester, NY; Angelina Brilliantova, Rochester Institute of Technology, Rochester, NY; Kimberly Fluet, University of Rochester, Rochester, NY;