Session 9: Supporting CS Education

Andrea Adamoli

Abstract: While the main advantage of CIs is responsiveness in analyzing answers, their measurement speed (the number of misconceptions that can be tested in a time frame) is poor. In this paper, we introduce an improved CI methodology that allows for accurate detections of broader sets of misconceptions in classes and, thus, to obtain detailed pictures of the students’ difficulties. Our methodology is based on observing the answers to the individual options of the multiple-choice questions, rather than to the question as a whole. We are therefore able to get more information from an item and thus improve the measurement speed. We integrated our methodology in a CS1 Java Programming Language course, and we tested 89 distinct misconceptions in 15 sessions of 30 minutes. Our methodology showed a 4x speed up in the measurement speed compared to state-of-the-art CIs, while preserving satisfactory accuracy. Thanks to this extensive coverage of misconceptions, we built a new metric, the “knowledge fitness”, to objectively measure student difficulties.

📄 Read More: https://link.springer.com/chapter/10.1007/978-3-031-42682-7_1

Aleksandar Karakaš and Denis Helic

Abstract: Students’ previous knowledge and self-regulated learning are important predictors of academic success. A growing body of literature studies these predictors with respect to introductory programming courses. Especially in the first semester, cohorts exhibit a wide range of backgrounds with many students having no prior programming experience at all. Furthermore, many first semester students lack self-regulated learning capabilities. In the light of high drop-out rates in introductory programming courses, it is thus crucial to also consider student characteristics, such as previously acquired programming skills or self-regulated learning capabilities. In this work, we collect data on such student characteristics via surveys and investigate the relation between survey data and student’s use of a version control platform during a first semester programming course at a European university. We also relate the survey data to the number of test cases students pass in their assignments. Using random forests, we investigate, how version control data can be used to predict students‘ success in an assignment and to what extent additional survey data can improve such predictions. Our results show that especially in an early phase of an assignment, data on previous knowledge and self-regulated learning can help predict students‘ success.

📄 Read More: https://link.springer.com/chapter/10.1007/978-3-031-42682-7_10