PPIG 2020 - 31st Annual Workshop
Exploring the Coding Behaviour of Successful Students in Programming by Employing Neo-Piagetian Theory
Natalie Culligan, Kevin Casey
Abstract: We have collected data from approximately 300 students in their third-level first year Introduction to Programming module as they learn to write code using our in-house pedagogical coding environment, MULE. This data includes performance in lab exams and pseudocode questions, and data on code compiled, code run, and code evaluated, which we call CRE data. Evaluations are automatically graded and feedback is provided to students on their code. The student can only evaluate their code in the scheduled lab place and times but can evaluate as many times as they wish without penalty. The pseudocode questions are used to examine the students’ understanding of programming concepts, by removing the use of the compiler and comparing their performance in pseudocode questions to CRE data. Using a Neo-Piagetian framework, we examine pseudocode performance, lab exam performance and programmer behaviour in terms of CRE data. We investigate CRE data as signs of a student’s progression through the three stages of Piagetian understanding and build a series of Deep Neural Net binary classifiers to test if this passively collected behavioural data can be used to detect students in danger of failing.