CS1: how will they do? How can we help? A decade of research and practice

ABSTRACT

Background and Context: Computer Science attrition rates (in the western world) are very concerning, with a large number of students failing to progress each year. It is well acknowledged that a significant factor of this attrition, is the students’ difficulty to master the introductory programming module, often referred to as CS1.

Objective: The objective of this article is to describe the evolution of a prediction model named PreSS (Predict Student Success) over a 13-year period (2005–2018).

Method: This article ties together, the PreSS prediction model; pilot studies; a longitudinal, multi-institutional re-validation and replication study; improvements to the model since its inception; and interventions to reduce attrition rates.

Findings: The outcome of this body of work is an end-to-end real-time web-based tool (PreSS#), which can predict student success early in an introductory programming module (CS1), with an accuracy of 71%. This tool is enhanced with interventions that were developed in conjunction with PreSS#, which improved student performance in CS1.

Implications: This work contributes significantly to the computer science education (CSEd) community and the ITiCSE 2015 working group’s call (in particular the second grand challenge), by re-validating and developing further the original PreSS model, 13 years after it was developed, on a modern, disparate, multi-institutional data set.

KEYWORDS:

• Introductory programming
• predicting programming performance
• interventions
• CS1
• attrition rates
• programming performance
• programming self-efficacy
• machine learning
• growth mindset
• artificial neural networks

Disclosure statement

No potential conflict of interest was reported by the authors.