Assurance of Learning Through Testing

Programming education increasingly centres around tools that offer automatic marking of student progress through a set of exercises. Automated marking has become increasingly popular as it allows instructors to provide personalised feedback to students in real-time and is easily scalable to large cohorts. It can also be used to provide instructors with detailed insight into the class progression.

While this brings many new advantages it also creates new problems. Chief amongst those is the difficulty of marking long, dynamic and interactive programs which places a large burden on educators to develop sophisticated marking systems.

At the more extreme end of the problem, we may be faced with students who want to take shortcuts such as hard-coding, or using Python’s built-in functionality and third party packages, which may violate the spirit of the exercise. As a result these students may not be achieving learning outcomes.

In this talk we will discuss and demonstrate practical approaches to the aforementioned problems that all educators can use to simplify marking and to increase their assurance of learning. We will also discuss current limitations that we face and provide recommendations to those wishing to apply these approaches.