One of the main features of STACK is the potential response tree, which can be used not only to determine correct and incorrect answers to questions, but also to classify wrong answers.

We have attempted to analyse the process of answering a question, i.e., what kind of wrong answers students went through before arriving at a correct answer, by using this function of classifying wrong answers.

Kurihara et al. represented the answer process of a student as a directed graph, in which the correct and wrong answers are nodes and the nodes are connected by directed edges, and visualised the overall tendency of the answer process by overlaying the directed graphs of all students in the class.

However, there is a drawback that detailed temporal information is unknown.

We compensate for this by analysing the tendency towards correct answers by visualising the time variation of the types of wrong answers.

We also want to investigate how feedback influences the students’ answer process.

## Resources

- Slides (.pdf)