Interaction and Feedback in Automatically Assessed Algorithm Simulation Exercises

Abstract

Executive Summary Feedback is an essential element of learning. Students need feedback on their work and their solutions to assignments both when they work manually and while they use a computer. A number of tools have been implemented to automatically assess and give feedback, for example, on programming exercises and algorithmic exercises. However, one problem of the provided feedback is that in most cases its scope is too narrow to support the needs of different types of learners. For example, many systems provide purely verbal feedback. In this paper we consider how exercises with automa tic feedback should be designed to support a broader scope of learners. We discuss the Felder-Silverman learning model, which we use as the framework for our discussion. The model categorizes learners with four different axes: sensing vs. intuitive learners, visual vs. verbal learners, active vs. reflective learners, and sequential vs. global learners. We discuss how all dimensions of the model can be taken into account when designing assignments and automatic feedback. Moreover, we use two modern automatic assessment systems, PILOT and TRAKLA2, as example systems to demonstrate our ideas. We strongly believe that incorporating analysis of learners' preferences into design of courses, automatic feedback systems, and learning environments leads to better learning. As teachers, we should better support the needs of our students, and also train their skills to process information in more versatile ways. Our discussion concentrates on algorithmic assignments. However, in the conclusion we briefly illuminate how similar approach could be used to design better assignments and feedback for programming exercises, as well.