A Holistic approach to assessing students' laboratory performance using Bayesian networks

Abstract

Laboratory activities have a central role in engineering teaching and learning. They are used to involve students in practical experimentations with concepts. This central role makes the ongoing challenge of performance assessment of students' laboratory work one of utmost importance. Laboratory work is traditionally assessed by the teacher marking students' written report of the laboratory activity. Assessment based on written evidence often overlooks the fact that laboratory work involves specific abilities/skills. The latter are highlighted by both QAA and ABET as part of engineering teaching and learning outcomes. There is a need to develop a model for laboratory performance assessment, which not only assesses what a student knows and understands about the concept addressed by a laboratory activity, but also the student's laboratory abilities/skills through analysis of the laboratory work process. This paper presents the work so far on the implementation of such a model. The model harnesses the strengths of Bayesian networks and integrates with a virtual electronic laboratory, the RealLab, which was designed and implemented in the course of this work