Using Agent-Based Modelling for Preliminary Eer Experimental Design

Abstract

In this paper, we explore the use of agentbased modelling for preliminary engineering education research design. The motivation for this work is to help researchers understand the behaviour of the system being studied (e.g., classroom, program, etc.) before embarking on the, often, long process of an engineering education research study. We provide an example model of the effect of homework difficulty on homework completion and study group formation to demonstrate this approach. The results indicate this approach holds considerable promise as a preliminary modelling tool that can help focus longer-term engineering education research studies.

Using Individual-Based Modeling to Better Understand the Hidden Curriculum of Engineering

“…To mitigate this, they found creating PBL teams with the greatest diversity of students increased the sharing of experiences and knowledge, leading to stronger team dynamics and project outcomes [16]. Within our lab, we are using IBM to model cheating behaviours in engineering classrooms, where we were able to better understand the impact of harder assignments, a cheating culture, and poor prerequisites on engineering students' cheating behaviours [3,13].…”

“…Previously, we have described our approach for exploring the use of IBMs within engineering to provide an experimental method for easily testing research questions and interventions without physically using real participants and classroom environments [3,13]. Although IBMs are not a replacement for these, they provide a first step at understanding how individuals and the system might react to changes, which can inform experimental design.…”

Simulating Classrooms: Experience With Agent-Based Models in Engineering Education

“…Over the past two years, we have explored the application of agent-based modelling (ABM) to problems in engineering education. In particular, we have investigated the use of ABM as an engineering education research (EER) design tool [3,13]. Given that EER studies involve considerable uncertainty, and typically require sufficiently large sample sizes and timeframes to validate the effects of the study, there is always the risk that the time and effort put into the study can be wasted if errors are made in the early design phase of the study.…”

“…In this paper, we build on our previous papers [3,13] and share our experiences applying ABM to engineering education problems. Although ABM, a form computational modelling where a system's behaviour is modelled by software agents and their interactions [22], has been successfully applied to a range of natural science, engineering science, and social science problems, its application to the scholarship of teaching and learning is in the early stages.…”

“…As summarised in Table 1, these models include our original homework and cheating behaviour models [3,13], a model to explore summer camp parking logistics in the context of the COVID-19 restrictions, a set of models to explore COVID-19 transmission in the university environment [2,12,20], and models to explore student teams and student motivation [5]. Overall, these models provided us with test cases on the feasibility of using ABM to answer educational research questions.…”