A hybrid simulation model of individual and team performance in software project environment

Abstract

Software development is a human- intensive effort, hence software project success heavily depends on performance of the members of its development team. A team member’s performance can be influenced by many factors, such as level of technical skill, motivation, stress level as well as team culture and organizational environment. Understanding the impact of these factors on software project outcome can lead to better project management and human resource management decisions, and as a result improve project success rate. However, it is a non-trivial task to evaluate such impact, due to difficulties in quantifying human behaviour in a complex environment such as that of software project, hence little literature exists on this research topic. This study attempts to explore the impact of individual and team performance on software project outcome with a simulation model that introduces human factors to a virtual software project environment. The simulation model includes the interplay of three aspects typical of a software project environment, namely: a) individual behaviour of each developer and communication among team members; b) software development workflow of activities and project management practices; and c) organizational environment including culture and HR practices. A hybrid model that includes the application of agent-based, system dynamics and discrete event approaches has been developed. Agent-based modelling (ABM) is one of the most suitable techniques for investigating the complexity of agents’ behaviour, actions, and interactions, thus allows us to observe the emergent behaviour of each individual agent; system dynamics (SD) modelling is suitable for observing, at an aggregated level, the cumulative and dynamic effect of varying conditions that influence the behaviours of model components; and discrete event simulation (DES) is suitable for capturing work flow under resource and/or other constraints. In our hybrid model, the agent-based component is developed as the core to simulate each team member’s behaviour and states of each unit of work, the system dynamics component is used to capture the cumulative mood of each developer while working as a member of the project team, and the DES component is used to simulate the integration process. Various experiments (“what if” scenarios) can be conducted by changing model parameter values and observe the impact of such changes on project outcome. A sample application is provided to demonstrate the impact of team performance under different levels of requirement volatility on a relative scale. This paper presents the design, development as well as a brief discussion on validation of such a hybrid simulation model. The research not only benefits those who are interested in understanding the impact of individual and team performance on software project outcome, but also those who are considering using hybrid simulation modelling approach to study human factors in other complex social-technical systems.