Capturing the effect of personality on teams with agent-based modelling
Continuing advances in information and communication technology (ICT) have changed the landscape of project management. Now there are increasing occurrences of short-term projects staffed by ad hoc assemblies of temporary team members who have been quickly recruited from a candidate population. However, there is little in the way of general guidelines available concerning how to manage these volatile situations. In particular there are no established approaches for more effective assembly of ad hoc project teams with respect to the collective psychological makeup of the team members. This thesis makes a contribution in this area by providing an examination into improved ways of assembling ad hoc project teams with respect to the psychological (personality) profiles of team members in order to produce more effective project outcomes. This thesis is divided into three main sections. In the first section, we investigate how the strategies that determine the composition of teams can affect team performance. Because of the autonomous nature of team members, we employ agent-based modelling techniques that can be used to predict the assembly of teams and their ensuing performance. Our agent-based simulations in the first section of this thesis demonstrate emergent effects based on different parametrisations. In order to compare the outcomes of these models with real-world situations, a practical method of simply determining individual personality types is needed. In this regard, we have used the Myers-Briggs Type Indicator (MBTI) index to identify personalities. In the second part of the thesis, we develop a team formation model to explain how self-assembly teams tend to evolve in the area of software development. In order to develop an agent-based model intended to predict the teams’ compositions, we describe our assumptions about the factors affecting team formation. A model is developed to explain the mechanism behind team formation and the extent to which our assumptions can predict the compositions of teams. Our model has been validated against a case study known as the “Python Enhancement Proposal” (PEP), which is used by small ad-hoc software teams to enhance the Python programming language. In order to discover the personality of a PEPs developer, we make an additional contribution in this thesis: that is, developing a novel model that infers the MBTI specification of personality from the candidate team members’ writing styles. By comparing PEPs data with the results produced from our agent-based simulations, we can identify the factors that explain the mechanisms behind team formation. In this study, we identified four significant input factors that affect team composition and performance: previous performance, teammate familiarity, MBTI Feeling personality, and MBTI Perceiving personality. The third part of this thesis focuses on the relationships between the personalities of a team and the team’s group performance. We introduced a data-driven methodology that can be customised for different organisations to discover the relationship between personality and team performance. In addition, we identified the team compositions that can result in better performance. One hypothesis that was tested and confirmed in this connection is the positive effect of personality heterogeneity on the performance of software development teams. The thesis makes several methodological and practical contributions. In this thesis, not only have we developed and tested how people do form into a team, but also we investigate how people should form into a team. The models and techniques developed in this thesis can be used to guide and help managers to investigate the assembly and evolution of temporary ad-hoc work teams. Managers can apply these models in connection with conducting various “what-if” analyses by simulating the behaviour of teams under different circumstances.
Advisor: Purvis, Martin; Purvis, Maryam; Bastin Roy Savarimuthu, Tony
Degree Name: Doctor of Philosophy
Degree Discipline: Information Science
Publisher: University of Otago
Keywords: Agent-based; modelling; Personality; MBTI
Research Type: Thesis