Exploring the Influence of Team Cohesion on Teamwork in Software Engineering Education: The ASEST Framework

Engineering education is critical for the development of industry and hence of society in general in any country. Nowadays it is generally recognized that in order to be an effective software engineer in the modern software industry one requires skills for performing in a team. A large number of published papers confirm that teams have been widely researched in software engineering education (SEE). However, famous industry Chaos Reports have shown that software project success is around 30%. More than two decades have passed since Demarco and Lister, in their classic book “Peopleware, Productive Projects and Teams”, stated that when software projects fail it is generally more because of teamwork problems than technical issues. Their statement remains a current concern for progress discussion on human and social aspects of software engineering (SE). They showed then that for typical large software projects team working is about 70% of the project time. Nowadays this could be even higher because of the general use of agile methods by the software community as they base their success on self-managed teams. Accordingly, more attention should be paid to bridging the gap between what software engineers are learning about teamwork and what is required to properly perform in teams in the modern software industry.

Some authors have studied team performance in SEE. However, diverse conceptualizations have been used; and they all put emphasis on the process results rather than on the behaviors that allow software engineers to obtain team objectives. Thus, this dissertation addresses the following scientific problem: How to improve team performance behaviors in software engineering education? Cohesion is considered one of the most important factors that influence software teams. It has been found important for aspects like team effectiveness, motivation, productivity, team collaboration, and agile practices. Cohesion has also been found strongly correlated with performance.  However, cohesion has been scarcely studied in SEE and there is no empirical evidence on the relationship with team performance behaviors.

Since overlapping conceptualizations have been proposed by different scholars, in this research we examine both team performance outcomes (referred to as ‘team performance’) and team performance behaviors in terms of team learning.  The overall aim is to provide a teaching-learning framework named Agile Software Engineers Stick Together (ASEST) that aims to develop team cohesion, leading to better team learning and team performance of SE student teams. The process followed to address the objective of this research included two iterations. Iteration aimed at determining the basis of the framework, setting up a preliminary version (ASEST 0), and testing it. Three research questions are answered: RQ1: What current teaching and learning approaches can be used to set up the framework?; RQ2: Does the application of the ASEST0 improve perceived team cohesion, team learning, and team performance? And; RQ3: What are the students’ perceptions of ASEST0? Iteration 2 aimed at improving the preliminary proposal and validating the final framework (ASEST+). The following research questions are answered: RQ4: What are the cohesion antecedents for collocated teams in SEE?; RQ5: What are the most relevant identified antecedents for agile collocated student teams?; RQ6: What are the approaches in Agile Software Development regarding the relevant antecedents to improve the framework?; RQ7: What approaches regarding agile teamwork in SEE can be used to improve the framework learning strategies?; RQ8: Does the application of the ASEST+ improve perceived team cohesion, team learning, and team performance?; RQ9: Does team cohesion have a mediational role through the application of ASEST+? And, RQ10:  What are the perceptions of teachers on ASEST+?

The ASEST0 combines team-based learning, project-problem-based learning, and role-playing game learning strategies in three phases and eight steps. The first phase aims to establish the learning environment and prepare the students on team working skills. The core of the framework is the second phase with an establishment of an agreement on team rules that support communication and conflict management. The third phase focuses on adjusting the agreement via a self and peer evaluation of member contributions. A quasi-experiment of one group of students that applied the ASEST0 indicated that their positive perceptions of team cohesion, team learning, and team performance increased significantly compared with the perceptions of the students in a control group. A quasi-experiment of ASEST0 involving teams performing in a company setting also showed ASEST0 to be effective. A survey showed students’ acceptance of our proposal.

ASEST+ aims to solve the identified difficulties from iteration 1 and considers cohesion antecedents. It is built around Scrum teams and combines learning strategies to train students in collaborative and technical agile practices. ASEST+ establishes policies for role allocation and team rule agreements to regulate communication and address conflict management linked to agile practices. ASEST+ addresses personality, conflicts, and task interdependence as these antecedents are identified as the most important. A quasi-experiment showed that the use of ASEST+ significantly increases the students’ positive perceptions on team cohesion, team performance, and team learning compared with a control group. Another quasi-experiment over one group of students without the researcher’s participation replicated the intervention. It also showed ASEST+ to be effective. In addition, a mediation longitudinal analysis along this study confirmed cohesion to be a mediator between the antecedents personality, conflicts, and task interdependence and the outcomes team learning and team performance.

The main contributions of this dissertation are the identification of trends on teamwork in SEE, the identification of the relevant aspects affecting cohesion of SE student teams, the foundation of ASEST framework to develop cohesion for SE student teams, and experimental validation with ASEST framework. Although some limitations in terms of generalization should be considered, it is our expectation that the findings presented in this dissertation can contribute to delivering better-prepared software engineers to the industry and open new doors to future related research.