MAKING SENSE OF CONFLICT IN DISTRIBUTED TEAMS: A DESIGN SCIENCE APPROACH

ABSTRACT

Conflict is a substantial, pervasive activity in team collaboration. It may arise because of differences in goals, differences in ways of working, or interpersonal dissonance. The specific focus for this research is the conflict in distributed teams. As opposed to traditional teams, participants of distributed teams are geographically dispersed and rely heavily on computermediated communication. Understanding and managing conflict is a crucial task in these settings because it is at the crux of many other managerial issues (e.g., trust, leadership, and knowledge transfer). Conflict in distributed teams hides in asynchronous communication among team members. While this causes extra burdens in detecting and understanding conflict, it also makes it possible to rebuild conflict scenarios for the purpose of making sense of, and managing conflict in these teams.

The objectives of this study are to make progress toward an initial design theory in the form of a novel conflict detection and analysis approach; to implement a faithful instantiation of this evolving theory; and to employ it to gain insight about the conflict phenomenon in the new setting. The work is organized and reported in three essays.

Essay 1 develops a meta design of the approach (CM2) and describes its implementation as a novel conflict management system. The design builts on a new construct – the Vignette (vivid yet analytical, theory-laden descriptions of past experiences) – that describes conflict situations. A meta-model developed on basis of prior kernel theories underpins the conflict vignettes (the proposed conflict management approach) and CM2 (the implementation). The essay also describes an initial empirical pilot investigation that serves as the impetus for the study, and demonstrates the anticipated use of CM2 with a detailed use scenario.

Essay 2 is aimed at automating the process of extracting conflict information from computer-mediated communication (CMC) data, such as emails and instant messages. The work focuses on the basic action in conflict – argument – and develops an automatic argument detection solution. Drawing on the argumentation theory, I propose a model for argument detection composed of features that reflect five categories of argumentation functions including: announcement; reasoning; modality; transition; and, affect, along with another set of language features that are informative for recognizing arguments. The evaluation results show that the model achieves higher accuracy and recall in detecting arguments in message sets compared to baseline models.

Essay 3 extends prior design efforts to explore the conflict phenomenon in distributed settings. The work creates instruments to measure conflict elements as specified in the meta model and based on automatically detected argument information. Following a grounded theory approach, I employ the instruments to analyze 23,094 conflict situations in Bugzilla, an open source software development community. The analysis results reveal patterns associated with the occurrence of conflict, participants and their behaviors, and the temporal evolution of conflict. I then interpret these findings with the help of prior work to develop theoretical propositions that explain conflict in crowd collaboration settings that can provide directions for future research.

Chapter 1 

Introduction and Motivation

If two people on the job agree all the time, then one is useless. If they disagree all the time, then both are useless.                                                 

                                                                                                – Dale Carnegie

 

Research Background

Conflict is a pervasive phenomenon in the work place (De Dreu & Gelfand, 2008).

Conflict refers to a process that begins with an incompatibility between interdependent parties (Putnam & Poole, 1987; Rahim, 2010; Thomas, 1992) that may arise because of differences in goals, differences in ways of doing things, or because of interpersonal dissonance (Jehn & Mannix, 2001; Rahim, 2010). It is then manifested as behavior that seems to adversely affect the parties in the conflict as well as the surrounding organization. Scholarly work sometimes describes conflict as the “dark side of the workplace” (Raver & Barling, 2008) because it produces tension and antagonism, and distracts conflict parties away from their assigned tasks –  which in turn reduces team performance and individual satisfaction (Hackman & Morris, 1975; Pondy, 1967; J. A. Wall Jr & Callister, 1995). However, it has been recognized that conflict also has functional, positive impacts, and these impacts have been getting more attention. This work shows that a moderate amount of conflict can, in fact, encourage different perspectives, stimulate innovation and creativity, and improve decision making (Levine, Resnick, & Higgins, 1993; Nemeth & Staw, 1989; Tjosvold, 1997; Tjosvold & Johnson, 1977) while preventing teams from succumbing to stagnancy, mediocrity, and group-think (Nemeth, 1986). These potential positive influences of conflict – along with the realization that conflict is inevitable – suggest that managing conflict is a critical problem in organizations.

Contemporary responses to this problem range from “best managerial practices” that seek to avoid conflict and resolve its negative effects (APM, 2006; PMI, 2004), to popular writings that delineate the functional impacts of conflict, propose productive interaction, and suggest negotiation strategies (Eisaguirre, 2002; Joni & Beyer, 2009; Weiss & Hughes, 2005). The scholarly community has contributed to the discussion with investigations of conflict, generic strategies for conflict management, and by describing the factors that matter in conflict process (Jehn & Mannix, 2001; Rahim, 2010; Thomas, 1992). While some social systems, including organizational structures and conflict report and resolution mechanisms, have been proposed for conflict management (Yarn, 2014),  the application of information technologies has not received as much attention from the scholarly community, and even less from researchers from the information system communities.

Research Setting and Problem Statement

This research deals with conflict in the workplace, as opposed to the conflict that occurs inside families or between social groups. In particular, I select distributed teams as the focal setting for the research because managing conflict is crucial for realizing the potential of these teams and preventing project failure (Hinds & Mortensen, 2005; Montoya-Weiss, Massey, & Song, 2001). The availability of archived communication data between team members also provides the opportunity to explore conflict management approaches in these settings.

Distributed Teams

A distributed teams is defined as a group of people who share a common purpose, carry out interdependent tasks, work across space and time, and interact primarily through computermediated communication (Cramton, 2001; Majchrzak, Malhotra, & John, 2005). Driven by the global economy and enabled by information and communication technologies (ICTs), distributed teams have become a popular organizational form in contemporary companies, especially in hightech industries that employ knowledge workers scattered around the world (Berry, 2011a; McDonough III, Kahn, & Barczak, 2001). Examples of distributed teams include new product development teams (McDonough III et al., 2001), global software development teams (Espinosa & Kraut, 2002; Vlaar, Fenema, & Tiwari, 2008), and open collaboration communities (Filippova & Cho, 2015; J. Wang, Shih, & Carroll, 2015).

In spite of the fact that an increasing number of projects are conducted through distributed teams, these teams are often found to be difficult to manage and fall short of performance expectations (Hinds & Mortensen, 2005). Previous studies show that distributed teams are subject to coordination problems (Cramton, 2001), crises of trust (Jarvenpaa & Leidner, 1998), unhealthy team dynamics (Armstrong & Cole, 2002), and severe conflicts between team members (Mannix, Griffith, & Neale, 2002). While these problems are also observed in collocated teams and can be explained by generic models, it has been argued that separate studies are required for distributed teams because of their unique characteristics and dynamics (Hinds & Mortensen, 2005). To help realize the potential benefits of distributed teams, many studies have been conducted to investigate the salient managerial issues that tend to be found in distributed teams, such as developing trust between team members, establishing leadership, enhancing knowledge transfer, and managing conflict (See Table 1-1).

The present research focuses on conflict and conflict management in distributed teams for three reasons. First, conflict is inevitable in these teams. Distributed teams are composed of members who are drawn from different departments or organizations; have various knowledge and cultural backgrounds, and who often know very little about each other at the outset. Conflict arises when they make decisions about the goals, strategy, and schedule for their work (Cramton, 2001), and when misunderstandings are not successfully resolved through electronic communication (Armstrong & Cole, 2002).

Table 1-1 Managerial Issues in Distributed Teams

Issues	Descriptions	Approaches	Example References
Trust	Trust building is inhibited by geographical, temporal, and technical constraints in distributed teams	•       Initial trust •       Trust transition •       Reward	(Rousseau et al., 1998) (Berry, 2011a) (Kuo & Yu, 2009)
Leadership	Leadership becomes particularly complex in distributed environments, (e.g. physical location and number of leaders, process used to determine leaders, the presence and control of followers).	•       New roles •       Initial pressure	(Ocker et al., 2011) (Zigurs, 2003) (Yoo & Alavi, 2004)
Knowledge Transfer	It is hard to achieve knowledge sharing and integration in distributed teams due to stickiness of information, unique local working norms, and limited opportunities for discussion and debate.	•       Shared passion •       Rapport •       Knowledge repository	(Lee et al., 2014) (Oshri et al., 2008)
Conflict	Conflict in distributed teams tends to be detrimental and resist resolution due to lack of shared identity and context, and because of information diversity.	•               Swift trust •               Spontaneous communication •            Coordination Mechanisms	(Cramton, 2002) (Hinds & Mortensen, 2005) (Mannix et al., 2002)

 

Second, managing conflict is at the crux of addressing other managerial issues in distributed teams. Previous studies have identified a variety of managerial issues that tend to be salient and unique in distributed teams. Conflict has been repeatedly identified as one of the most important concerns in solving other managerial problems. For example, conflict inhibits the development of trust in distributed teams. Severe and escalated conflict between distant team members can weaken their confidence in the project and ferment tension between them, which in turn can prevent the development of trust (Berry, 2011b; Kuo & Yu, 2009). Conflict also causes obstacles to knowledge transfer in distributed teams because it impairs rapport and communication among team members (Kotlarsky & Oshri, 2005; Lee, Bassellier, & Faraj, 2014). Monitoring and resolving conflict therefore represents an important task that the leader of a distributed team must consider in order to maintain optimal team functionality, as well as building morale and relationships within the team (Mannix et al., 2002; Ocker, Huang, Benbunan-Fich, & Hiltz, 2011).

Third, evidence shows that conflict is not only detrimental in distributed teams, but particularly difficult to cope with when compared with collocated teams (Bergiel, Bergiel, & Balsmeier, 2008; Cramton, 2001; Hinds & Mortensen, 2005; Montoya-Weiss et al., 2001). Several factors make managing conflict in distributed teams challenging. One is that members in distributed teams tend to misunderstand each other and take a competitive stances because of a lack of shared identity and shared context (Espinosa & Kraut, 2002; Mortensen & Hinds, 2001). Another factor is that conflicts in distributed teams might not be easily handled by traditional organizational structures and managerial approaches (De Dreu & Gelfand, 2008; Rahim, 2010). For example, “appeal to authority” (a traditional way to solve conflict between peers) is invalidated in distributed teams since they consist of members from different departments who have different reporting lines. The third reason is that conflicts in distributed teams are hidden in asynchronous and discontinuous electronic communication streams. Constrained by time and attention, participants are likely to overlook conflict information, to lose track of the progress of a given conflict, and to take inappropriate, even counterproductive action (Kahneman, 2011; Thomas, 1992).

In response to these challenges, a variety of approaches and mechanisms haven been proposed to manage conflict in distributed teams. Examples includes developing swift trust between team members (Mannix et al., 2002), creating transative memory within the team (Hinds & Mortensen, 2005), and enforcing coordination mechanisms (Montoya-Weiss et al., 2001).

While informative and useful, these approaches are subject to three limitations. First, they focus on preventing negative conflict, but have nothing to say about embracing and exploiting productive conflict; some level of conflict is necessary to elicit different perspectives, stimulate creativity, and improve decision making in distributed teams. Second, these approaches are rarely tied to the generic conflict management knowledge accumulated in previous studies. Therefore, the effectiveness of these approaches is in doubt since they do not directly tackle problems known within conflict management practice. Third, these approaches fail to incorporate IT support. Ironically, although technology provides the foundation for distributed teams, advanced applications (such as analytical tools for supporting conflict management) have rarely been investigated and tested to deal with conflict in distributed teams. In the absence of such applications, the difficulty in tracking and understanding conflict increase exponentially as conflict parties are overloaded with electronic communication data.

Conflict and Conflict Management

Organizational conflict has been intensively studied in the fields of organization, psychology, and communication (De Dreu & Gelfand, 2008; Putnam & Poole, 1987; Thomas, 1992). These studies have generated a series of theories that help to understand and manage conflict in organizational settings.

Understanding a conflict starts with examining its cause. Prior work suggests three broad types of conflicts: task, process and relationship (Jehn & Mannix, 2001). Task conflicts arise because of incompatible viewpoints and opinions pertaining to essential ideas of the task, such as its goal, logic, and value. Process conflicts are triggered between people who have different viewpoints and opinions pertaining to duty, resource delegation, and schedule. Relationship conflicts can be traced to adverse emotions and feelings between people. Understanding the cause of a conflict is important because it can both predict the outcome as well as suggest strategies for mitigating the consequences. For example, prior studies show that task conflicts have the potential to lead to functional outcomes while process and relationship conflicts often lead to dysfunctional consequences (Jehn, 1995).

Conflict also needs to be understood and managed as a process. Conflict may be conceptualized following a sequential model, or a more nuanced, punctuated model. The punctuated process model emphasizes events that occur as a conflict unfolds. In line with the punctuated equilibrium theory, the punctuated process model describes the conflict process as alternating between stasis and transition (Gersick, 1988; Putnam, 2004). Here, a transition refers to a short period of time when a conflict shifts substantially in terms of its intensity, or its progress toward resolution. A transition is characterized by the behavior and emotions of the parties in the conflict. The conflict escalation theory suggests that a conflict may undergo a combination of two types of transitions – conflict escalation and conflict de-escalation (Pruitt, 2008). Conflict escalation occurs when a conflict is intensified with mechanisms such as more extreme strategies, emotional involvement, issue generalization, or through an increased number of participants. Conflict de-escalation occurs when conflict parties re-assess the conflict situation and change their language and behavior to more moderate and collaborative.

Knowing the roles played by conflict parties is a precondition for understanding their behavior. The parties involved in a conflict can be broadly classified into two categories: principal conflict parties and third parties. The term ‘principal conflict parties’ refers to individuals or groups who are interdependent with one another in the pursuit of their interests in spite of the incompatibilities between them (Putnam & Poole, 1987). The term ‘third parties’ refers to individuals or groups who intervene in the conflict either to fulfill their duties or because they have a stake in the conflict (Pruitt & Kim, 2004). Understanding a conflict situation, thus, requires recognizing the perspectives and actions of both principal conflict parties as well as third parties who are affected by the conflict.

Prior work also outlines high-level conflict management strategies to characterize the general intention of the principal conflict parties toward a conflict and the approaches they use for coping with the conflict (Thomas, 1992). Scholars have conceptualized conflict management strategies in a space defined in two dimensions – concern for self and concern for others. Combining these two dimensions results in five styles or strategies that the principal conflict parties can use: collaborating (H-H), compromising (M-M), yielding (L-H), contending (H-L), and avoiding (L-L) (Olekalns, Putnam, Weingart, & Metcalf, 2008). A conflict management strategy, thus, represents a composite construct that characterizes the general orientation of a conflict party, not a specific action.

Conflict management is a task full of uncertainty. Although existing conflict, and conflict management theories provide basic guidelines for understanding and coping with conflict, specific conflict management approaches that can help practitioners put these theories to practical use are still needed. For example, conflict management strategies provide high-level explanations of conflict behavior. However, there is lack of discussion about how to apply these strategies in practice. One important reason is that conflict party behavior is influenced by a very large number of internal and external factors. The relationship between these factors and the resulting behaviors cannot be captured in structured and un-ambiguous rules. The absence of this kind of knowledge has lead toa gap between research and practice.

IT Support in Conflict Management

It is possible to conceptualize managing and resolving conflicts – like other cognitively demanding tasks – as a task that may benefit from technological support. In spite of the inherently messy and often unpredictable nature of conflict, scholars have suggested such uses of information technology for conflict management. These prior scholarly efforts have developed two genres of approaches for managing organizational conflicts – communication-driven and decision-driven.

Communication-driven approaches are aimed at facilitating interactions between the conflict parties and supporting information processing activities. The implementations of these approaches are often directly adopted from group decision support systems (GDSS) (Chiravuri, Nazareth, & Ramamurthy, 2011; El-Shinnawy & Vinze, 1998; Sambamurthy & Poole, 1992) or

Negotiation Support Systems (NSS) (Delaney, Foroughi, & Perkins, 1997; Johnson & Cooper, 2009; Wolfe & Murthy, 2006). These systems are adopted to provide features such as anonymous posting (El-Shinnawy & Vinze, 1998; Sambamurthy & Poole, 1992), procedural enforcement (Chidambaram, Bostrom, & Wynne, 1990), problem visualization (Chiravuri et al., 2011), team organization (Miranda, Bostrom, & Robert, 1994), and communication channels (Anson & Jelassi, 1990; Wolfe & Murthy, 2006). These features have been shown to contribute to encouraging collaboration, reducing conflict, and facilitating conflict resolution, although they are not entirely free of negative impacts (Chiravuri et al., 2011; Johnson & Cooper, 2009). However, these GDSS and NSS approaches tend to support deliberations within groups that are collocated, and are not primarily designed for longer-duration decision-making or negotiation. Although they may (arguably) be extended for distributed teams, their applicability remains open to question.

Decision-driven approaches refer to those that provide useful information to conflict parties to better understand the conflict at hand and to make decisions. Implementations of these approaches provide features such as visualization (Flack & Summers, 1971; Hine & Goul, 1998; Taratoukhine, 2002), and identifying potential and optimal resolutions (Fraser & Hipel, 1981; Klein, 1991; Madani, Rouhani, Mirchi, & Gholizadeh, 2014). Like the features outlined for the first category (communication-driven approaches), these features have also been shown to contribute to conflict resolution (Hamouda, Kilgour, & Hipel, 2006; Klein, 1993). However, these approaches are often aimed at only a small component of the overall conflict process (e.g. the decision phase), and do not help users in tasks such as detecting and tracking conflicts or exploring different conflict resolution strategies. Much like the approaches in the first category, it is possible to argue that the approaches in the second category may be extended to cover more conflict phases and new conflict situations. However, their applicability also remains limited and open to question.

The two genres of approaches, thus, have some inherent limitations. Each provides only a fractional and static view of conflict. Neither incorporates an underlying model that can represent the entire conflict process. These critiques are not surprising because both approaches were not originally designed to support the management or resolution of conflict. Rather, they have been appropriated for conflict management from their original domains – coordination and decisionmaking. Because of this, they do not directly account for or conceptualize ideas such as events that occur as a conflict situation unfolds and the different behaviors of conflict parties. As a result, these systems cannot provide users the ability to understand diversity across conflict situations, or to make effective conflict management choices based upon them.

The limitations described above shows that the design of conflict management systems and approaches remains understudied in the literature. One key obstacle constraining the development of two genres of approaches and other novel approaches, to my observation, is the difficulty in locating data sources that capture a wide range of conflict information. In the traditional conceptualization, the negotiations through which conflict arises and is resolved are conducted via fleeting verbal dialog between the conflict parties. It is not only technically difficult to capture these negotiations into usable data, but also socially inappropriate to do so as conflict is usually considered as an embarrassing and sensitive topic (Jehn, 1997; Raver & Barling, 2008). Without data support, the development of conflict management approaches has been concentrated on well-understood conflicts, processes, or solutions of that could be codified and optimized. Social and behavioral factors have been excluded from the design. However, the generation of CMC data in distributed teams provides an opportunity to overcome this obstacle and to design new conflict management approaches that can be used not only to broaden the reach of current conflict management practice, but to address the specific problems arising within distributed teams.

Research Objectives and Questions

The present research opportunity lies at the intersection between conflict studies and distributed team studies. As detailed in the literature review, managing conflict is a critical task in distributed teams because these teams are more subject to detrimental conflicts. Understanding and managing conflict is necessary to prevent the negative effects of these conflicts, and is at the crux of addressing other cooperative issues within teams. Although CMC tools lay the foundation for distributed teams and generate a large amount of rich data in practice, the data has not been exploited to support participants and researchers in making sense of conflict because of a lack of analytical tools. From the other side, prior conflict studies have generated a great deal of knowledge and many models relating to conflict and conflict management in organizations, but, the design of conflict management systems has rarely incorporated them. This is a result of the fact that system design is hindered by lack of data sources that can capture the process of conflict (such as the events that occur as a conflict unfolds, and the action taken by conflict parties to tackle the conflicts). I contend that investigating novel artifacts to extract and make sense of conflicts that occurred in distributed teams will contribute to both fields of study.

 

The objective of this study is to design, implement, and apply a design theory in the form of a novel conflict detection and analysis approach that can support people in making sense of conflicts that occurred in computer-mediated communication. To fulfill the objective, this research will be conducted to answer a set of research questions.

RQ1: What is the information necessary to understand a conflict? How should the information be organized and presented to promote sense-making in conflict situations?

RQ2: How to extract conflict information from communication data?

RQ3: How to apply the proposed approaches to analyze conflict situations?

Structure of This Proposal

This proposal is structured as follows. In Chapter 2, I first review literature about conflict and conflict management to develop a generic and systematic view of conflict, followed by the discussion about its uniqueness in distributed teams. I then turn to an investigation of existing approaches that have been proposed for conflict management in order to explore shortcomings in the system design. In Chapter 3, I describe design science research (the methodology used for this research) and discuss the philosophical foundations of the methodology. After that, I outline the detailed research plan that was followed to organize the present research. In Chapters 4 to 6, I develop three essays, each of which answers one of the research questions. Specifically, Chapter

4 describes the proposed novel conflict management system (CM2) and its implementation. Chapter 5 describes a methodology aimed at automating the process of extracting conflict information from raw communication data. Chapter 6 presents the results of the analysis that I conducted, and extends prior design outcomes to understand the conflict phenomenon in Bugzilla, an online collaboration community. In Chapter 7, I conclude the dissertation with a discussion of the research outcomes.

MAKING SENSE OF CONFLICT IN DISTRIBUTED TEAMS: A DESIGN SCIENCE APPROACH

Sharing is caring!