Global virtual teams coordination mechanisms: building theory from research in software development

ABSTRACT

The advent of globalisation has led to the growing use of global virtual teams (GVT) for software development. The use of such teams enables organisations to operate across national, economic and social, and cultural boundaries; this new form of teamwork presents challenges for traditional coordination mechanisms. Hence, a range of new operational problems for the coordination of software development teams have emerged due to the nature of virtual work: these are related to issues of geographical distance, language differences, time zone(s) differences, cultural differences, and trust. This paper applies a theoretical model drawn from prior research to explore the coordination mechanisms employed by a global virtual software development team in a major multi-national telecommunications organisation. The study analyses the impact that the aforementioned issues have on the effectiveness of project team coordination mechanisms and then develops a refined conceptual model to guide future research on global virtual software development teams. The findings also inform practice on the problems encountered in ensuring the effective coordination of such teams.

KEYWORDS:

• Global virtual teams
• coordination mechanisms
• software development
• project management

1. Introduction

In recent years, global virtual team (GVT) use for software development has become standard practice for most organisations; Jimenez et al. (2017) for example, highlight the increased incidence of global software development projects. A GVT is a specific type of virtual team which is typically geographically, temporally and organisationally dispersed, and, also, culturally diverse (Fleischmann, Aritz, and Cardon 2020; Maznevski and Chudoba 2000; Ramachandran 2005; Zahedi, Shahin, and Babar 2016). While GVTs are not a new phenomenon, advances in the reach and range of information and communication technologies (ICT), such as Slack, MS Teams, Zoom etc. (Stray, Moe, and Noroozi 2019), coupled with the trends toward globalisation, facilitate and motivate the use of GVTs in organisations, particularly in software development projects (Boyer O’Leary, Wilson, and Metiu 2014). However, while the use of GVTs for software development has rapidly become commonplace, research on certain aspects of GVTs (planning and strategy, creativity, subgroup formation, new ICT technologies, etc.) has received relatively little attention (Gilson et al. 2015). In addition to globalisation and ICT advances, other driving forces behind the growing use of GVTs include the need for cost reduction, increased competitiveness, and the possibility to share resources on a global scale (Kroll et al. 2018; Levina and Vaast 2008; Zahedi, Shahin, and Babar 2016). Thus, software development (SD) teams that are globally dispersed rely on: (i) shared knowledge; (ii) shared assets; and (iii) the utilisation of the unique skills and capabilities of a variety of team members (Maynard et al. 2019; Nordbäck and Espinosa 2019; Vlaar, Van Fenema, and Tiwari 2008). The possibility of team-working over twenty-four hour days (Kroll et al. 2018), optimum team selection (Vallon et al. 2018), and utilisation of local knowledge are attractive to organisations (Hanisch and Corbitt 2007; Kim and Roberts 2019); however, global virtual SD teams give rise to operational problems that can override the potential benefits (Kotlarsky and Oshri 2005; Sangaiah, Subramaniam, and Zheng 2015).

Prior research on GVT SD projects highlight the following issues (Table 1) as being important to their management: cultural differences (Dubé and Paré 2001; Maznevski and Chudoba 2000; Paul, He, and Dennis 2018; Sarker and Sahay 2004; Zakaria 2017), communication technology (Bartelt and Dennis 2014; Gilson et al. 2015; Majchrzak et al. 2000; Petter et al. 2020), geographical distance (Carmel and Agarwal 2001; Espinosa et al. 2003; Herbsleb et al. 2000; Zahedi, Shahin, and Babar 2016), language differences (Dubé and Paré 2001; Fleischmann, Aritz, and Cardon 2020; Gilson et al. 2015), team leadership (Kristof et al. 1995; Nordbäck and Espinosa 2019), time zone differences (Cramton 2001; Herbsleb et al. 2000; Jiao et al. 2016), knowledge sharing amongst dispersed team members (Alsharo, Gregg, and Ramirez 2017; Cramton 2001; Griffith, Sawyer, and Neale 2003; Kanawattanachai and Yoo 2007; Kotlarsky and Oshri 2005; Sangaiah et al. 2017), shared identity amongst team members (Hinds and Cramton 2014; Medappa and Srivastava 2019; Boyer O’Leary, Wilson, and Metiu 2014) and trust between team members (Alsharo, Gregg, and Ramirez 2017; Jarvenpaa and Leidner 1999; Maznevski and Chudoba 2000).

Table 1. Global virtual team issues.

Cultural differences

Communication technology

Geographical distance

Language differences

Team leadership

Time zone differences

Knowledge sharing

Shared identity

Trust

The issues identified in this body of research lie in stark contrast with those identified in traditional co-located SD teams: for example, such SD projects are characterised by problems relating to budgets, schedules, and system quality (Ewusi-Mensah 2003; Johnson 2000; Standish Group 2004). Thus, the findings of previous research attribute the causes of failure to poor project management (Lauesen 2020; Tesch, Kloppenborg, and Frolick 2007; Saleem 2019; Standish Group 2004; Wallace and Keil 2004). It is clear from the research cited above that the increased use of GVTs and the development of increasingly complex software artefacts have combined to bring new challenges to the SD environment (Sangaiah et al. 2018). In particular, the challenge of how such team work might be effectively coordinated remains very much an open question (Chamakiotis et al. 2020; Dubé and Robey 2009; Morrison-Smith and Ruiz 2020; Watson-Manheim, Chudoba, and Crowston 2012). GVTs, by their very nature, create issues for coordination of work (Medappa and Srivastava 2019; Watson-Manheim, Chudoba, and Crowston 2012; Powell, Piccoli, and Ives 2004). When that work necessitates high levels of coordination for successful project outcomes, such as SD, the optimal approach to coordinating the team is not always clear.

While prior research has explored the challenge of coordination in GVTs, there has been minimal research at the level of coordination mechanisms (McBride 2008; Nguyen-Duc, Cruzes, and Conradi 2012). Therefore, there is a limited understanding of the role of different types of coordination mechanism and the specific impact of GVT issues. A coordination mechanism may be defined as any administrative tool for achieving integration among different units within an organisation (Martinez and Jarillo 1989). A research study that focuses on the impact of GVT issues on coordination mechanism efficacy is warranted given (1) that coordination mechanisms are the medium through which coordination is achieved; and (2) existing research found coordination to be particularly problematic in GVTs (and specifically software development work) due to the presence of various GVT issues.

Approaching the coordination issue through a mechanism-based lens may offer insights into GVT coordination. Indeed, given the current global landscape (where traditional colocation and/or in-person site visits are not available to address the coordination shortfalls that tend to exhibit in GVTs), consideration of different types of coordination mechanism effectiveness in this context is of critical importance. Therefore, the objective of this study is to investigate the use of traditional coordination mechanisms in GVT SD projects and the impact that GVT issues such as geographical distance, trust, language differences, cultural differences and time zone differences have on their effectiveness. In doing so, the study contributes to the body of knowledge as it uncovers the ineffectiveness of specific categories of coordination mechanism in GVTs. Further, the study contributes to practice by highlighting the gaps in knowledge that may be addressed for GVT coordination as well as offering practitioners clarity on the specific mechanisms that should be tailored or substituted for GVT SD projects. To facilitate this objective, a conceptual model with associated propositions will be built from existing literature. The resultant model will be used as a synthesising lens through which to study the phenomena and derive a set of hypotheses for further research. We begin the conceptual model development in the proceeding section by examining existing knowledge and opportunities for advancing that knowledge. This is followed by a theory-building section that presents a conceptual model developed from prior research studies. We then use a single-site, in-depth case study to refine the conceptual model, to identify construct indicators, and to derive hypotheses. Finally, we conclude by discussing both the contributions of our work, limitations, and future research directions.

2. Background and prior work

The increasing sophistication of information communication and technology (ICT) combined with increasing competitive pressures has led to the widespread use of Global Virtual Teams (GVTs) (Aritz, Walker, and Cardon 2018; Ramasubbu et al. 2011; Stray, Moe, and Noroozi 2019; Zhang and Venkatesh 2013) as ways of organising work. GVTs operate across different geographical locations, a range of time zones, and are generally culturally diverse (Gilson et al. 2015; Nordbäck and Espinosa 2019; Powell, Piccoli, and Ives 2004). They are growing in prevalence as organisations seek to move to 24-hour working across geographically dispersed sites, leverage highly skilled team members, exploit local knowledge, facilitate new ways of work and reduce labour costs (Kroll et al. 2018; Ó Conchúir et al. 2006; Palacio et al. 2011). Consequently, GVTs are widely used in diverse contexts such as manufacturing, research, retailing, and customer support. Indeed, it is argued that the increasing use of GVTs as a distributed work structure is transforming the business world (Gilson et al. 2015; Maznevski et al. 2012; Morrison-Smith and Ruiz 2020).

GVTs are widely used in software development. However, the complexity of the software development process presents particular challenges for GVTs (Ó Conchúir et al. 2009). Software development is an innovation activity that typically involves complex tasks with high interdependencies. These interdependencies require rich communication and coordination (Kraut and Streeter 1995). Therein lies a paradox concerning the use of GVT in organisations, as the very nature of GVTs appears to be at odds with the effective coordination required for software development projects. In particular, GVTs tend to use lean communication channels that prohibit the rich coordination interactions necessary between team members engaged in complex software development work (Jiménez, Piattini, and Vizcaíno 2009; Niinimäki et al. 2012; Noll, Beecham, and Richardson 2010). Despite the increasing sophistication of collaborative ICT platforms in recent years, interaction richness is not always at the optimal level to facilitate tightly-coupled, collaborative software development work (Aritz, Walker, and Cardon 2018; Stray, Moe, and Noroozi 2019). Take for instance projects employing the agile method for development which may find it challenging to replicate the informal communication flexibility and speed of co-located pair programming (Layman et al. 2006; Niazi et al. 2016b). Therefore, it is unsurprising that the coordination of software development work in GVTs is proving problematic for organisations (Medappa and Srivastava 2019; Noll, Beecham, and Richardson 2010).

Current research does little to help enhance our understanding of how to overcome the challenges facing practitioners in using GVTs for software development. Existing studies focus on GVT issues such as time zone differences, geographical distance, language differences, cultural differences, and trust (Espinosa et al. 2003; Dennis et al. 2012; Gilson et al. 2015; Lin, Standing, and Liu 2008; Morrison-Smith and Ruiz 2020; Powell, Piccoli, and Ives 2004; Kanawattanachai and Yoo 2007). Furthermore, these studies indicate that the aforementioned GVT characteristics adversely affect team performance (Noll, Beecham, and Richardson 2010; Palacio et al. 2011). Research also reveals that GVTs suffer from coordination issues such as low levels of individual commitment, control problems, chronic misunderstandings, communication problems, delayed response times, role overload, role ambiguity, absenteeism, and social loafing (Carmel and Agarwal 2001; Jarvenpaa and Leidner 1999; Nguyen-Duc, Cruzes, and Conradi 2012; Shameem, Kumar, and Chandra 2017; Wilson, Crisp, and Mortensen 2013). However, while the impact of individual GVT issues (e.g. cultural diversity) on coordination and team performance has been the subject of study (cf. Kayworth and Leidner 2000; Powell, Piccoli, and Ives 2004), such studies have failed to provide a comprehensive understanding of how GVT issues impact the effectiveness of specific coordination mechanisms (Nguyen-Duc, Cruzes, and Conradi 2015). Thus, researchers argue that existing studies shed little light on the practicalities of implementing effective coordination in GVTs (Espinosa et al. 2007; Taweel et al. 2009; Vizcaíno et al. 2019) as the challenge of coordinating ICT-enabled work in GVTs has not been adequately addressed (cf. Dubé and Robey 2009; Medappa and Srivastava 2019; Watson-Manheim, Chudoba, and Crowston 2012). This is particularly evident from those studies that focus on coordination in GVTs, which, in contrast to studies of traditional co-located software development (e.g. Kraut and Streeter 1995; Andres and Zmud 2002), in general, do not identify the mechanisms by which coordination is achieved in GVTs (Nguyen-Duc et al. 2015; Sharma 2003). It is of critical importance that the specific impact of GVT issues on particular categories of coordination mechanisms is investigated and identified. Coordination mechanisms are the mode by which coordination is achieved and maintained in GVTs. If GVT issue impact can be uncovered at the mechanism level, solutions can be identified and applied at this base level to ensure effective coordination is consistent and achievable for distributed software development projects.

3. Towards a conceptual model for studying the coordination of virtual teams

As a first step in addressing this gap in knowledge, this study adopts an existing classification of coordination mechanisms in order to explore how GVT issues have been reported as impacting those mechanisms in prior research studies. The study employs’ Sabherwal’s (2003) classification of coordination mechanisms as a foundation for our conceptual model. Using a classification for coordination mechanism is widely recognised as a useful lens to observe coordination practices in team projects (Andres and Zmud 2002; Kraut and Streeter 1995; Nidumolu 1995). While Sabherwal’s study does concern itself with coordination mechanisms, its primary focus is on the relationship between client and vendor in outsourced SD projects and how coordination mechanisms evolve in the course of the relationship. The classification of coordination mechanisms described by Sabherwal is a synthesis of existing research on coordination in software development. Four constructs are posited by Sabherwal as key mechanisms to coordinate teamwork in software development teams (cf. Kumar and Van Dissel 1996; Kraut and Streeter 1995; Mintzberg 1979; Thompson 1967):

1. Coordination by standards: coordination by standards refers to those mechanisms which are used to direct team members to uniform practice such as methodologies, codes of practice etc.

2. Coordination by plans: coordination by plans refers to any documentation which may be employed to coordinate and direct team members (schedules, project plans, etc.).

3. Coordination by formal mutual adjustment: coordination by formal mutual adjustment are those mechanisms that require team members to interact in a pre-defined manner, such as project meetings.

4. Coordination by informal mutual adjustment: coordination by informal mutual adjustment involves team members interacting in an informal manner through ad-hoc meetings, impromptu communications, or co-location.

Several factors or issues are identified as influencing project outcomes in global virtual teams (Table 1). This study selected five of these issues to study their impact on coordination mechanisms in a GVT. There is significant support in existing literature to suggest that geographical distance, time zone differences, language differences, cultural differences and trust adversely impact the coordination of a GVT as well as influencing project outcomes (Noll, Beecham, and Richardson 2010; Boyer O’Leary, Wilson, and Metiu 2014; Palacio et al. 2011; Sievi-Korte, Beecham, and Richardson 2019). While issues such as leadership and knowledge sharing influence project outcomes, their relationship to coordination is different, with leadership acting as a facilitator of coordination (Eisenberg and Krishnan 2018; Gilson et al. 2015; Malhotra, Majchrzak, and Rosen 2007; Zakaria, Amelinckx, and Wilemon 2004) and knowledge sharing an outcome of effective coordination (Ghobadi 2015; Gilson et al. 2015; Rosen, Furst, and Blackburn 2007; Zakaria, Amelinckx, and Wilemon 2004). Leadership and knowledge sharing were excluded from the study to focus instead on GVT issues that have been predominantly reported as having a direct, negative impact on coordination (Noll, Beecham, and Richardson 2010; Palacio et al. 2011). Geographical distance hampers the effectiveness of coordination measures that typically rely on the proximity of team members (Vizcaíno et al. 2019). Similarly, time zone differences disrupt the typical coordination processes that ensure team members successfully interact to ensure a successful project outcome (Espinosa and Carmel 2003). GVTs often result in team members using a foreign or second language to communicate. Consequently, miscommunications often ensue, and these impact coordination measures that rely heavily on verbal interactions (Sievi-Korte, Beecham, and Richardson 2019). Cultural diversity amongst team members has also been reported as having an adverse impact on effective coordination, with varying national and organisational cultures causing missteps between team members (Zakaria and Yusof 2020). Trust between GVT members has received significant attention in prior studies (Powell, Piccoli, and Ives 2004) and is viewed as having great sway over the effectiveness of coordination measures employed by the team (Shameem, Kumar, and Chandra 2017).

In order to build theoretical propositions, these five issues, identified from previous research, are presented as constructs:

• Geographical Distance: Physical separation of team members across geographically dispersed locations (Saunders, Van Slyke, and Vogel 2004).

• Time Zone Differences: Time differences in locations of team members (Herbsleb et al. 2000).

• Language Differences: The working language of the team is not the native language for all team members (Dubé and Paré 2001).

• Cultural Differences: Team members exhibit diverse ethnic, national, and organisational backgrounds (Carmel and Agarwal 2001; Kayworth and Leidner 2000; Kotlarsky and Oshri 2005).

• Trust: Willingness to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trusting party, irrespective of the ability to monitor or control that trusted party (Mayer, Davis, and Schoorman 1995)

Though it is widely assumed that these issues can collectively seriously impact the successful outcome of projects via coordination mechanisms, there is a paucity of supporting evidence in the literature. As previously noted, there are very few studies that have explored the issue of coordination at a mechanism level (McBride 2008; Nguyen-Duc et al. 2014). In addition, the existing research on GVTs has its limitations. While there are a number of GVT studies which focus on software development, short-term projects have been predominant (Alaiad, Alnsour, and Alsharo 2019; Ale Ebrahim, Ahmed, and Taha 2010; Gilson et al. 2015; Powell, Piccoli, and Ives 2004; Scott and Wildman 2015). Few studies look at GVT projects that span longer time periods, which appear to be now more numerous (Hacker et al. 2019; Maznevski and Chudoba 2000). In addition, many existing studies into GVTs employ simulated virtual team projects using university students (Gilson et al. 2015; Kayworth and Leidner 2000; Raghuram et al. 2019; Sutanto, Kankanhalli, and Tan 2004). While such studies may contribute to theory building, their findings are not generalisable to industry and may not be relevant to practitioners. There is, therefore, a need to look specifically at GVTs involved in long term SD projects of 12 months or more and to identify the specific impact of GVT issues on different classifications of coordination mechanisms (standards, plans, formal mutual adjustment, and informal mutual adjustment).

To address the gap in our knowledge of GVTs, this research study focuses on the impact of GVT issues on coordination mechanisms in particular. In the next four sections, we use prior research on GVTs and team coordination to conceptualise the Global Virtual Team Issues-Coordination Mechanism (GVTI-CM) model, which maps the impact that GVT issues (identified above) have on the effectiveness of different types of coordination mechanism (see Figure 1). We use the constructs here defined to develop several theoretical propositions to inform theory building and refinement in the case chosen for study (cf. Eisenhardt 1989).

Figure 1. GVTI-CM conceptual model.

The issues of geographical distance, time zone differences, language differences, trust, and cultural differences are argued to negatively impact a virtual team’s ability to successfully interact and complete a software development project. The problems that arise from these issues create obstacles for the coordination of teamwork in particular (Montoya-Weiss, Massey, and Song 2001; Morrison-Smith and Ruiz 2020). There is general agreement that team coordination is a key activity in producing successful outcomes for project teams (Hsu et al. 2012; Kroll et al. 2018; Maznevski and Chudoba 2001; Parolia et al. 2007).

3.1. Standards as coordination mechanisms

Standards are one category of coordination mechanism employed in SD teams (Sabherwal 2003). Standards might include development methodologies, document templates, guideline documents, data dictionaries, error tracking procedures, and lifecycles. Typically, they are established at the start of a project and communicated to the entire software development team and are altered only in exceptional circumstances. However, while standards are generally held static and do not require a high degree of team member interaction, they are not immune from the impact of global virtual team issues. Evidence in existing literature suggests that geographical distance between team members creates problems in enforcing and ensuring adherence to a project’s standards (Cramton 2001; Espinosa et al. 2003; Raghuram et al. 2019; Ramesh and Dennis 2002). Multiple time zones play havoc with the team’s ability to follow error tracking procedures across sites (Niazi et al. 2016a; Powell, Piccoli, and Ives 2004). Existing research also suggests that cultural differences can lead to team members at different geographical locations interpreting standards (templates, methodologies, etc.) differently due to locally held practices and assumptions (Maznevski and Chudoba 2000; Niazi et al. 2016a; Sutanto, Kankanhalli, and Tan 2004; Zakaria et al. 2020). The language typically used as the official working language of the virtual team is not necessarily the first (or in some cases second) language of all team members (Gilson et al. 2015; Morrison-Smith and Ruiz 2020; Mortensen and Hinds 2001; Kayworth and Leidner 2000; Vizcaíno et al. 2019). Differing levels of language proficiency leads some team members to misuse or misinterpret standards such as templates and data dictionaries. This hinders the understanding by team members of project standards (Vizcaíno et al. 2019). Trust is difficult to engender and maintain in a virtual team. Team members can mistrust other team members at other project sites and their ability to adhere to standards set out for the team (Chang, Hung, and Hsieh 2014; Mortensen and Hinds 2001; Kayworth and Leidner 2000; Vizcaíno et al. 2019). Based on the foregoing analysis of existing research, we present our first proposition:

Proposition 1: The effectiveness of standards is impacted by geographical distance, trust, language differences, time zone differences and cultural differences.

3.2. Plans as coordination mechanisms

Plans are identified as another category of coordination mechanism (Kraut and Streeter 1995; Sabherwal 2003). Plans typically include project milestones, requirements specification, sign-offs; test plans, project plans, project estimates, etc. (Kraut and Streeter 1995; Sabherwal 2003). While plans are formulated before the commencement of a project and communicated to the entire virtual team, they are subject to ongoing modification throughout a project. Ideally, they should be held static, but this rarely occurs within a project; thus, they require a high degree of communication among team members.

Existing research suggests that geographical distance between team members during project planning causes delays in decision making and overall project progress (Herbsleb et al. 2000; Mani, Kannan, and Bharadwaj 2014; Morrison-Smith and Ruiz 2020; Mortensen and Hinds 2001). Creating the initial project plan and subsequent modifications are made more difficult and progress at a slower pace due to the physical separation of team members (Espinosa et al. 2003; Niazi et al. 2016a; Piccoli et al. 2004). There is evidence to suggest that different time zones also cause delays in relation to plans (Espinosa et al. 2003; Nguyen-Duc, Cruzes, and Conradi 2015; Vallon et al. 2018). If a requirements specification is modified or newly introduced, there is a typical 24 h time delay in rolling out the new requirements specification to the entire team due to multiple time zones and working hours (Espinosa et al. 2003; Giuffrida and Dittrich 2015). Plans suffer from a lack of a shared language between team members (Fleischmann, Aritz, and Cardon 2020; Majchrzak et al. 2000; Sarker and Sahay 2004), with misinterpretations occurring regularly (Sievi-Korte, Beecham, and Richardson 2019). Team members sharing similar proficiency levels in the chosen language can help remove ambiguity in project plans and requirements specification, which might otherwise be open to misinterpretation (Cramton 2001; Vizcaíno et al. 2019; Zahedi, Shahin, and Babar 2016). Project milestones, estimates, and project plans are also impacted by cultural differences of team members at different geographical locations (Cramton and Hinds 2014; Kayworth and Leidner 2000; Maznevski and Chudoba 2000; Vizcaíno et al. 2019). For example, when creating a project plan, team members need to consider a global schedule that may be affected by national holidays, vacation times, and religious observances (Kroll et al. 2018; Sarker and Sahay 2004). Lack of awareness of cultural differences (social, national and organisational) can seriously impact the team’s ability to effectively plan and schedule the project (Ahmad and Lutters 2011; Zakaria et al. 2020). Levels of trust between team members at different geographical locations will impact the effectiveness of plans as a coordination mechanism (Sarker and Sahay 2004; Vizcaíno et al. 2019). Trust between team members in virtual teams can be difficult to create and maintain (Powell, Piccoli, and Ives 2004; Shameem, Kumar, and Chandra 2017). With the typically high level of modification associated with plans, team member trust may come under pressure. This may arise due to a reliance on team members at other project sites to maintain project plans, produce accurate estimates, and complete project requirements properly (Alsharo, Gregg, and Ramirez 2017; Chang, Hung, and Hsieh 2014; Kayworth and Leidner 2000). If team members fail to complete these coordination tasks correctly, then the levels of trust among team members may be reduced, with a negative impact on the effectiveness of such mechanisms to coordinate the virtual team (Dennis et al. 2012; Dubé and Paré 2001; Jarvenpaa and Leidner 1999; Morrison-Smith and Ruiz 2020).

There is a paucity of research that specifically looks at the impact of GVT issues on plans as a coordination mechanism. While existing research does provide evidence that coordination mechanisms are impacted by a GVT structure, few studies focus on the use of plans to coordinate team effort. Drawing on the foregoing, we now present our second proposition:

Proposition 2: The effectiveness of plans is impacted by geographical distance, trust, language differences, time zone differences and cultural differences.

3.3. Formal mutual adjustment as coordination mechanisms

Existing research identifies formal mutual adjustment as the third type of coordination mechanism employed in project teams (Kraut and Streeter 1995; Sabherwal 2003). Unlike standards and plans, formal mutual adjustment depends on a high degree of team member interaction and communication to be an effective coordination mechanism. Formal mutual adjustment refers to any institutionally sanctioned interaction between team members. Examples of this type of mechanism are project team meetings, code inspections, hierarchies (reporting structures), and review meetings (Sabherwal 2003). Formal mutual adjustment seeks mutual adjustment between team members in a more structured fashion than informal mutual adjustment (see next section).

The geographical distance separating team members affects the effectiveness of formal mutual adjustment due to the high levels of team member interaction and communication required to coordinate team activities (Maznevski and Chudoba 2001; Niazi et al. 2016a; Shameem, Kumar, and Chandra 2017). Team members at different geographical locations cannot readily meet face-to-face for meetings (Espinosa and Carmel 2003; Ramesh and Dennis 2002; Sievi-Korte, Beecham, and Richardson 2019; Vallon et al. 2018), nor can project managers oversee the work of team members working in different locations. While advances in ICT improve virtual interactions through media richness, they do not, as yet, offer the same speed and flexibility as colocation for complex tasks (Jimenez et al. 2017; Scott and Wildman 2015). Given the nature of formal mutual adjustment and the higher degree of team member interaction, cultural differences may hold greater sway, which can lead to difficulties (Johansson, Dittrich, and Juustila 1999; Kayworth and Leidner 2000; Maznevski and Chudoba 2001; Sidhu and Volberda 2011; Xia, Dawande, and Mookerjee 2016). The greater the cultural diversity, the more incompatibilities between team members in terms of relationship values, time perception, and norms may lead to difficulties in coordination (Jensen and Nardi 2014; Kayworth and Leidner 2000; Maznevski and Chudoba 2000; Morrison-Smith and Ruiz 2020; Sutanto, Kankanhalli, and Tan 2004).

Time zones also impact formal mutual adjustment as they force team members to organise and schedule team meetings at specific times (Espinosa and Carmel 2003; Li and Maedche 2012; Niazi et al. 2016a). The greater the distance between team member project locations, the smaller the team's window of time to organise and participate in team meetings (Giuffrida and Dittrich 2015; Kayworth and Leidner 2000; Sarker and Sahay 2004). The burden of facilitating meeting times often falls more on one team site than others, with team members at those sites operating outside of local office hours to accommodate their virtual team members (Jensen and Nardi 2014).

In addition, formal mutual adjustment may also be affected by differing levels of language proficiency among team members, especially in the team's agreed working language. Team members who are not as confident or experienced with the project language will feel constrained in their interactions with native speakers of this language (Mani, Kannan, and Bharadwaj 2014; Niazi et al. 2016a; Sarker and Sahay 2004). Consequently, this can lead to some team members failing to participate fully in meetings, conference calls, or code inspections (Kayworth and Leidner 2000; Vizcaíno et al. 2019).

Team members at different project locations take time to build trust with team members in other project locations (Dennis et al. 2012; Maznevski and Chudoba 2001). In addition, if this trust is lost, it takes longer to rebuild when team members are physically separated. This is important as the effectiveness of formal mutual adjustment mechanisms depends on healthy levels of trust between team members (Breuer, Hüffmeier, and Hertel 2016; Kayworth and Leidner 2000). For example, if a team member does not trust other team members participating in conference calls, they are less likely to share bad news or participate in project discussions (Jarvenpaa and Leidner 1999; Vizcaíno et al. 2019; Wang, Wang, and Redmiles 2019). Hence, research provides evidence that geographical distance, time zone differences, cultural differences, language differences and trust may have a negative impact on the effectiveness of formal mutual adjustment as a coordination mechanism in virtual teams. Given the foregoing observations, we posit a third proposition:

Proposition 3: The effectiveness of formal mutual adjustment is impacted by geographical distance, trust, language differences, time zone differences and cultural differences

3.4. Informal mutual adjustment as coordination mechanisms

Informal mutual adjustment is another type of coordination mechanism (Kraut and Streeter 1995; Sabherwal 2003). Informal mutual adjustment, like formal mutual adjustment, depends on appropriate team member interaction and communication to work effectively. Informal mutual adjustment refers to any informal interaction between team members. Examples of this mechanism are impromptu communications, informal meetings, or joint development (Sabherwal 2003). Informal mutual adjustment coordinates team members’ activities through ad-hoc interactions. For example, a team member might email another team member unexpectedly to share information or to seek advice, or make a request for information.

Traditionally, informal mutual adjustment relies on team members working in the same physical space. Co-location allows team members to adjust by face-to-face interaction mutually. However, there is general agreement in previous studies that co-location as a coordination mechanism cannot be used, in any feasible sense, in a virtual team (Espinosa and Carmel 2003; Gilson et al. 2015; Malhotra and Majchrzak 2012; Powell, Piccoli, and Ives 2004). This influences the effectiveness of informal mutual adjustment as a coordination mechanism. Instead, team members rely on electronic communication, such as emails, texts, instant messages, online chat, and phone calls, to achieve a similar effect (Stray, Moe, and Noroozi 2019). However, studies suggest that these impromptu communications do not have the power of face-to-face interaction (Maznevski and Chudoba 2001; Sievi-Korte, Beecham, and Richardson 2019; Zahedi, Shahin, and Babar 2016).

Similarly, when team members operate in different time zones, the effectiveness of informal mutual adjustment is negatively affected. Since team members work in different global locations, and therefore have different office hours, an impromptu phone call is not always possible (Fleischmann, Aritz, and Cardon 2020; Espinosa and Carmel 2003; Niazi et al. 2016a; Sievi-Korte, Beecham, and Richardson 2019), while emails will have a delayed response. In addition, informal meetings may not be possible when time zones do not overlap due to significant geographical distance between team sites (Giuffrida and Dittrich 2015; Morrison-Smith and Ruiz 2020; Nguyen-Duc, Cruzes, and Conradi 2015; Sutanto, Kankanhalli, and Tan 2004). Thus, every communication with different project locations has to be planned.

There is evidence in existing literature to suggest that the use of different languages by project team members reduces the effectiveness of coordination mechanisms such as informal mutual adjustment (Dubé and Paré 2001; Boyer O’Leary, Wilson, and Metiu 2014; Vizcaíno et al. 2019). Team members with a poor grasp of a project’s working language may avoid phone calls. They will, instead, opt for emails, instant messages and/or online chat which afford them the opportunity to check language, but operate at a slower pace. This is problematic with urgent project decisions or if a project query is a straightforward one (Sarker and Sahay 2004; Zahedi, Shahin, and Babar 2016).

There is a general agreement that cultural differences lead to coordination difficulties (Giuffrida and Dittrich 2015; Kayworth and Leidner 2000, Maznevski and Chudoba 2001; Vizcaíno et al. 2019). Hence, cultural differences create obstacles to effective communication. Research finds that team members in different geographical locations display a preference for emails or other forms of formal correspondence over more informal phone calls, mobile communications, or instant messaging (Herbsleb and Moitra 2001; Jimenez et al. 2017; Raghuram et al. 2019; Shameem, Kumar, and Chandra 2017; Zakaria, Amelinckx, and Wilemon 2004).

In a similar manner to formal mutual adjustment, levels of trust between team members can influence the effectiveness of informal mutual adjustment as a coordination mechanism (Breuer, Hüffmeier, and Hertel 2016; Jarvenpaa and Leidner 1999; Morrison-Smith and Ruiz 2020; Sarker and Sahay 2004). Team members can use casual communications to progress project work. However, due to the informal nature of such communication, team members must trust that decisions made, or any information exchanged, will be communicated to the team as a whole. Trust between team members takes longer to build within a virtual team environment when team members have not met face-to-face (Maznevski and Chudoba 2001; Niazi et al. 2016a). It is with these issues in mind that we present our fourth proposition:

Proposition 4: The effectiveness of informal mutual adjustment is impacted by geographical distance, trust, language differences, time zone differences and cultural differences

The independent and dependant constructs described above are presented in the GVTI-CM conceptual model presented in Figure 1: the relationships between these constructs posited in the above propositions.

4. Research method

The objective of this research study is to investigate the use of traditional coordination mechanisms in virtual SD projects and to explore the impact that the virtual team issues of geographical distance, trust, language differences, cultural differences, and time zone differences have on their effectiveness. As an initial step in theory building, a conceptual model (based on existing research) with associated propositions was presented in the previous section. Following the specification of the propositions, the next step in theory building is to determine empirical indicators and subsequently produce hypotheses for empirical testing (Wheeler 2002). We now describe the research design on which this study is based (Figure 2).

Figure 2. Research design.

A ‘single’ case study research approach was selected as the most appropriate vehicle for theory building and refinement (Carroll and Swatman 2000; Eisenhardt 1989; Yin 1994). Case studies are particularly well suited to IS research (Benbasat, Goldstein, and Mead 1987). In addition, case research, which emphasises understanding empirical data in natural settings, is a suitable method for studying IS issues and practices (Eisenhardt 1989). A single case study is viewed as being a potentially rich and valid source of data and is a particularly good fit for the purpose of exploring relationships between variables in context (Yin 1994).

The case organisation¹ is a Fortune 100 telecommunications company with an estimated 66,000 staff worldwide. The organisation was chosen for the study on the basis of its predominant use of GVTs for software development. One specific project was selected for study from several ongoing software development projects. The specific software development project was purposefully selected, as it met several important criteria: (1) it possessed a global virtual team that operated over several geographically dispersed project locations with multiple time zones; (2) the team members were linguistically and culturally diverse (Figure 2). The project focused on the software development of a network operations and maintenance interface.

The project had a planned 18-month duration with team members located in six geographical locations (Ireland, USA, India, Israel, Malaysia, and China). There were upwards of 48 team members working on the software in different locations at any point in time. The development team was split into a number of distinct development functions; requirements engineering, systems engineering, software test, development, customer support, deployment, quality control, project management etc. Team sub-units in different geographical locations had distinct responsibilities viz. Israel (development), USA (development), Malaysia (development/test), China (development/test), India (development) and Ireland (project management/development). Data collection was conducted over a twelve-month period and concluded towards the end of the project development cycle during the software testing and implementation phases. Interviews, observation and documentation review were employed for data collection. Site visits to the Irish team location facilitated researcher observation of project team interactions. Review of a range of project documentation (organisational charts, project planning, and requirements documents, etc.) facilitated a clear and in-depth picture of team operations. Interviewees were purposively selected (Onwuegbuzie and Leech 2007) using the key informant approach. Thus, participants (see Table 2) were selected from all project sites (excluding Malaysia, for which no participant was available during data collection) with those participating in the study performing a range of project roles within the team (project manager, development manager, software developer, system engineer, test team leader, software tester).

Table 2. Interview participants.

INTERVIEWEE	ROLE	LOCATION
DM1	Development Manager	Ireland
DV1	Developer	Ireland
PM1	Project Manager	Ireland
PM2	Project Manager	Ireland
PM3	Project Manager	Israel
SD1	Senior Developer	Ireland
SD2	Senior Developer	USA
SE1	Software Engineer	India
TA1	Technical Architect	Ireland
TL1	Test Team Lead	China

Each participant had a minimum of ten years’ experience working in global virtual teams. While ten participants is on the low side, each expert participant was purposefully selected to ensure the capturing of data from a diverse set of team members in terms of location, native language, time zone and culture. Due to physical proximity to team members based in Ireland, researchers had greater access to participants in that geographical area. As Ireland was also the centre for project management of the overall project, this proximity allowed the researchers to capture expert opinion from team members with key coordination knowledge and experience. As such, the team was sufficiently represented; furthermore, the analysis of interview transcripts indicated that data saturation was achieved (Seidman 2006).

The interviews were unstructured in format. However, an interview guide (Patton 1990) was used to ensure consistency and coverage of topic across all interviews with the ten participants (Figure 2). The interview guide was devised based on the conceptual model developed from prior research during the initial theory-building phase. The guide required the capturing of: (i) project role; (ii) team structure; (iii) coordination mechanism use; (iv) evidence of GVT issues; and (v) how GVT issues impacted effectiveness of each category of coordination mechanism. Follow-up interviews, emails, and phone calls with participants were used to clarify and refine issues that emerged from primary interview transcripts. Content analysis was conducted by a three-person research team and employed techniques proposed by Strauss and Corbin (1990). The first step (open coding) involved the data being examined ‘line by line’ to ascertain the main ideas. One researcher undertook this step. These ideas were then grouped by meaningful headings (informed by constructs developed in section 2) to reveal categories and sub-categories/properties. The next step (axial coding) was the process of determining hypotheses about the relationships between a category and its subcategories: e.g. conditions, context, action/interaction strategies, and consequences. This step was undertaken by all three researchers. The focus then turned to the data to assess the validity of these hypothesised relationships. Relational and variational sampling (cf. Strauss and Corbin 1990) was used to select data for this analysis. This process continued iteratively, and resulted in the modification of categories and relationships. Finally, Selective Coding was undertaken to identify the relationships between categories (constructs) using hypothesised conditions, context, strategies and consequences. Discriminate sampling (cf. Strauss and Corbin 1990) was used to select data to examine strong and weak connections between categories.

This process also facilitated the use of a rating scale (Miles and Huberman 1994) with respect to the impact (e.g. high, moderate, low) of specific GVT issues on coordination mechanisms classifications. Analysis of interview transcripts in NVivo allowed the researcher team to determine the volume of negative comments relating to specific GVT issues and coordination mechanisms. For issues that garnered high levels of negative comments in interview transcripts, constructs in hypotheses were considered as exerting a ‘highly negative impact’. All three researchers engaged in an iterative process, assessing the model and derived hypotheses veracity until consensus was achieved.

The issues of validity, reliability, and objectivity (cf. Yin 1994) were addressed through: (a) Prolonged engagement and persistent observation (data was collected over the course of 12 months, initially via interviews and then via site visits, conference calls and emails); (b) triangulation techniques, which were extensively used to provide insights into events, relationships, etc. between data sets (project documentation including organisation charts, planning documents, Gantt chart, requirements documents were accessed to give context and clarity to interview transcripts); (c) a data analysis approach based on rigorous coding and the use of memos, which together provided an audit trail; and, finally, (d) member checks were also employed as the findings were presented to project team members for subsequent feedback (cf. Patton 1990).

5. Case study analysis²

This section presents the findings of the case study. From this analysis we were able to (i) identify empirical indicators for virtual team issues (Table 3); (ii) identify empirical indicators for coordination mechanisms (Table 4) and (iii) illustrate how the effectiveness of coordination mechanisms is impacted by virtual team issues (geographical distance, time zone differences, language differences, cultural differences, and trust).

Table 3. Indicators for GVT issues (language differences, trust, geographical distance, time zone differences and cultural differences).

CONSTRUCT	EXPLANATION	INDICATOR
Language Differences	The working language of the team is not the native language for all team members	Team members think that others: have a poor command of the working language write the working language poorly comprehend the project working language poorly Team members do not understand others accents
Trust	Willingness to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trusting party, irrespective of the ability to monitor or control that other party	Team members assume that other team members: do not lie do not withhold information will do their work to the best of their abilities iv. will complete their work even if they are not physically present to monitor work
Geographical Distance	Physical separation of team members across geographically dispersed locations	Team members cannot: meet face-to-face in the same room regularly or not at all cannot monitor other team members work Team members do not have direct access to all team resources Team members operate in different countries
Time Zone Differences	Time differences in locations of team members	Team members: cannot work a full day together access to other team members and project sites is limited meetings will occur at irregular hours to facilitate different team locations experience a 24/48 h delay can occur for queries across project locations in some project locations must work irregular hours to facilitate other project locations operate across multiple time zones All team members cannot attend one joint project meeting
Cultural Differences	Team members exhibit diverse ethnic, national, and organisational backgrounds	Team members: perceive others as having different work and social etiquette to their own have different vacation time, religious holidays, public holidays across different project locations use different phrasing (slang) and employ language differently across project locations perceive others as having preference for specific electronic communication

Table 4. Indicators for coordination mechanisms (Standards, plans, formal mutual adjustment and informal mutual adjustment).

CONSTRUCT	EXPLANATION	INDICATOR
Standards	Those mechanisms which are used to direct team members to uniform practice such as methodologies, codes of practice	Team members conduct different project phases in parallel with other project phases Team members use designated templates to create documentation at each phase of the development project Team members pass project documentation from one sub-team to another (requirements engineering, software engineering, software test etc.) Team members review documentation to ensure consistency across the team Team members use set guidelines to author an email Team members follow the organisation defined development process Team members use designated synchronisation points to ensure the defined development process is being adhered to
Plans	Documentation which may be employed to coordinate and direct team members (schedules, project plans etc.)	The team authors and maintains a project plan for the project The team authors and maintains a software configuration management (SCM) plan for the project The team authors and maintains a quality control plan for the project The team calculates and publishes project estimates for the project The team authors and maintains a requirements document The team authors and maintains a Gantt Chart for the project
Formal Mutual Adjustment	Those mechanisms which require team members to interact in a pre-defined manner such as project meetings	Team members participate in a weekly scheduled project conference call Team members participate in scheduled, weekly status calls Team members at local sites attend weekly, scheduled site meetings Team members participate in weekly, scheduled meetings for their functional area (development, test, project management etc.) Team members maintain and share calendars Team members author, send and/or receive emails via mail aliases
Informal Mutual Adjustment	Those mechanisms which allow team members to interact in an informal manner through ad-hoc meetings, impromptu communications or co-location	Team members make and receive ad-hoc phone calls from other team members Team members author, send and/or receive instant messages to other team members Team members author, send and/or receive ad-hoc emails from other team members Team members make and receive ad-hoc mobile phone calls from other team members Team members at local project sites hold face-to-face conversations with other local team members

5.1. Coordination mechanisms – Standards, plans, formal mutual adjustment and informal mutual adjustment

All coordination mechanisms identified in the case organisation were mapped to categories highlighted by existing research; hence, we find support for our GVTI-CM conceptual model (see Table 4). The standards employed as coordination mechanisms in this project included: (1) an explicitly defined software development lifecycle; (2) an extensive repository of document templates; (3) a software development methodology with distinct synchronisation points for sign-offs; (4) guideline documents; (5) code inspections; and (6) test case creation. The presence of plans was indicated by the global virtual team’s use of: (1) a project plan that included a software configuration plan (SCM), a quality control plan for the project, project estimates, and a project Gantt chart; and (2) a requirements document with associated project dates and attached resources. Formal mutual adjustment was achieved in this team through institutionalised structures such as: (1) weekly, pre-scheduled conference calls; (2) weekly pre-scheduled status calls; (3) local weekly site meetings; (4) weekly development, test, software engineering, project management meetings; and (5) use of mail aliases and shared calendars. The coordination mechanisms that focused on informal mutual adjustment included: (1) regular ad-hoc phone calls between team members; (2) ad-hoc instant messages between team members; unscheduled emails between team members; (3) unscheduled mobile phone calls between team members; and (4) ad-hoc conversations between co-located team members.

5.2. The impact of virtual team issues on the effectiveness of standards

The GVTI-CM conceptual model (Figure 1) proposes that the effectiveness of standards is impacted by global virtual team issues, including geographical distance, time zone differences, language differences, cultural differences and trust. The model assumes that these virtual team issues impact on coordination mechanisms in equal measure. However, the findings indicate that some virtual team issues have a greater impact than others on specific coordination mechanisms. The analysis of case study data also revealed that standards (Table 4) are impacted by language differences, geographical distance, and team member trust. Also, the data indicate that time zones and cultural differences have no perceived impact on standards. Language differences had the greatest negative impact on project documentation, such as bug reports and templates. A project manager (PM1) explained that in a number of instances, documents had to be reviewed and that they had to work out what the non-English speaking team member meant before rewriting the document for general consumption (Table 3).

Team members also reported the creation of additional documentation, which acted as guidelines for team members across different project sites. The test team leader (TL1) created a set of documents for his team that detailed how to write an email, ask a question, and write a test case. He described this as ‘documents to show how to use documents.’ Concerning geographical distance, the impact is observed in different project sites that diverge in development processes and which apply differing methodologies. The development manager (DM1) was particularly concerned about this divergence. He explained, ‘when you have different teams in different geographic regions you may have different kinds of quality standards, and therefore there may be some non-uniformity in terms of the quality.’ In relation to project documentation and standards, team members expressed high levels of distrust in connection with team members at other project sites (Table 3). When documentation was not updated in a timely fashion, it decreased the confidence the team members had in the documentation and thus reduced its usefulness as a coordination mechanism.

The GVTI-CM conceptual model proposes that the effectiveness of standards is impacted by geographical distance, time zone differences, trust, cultural differences, and language differences. However, this is not fully supported by the findings of the study. The results (based on a number of negative statements in interview transcripts) indicated that language differences have a highly negative impact on the effectiveness of standards as a coordination mechanism. Trust was also found to have a high negative impact. Geographical distance was also found to have a moderately negative impact on the effectiveness of standards. However, case study analysis found no support for cultural and time zone differences as impacting on the effectiveness of standards (Table 4). We can therefore continue the theory building process by specifying three hypotheses:

H1: Greater geographical distance has a moderately negative impact on the effectiveness of standards

H2: Lower trust has a highly negative impact on the effectiveness of standards

H3: Greater language differences have a highly negative impact on the effectiveness of standards

5.3. The impact of virtual team issues on the effectiveness of plans

The case study data revealed that plans (Table 4) are greatly impacted by geographical distance, time zone differences, cultural differences, and team member trust. Case study analysis revealed that language differences appeared to have no impact on plans with no statements from interview transcripts relating to linguistic differences. The impact of geographical distance between team members on plans was evident in the requirement of team members at different project locations to maintain local project schedules and then feed into one global project schedule for the entire team. Also, project estimates were required to be submitted by each project location. Team members recognised the need for plans within the project team, but they also felt that there were ‘almost too many’ mechanisms employed. This meant that the project team spent a lot of time ‘going around in circles.’ Cultural differences were perceived as having a highly negative impact on the effectiveness of plans as a coordination mechanism (based on the high number of negative comments in interview transcripts), with some team members observing that other team members tended to avoid saying ‘no’ to work requests from other project sites which had serious ramifications for the project schedule and completion. One project manager (PM2) provided an example of team members in another location agreeing to a project task even though they knew from the outset it was not feasible with the given schedule. PM2 perceived this as being a ‘cultural issue’ as team members in the other project sites would ‘generally say no to work which was not feasible straightaway.’ This reluctance to communicate a ‘negative’ response to team members in distributed locations appears regularly in prior research on cultural differences in GVTs (Tan et al. 2003).

The case study data indicated that multiple time zones were perceived as having a highly negative impact on the effectiveness of plans as coordination mechanisms. PM1 explained the problem of receiving estimates or addressing project plan issues that needed to be answered quickly, but the relevant team members were not available because their site was offline (Table 3). The technical architect (TA1) agreed with this view, explaining that the team may be under pressure to get something done quickly, which depends on team members in different geographical locations. If that site has a holiday period, team members may not be available to take action. The impact of trust on plans was particularly evident. DM1 commented that it was difficult to trust any team members due to a number of project issues, which another project site had been slow to bring to light. PM2 at the same project site agreed with this view stating that ‘Our feature manager may say yes, our code is done and that the schedule is showing a date. I’d often go in and check the schedule myself and find that the schedule is not showing the date.’ Case analysis also revealed that in the initial stages of creating working relationships, team members would require other dispersed team members to ‘prove themselves.’ During this period, team members would also heavily monitor other team members at different project sites. The less physical contact a team member had with another team member, the more rigorous they would be in monitoring activities (Table 3). This supports the findings of prior research that reports on the initial distrust between new team members (Jarvenpaa, Shaw, and Staples 2004). We, therefore, continue theory building by specifying four hypotheses:

H4: Greater geographical distance has a moderately negative impact on the effectiveness of plans

H5: Lower trust has a highly negative impact on the effectiveness of plans

H6: Greater time zone differences have a highly negative impact on the effectiveness of plans

H7: Greater cultural differences have a highly negative impact on the effectiveness of plans

5.4. The impact of virtual team issues on formal mutual adjustment

The analysis of the case data confirmed that the effectiveness of formal mutual adjustment (FMA) (Table 4) was impacted by language differences, time zone differences, trust, cultural differences, and geographical distance. This finding matches the GVTI-CM conceptual model (Figure 1). However, the level of impact differs across the virtual team issues. While cultural differences and trust impacted the effectiveness of FMA mechanisms, geographical distance, time zone differences, and language differences had a greater perceived impact (based on the high number of negative comments identified in interview transcripts). Geographical distance negatively impacted on the effectiveness of FMA by the physical separation of team members across different team sites (Table 3). Co-location was not an option for FMA. Team members viewed this as a disadvantage. DM1 observed one project conference call where they noticed that a participant had ‘gone quiet.’ Because they were using a conference call, team members could not tell if the participant had dropped off or was annoyed. Since the development manager could not see the participant, it took them longer to realise the team member’s dissatisfaction with aspects of the project (Table 3). Another issue that arose concerning geographical distance and FMA was the high usage of conference calls. A software engineer (SE1) stated that large-scale conference call meetings could not focus on specific technical issues and missed these project problems ‘unless they are glaringly obvious.’ To resolve this issue, she often ended up having to call team members one-to-one in order to catch and resolve technical items.

Time zone differences were revealed as having a highly negative impact on the effectiveness of FMA as coordination mechanisms. All participants made a number of negative comments relating to time zones and formal interactions between team members. DM1 provided the example of team members in two locations taking project conference calls at 10:30 pm (local time) or hosting project calls to facilitate team members based in other project sites, as the useful communication window is small between these distinct sites. SE1 concurred with this view stating that scheduling interaction between team sites was difficult and that ‘Usually one team or the other is having to pay for it in terms of having very, very early in the morning meetings or very later night-time meetings.’ In relation to these long working hours, team members were concerned about performing their work at optimum levels. Team members at some project sites were concerned about the effect that conference calls scheduled late or exceptionally early had on their ability to contribute in a useful way (Table 3).

Language differences were found to have a highly negative impact on the effectiveness of FMA. The Indian software engineer (SE1) concurred with this view stating that ‘Even though everyone speaks English in meetings, phone calls … words and context are often confused and it is harder to get the message across’. A project manager (PM3) observed that team members from different geographical locations and cultures, where English is not the first language, can be quiet in conference calls. A senior developer (SD1) concurred, stating that the difficulty was that ‘some people have a better grasp of English than others. You often find people who can write reasonably legible emails from an English point of view, they might not speak the same. Just with accent and pronunciation it’s actually very, very difficult to understand what they are actually saying’. Cultural differences were also observed to have an impact on the effectiveness of FMA (Table 3). For example, TL1 stated that other project sites reported that team members at his site ‘don’t say anything. They just listen. They do not provide any active feedback!’. SE1 summed up this cultural use of language best by explaining, ‘I think that cultural differences have implications on how people communicate. There are also different characteristics of individuals in different regions as well’. Trust was also found to impact the effectiveness of FMA. Team members observed that it took a longer period of time to build trust with team members in other project locations. Team members tended to lose trust in dispersed team members who did not respond to emails, instant messages, or phone calls in a timely fashion and, once lost, found it was more difficult to rebuild the trust when they could not meet face-to-face. We, therefore, specify five more hypotheses:

H8: Greater geographical distance has a highly negative impact on the effectiveness of formal mutual adjustment

H9: Lower trust has a moderately negative impact on the effectiveness of formal mutual adjustment

H10: Greater language differences have a highly negative impact on the effectiveness of formal mutual adjustment

H11: Greater time zone differences have a highly negative impact on the effectiveness of formal mutual adjustment

H12: Greater cultural differences have a moderately negative impact on the effectiveness of formal mutual adjustment

5.5. The impact of virtual team issues on the effectiveness of informal mutual adjustment

Data from the case supported the proposition that informal mutual adjustment (IMA) (Table 4) is significantly impacted by issues of geographical distance, cultural differences, time zone differences, language differences, and trust. The analysis of the case study data also indicates that geographical distance and time zone differences have the greatest impact on IMA with other virtual team issues having a relatively lesser impact. The effectiveness of IMA is significantly impacted by geographical distance between team members (Table 3). Team members cannot have ad-hoc face-to-face meetings. Instead, face-to-face ad-hoc interaction is replaced with ad-hoc emails, instant messages, and phone calls. In addition, team members are obliged to record ad-hoc interactions that impact the project or change aspects of the project. Time zones were found to impact the effectiveness of IMA as they curtailed the availability of team members based in different project locations. In the same way, emails and instant messages had a delayed response due to multiple time zones. DM1 on the project explained that ‘You could find that an issue is raised in the test team who are located in two separate locations. They send out the issue, it’s their close of business. Development team members, who are located in three separate sites should reply. They send a reply saying I don’t fully understand this and need clarification so you are back into another 24 h’.

Language differences were also revealed as having a moderately negative impact on the effectiveness of IMA. As was the case with formal mutual adjustment, team members who were not native English-speakers tended to avoid phone calls (Table 3). They instead opted for emails and instant messaging, which allowed them time to correct their language, but were also ‘slower’ mechanisms for coordination due to different locations and time zones. The same held true for English-speaking team members who could not understand other team members’ accents on phone calls and instead requested communication via emails. Trust was also revealed as having an impact on the effectiveness of IMA. A software developer (DV1) explained that trust was difficult to build because she could not walk up to them in the office and quickly resolve an issue. She was dependent on the team member answering their phone, replying to her emails, and using Instant Messenger, which was entirely at their discretion. ‘In a virtual team everyone is isolated – we are all outsiders to other sites so trust is there but it is harder to build’. In addition, it was perceived that team members took a longer period of time to build trust with team members in other project locations. Cultural differences were observed as having a low impact on IMA (based on the few related comments from interview participants). One senior developer (SD2) felt that team members in some project locations acted deferentially in their interactions with him: ‘And you think there is no need to be apologetic. If you made a mistake, you make a mistake, we all make mistakes. And there is no need to beg my leave to do something or whatever’. PM3 provided the example of receiving multiple ad-hoc emails over a project action from an off-site member. They would not begin work until every question had been answered, even down to something as small as a typo (Table 3). We therefore conclude our theory building process by presenting five more hypotheses:

H13: Greater geographical distance has a highly negative impact on the effectiveness of informal mutual adjustment

H14: Lower trust has a moderately negative impact on the effectiveness of informal mutual adjustment

H15: Greater language differences have a moderately negative impact on the effectiveness of informal mutual adjustment

H16: Greater time zone differences have a highly negative impact on the effectiveness of informal mutual adjustment

H17: Greater cultural differences have a low negative impact on the effectiveness of informal mutual adjustment

In conclusion, the case study data allowed us to (i) refine our theoretical propositions, (ii) define empirical measures for our constructs (Table 3 and Table 4), and (iii) re-present this paper’s theoretical model with accompanying hypotheses (see Figure 3). Figure 3 presents each coordination mechanism classification associated hypotheses separately.

Figure 3. Revised GVTI-CM conceptual model.

6. Conclusion

This paper has analysed the impact that the GVT issues of geographical distance, language differences, time zone differences, cultural differences, and trust have on the effectiveness of project team coordination mechanisms. The resultant GVTI-CM conceptual model offers a novel lens through which future research may explore the efficacy of coordination mechanisms use in virtual teams. In doing so, this paper contributes to the cumulative body of research on the coordination mechanisms employed in GVTs. The study focused on a software development project in a multi-national telecommunications manufacturer that used a GVT to develop a software sub-system for its infrastructure product. Thus, the unit of analysis in this study provides unique insights into the complex nature of project management in a global context. This has implications for both research and practice.

6.1. Contributions to research

The paper first presented a-priori theory based on prior research on project coordination mechanisms and research on the constructs of geographical distance, time zone differences, language differences, cultural differences, and trust. This theory building activity resulted in (i) a bounded theory and (ii) the definition of constructs and their relationships in the form of propositions. We then employed the case study method to validate the propositions, refine the theory, and to derive empirical indicators for the constructs, and to specify hypotheses on the relationships between those constructs. This process resulted in a conceptual model of high empirical fidelity. The refined theoretical model and its related hypotheses indicate the following:

1. the effectiveness of standards is negatively impacted by greater language differences, greater geographical distance and lower trust;

2. the effectiveness of plans is negatively impacted by greater geographical distance, greater time zone differences, greater cultural differences and lower trust;

3. the effectiveness of formal mutual adjustment is negatively impacted by greater geographical distance, greater time zone differences, greater language differences, greater cultural differences and lower trust; and

4. the effectiveness of informal mutual adjustment is negatively impacted by greater geographical distance, greater time zone differences, greater language differences, greater cultural differences and lower trust.

In addition, we have shown that specific coordination mechanisms are more susceptible to the impact of global virtual team issues than traditional influences. The revised GVTI-CM conceptual model presented in this paper can be employed in future research as a lens through which to test the impact of global virtual team issues on coordination mechanisms.

6.2. Contributions to practice

Globalisation and the enabling role of ICT are pushing the operational boundaries of global virtual teams (GVTs) to limits not previously experienced in software development. Locating project team members to geographical locations across space and time are now viable for organisations looking to have 24 × 7 development of software artefacts, while lowering project costs, and maximising knowledge transfers between experienced software engineers globally. While the ‘uptake’ in the use of GVTs has increased, our understanding of the underlying team structures and processes is still limited, as is our knowledge of project management, coordination, and control, of such initiatives. Thus, this study’s findings have a number of implications for practice. First, organisations looking to use virtual teams for global software development projects need to be vigilant regarding the impact of the issues described herein on the effectiveness of coordination mechanisms. Second, organisations should review the coordination mechanisms employed in their global virtual teams to ensure fit for purpose. For example, there is evidence that standards used to coordinate software projects are seriously impacted by the number of team members whose first language is not that chosen to be the working language (written and oral) of the project. Hence, practitioners need to consider how standards should be modified to minimise the challenges that language differences pose. Alternatively, they may need to consider instituting new standards for GVT environments to accommodate language differences. In a general context, if traditional coordination mechanisms (or those presently in use) are heavily impacted by issues such as those described, then what implications does this have for controlling project outcomes?

Finally, we call for future research that approaches GVTs from a mechanism perspective. To date, there has been very limited research at this level of analysis. Coordination mechanisms are a critical operational component for successful project outcomes. Therefore, coordination mechanisms efficacy cannot be taken for granted in a GVT and merits further investigation. Findings from this study indicate that some coordination mechanisms are subject to significant negative impacts from GVT issues. This points to the need to explore tailoring existing mechanisms, or developing novel mechanisms for a GVT environment.

The advances in ICT and collaborative platforms and how coordination mechanisms are aided (or hampered) by such tools should also be explored. Criticism of such tools in the past was that they lacked the richness necessary to support the complex interactions required between team members for software development. Recent increased use of platforms such as MS Teams, Slack, and GitHub should: (i) illuminate coordination mechanism limitations; (ii) engender experimentation; and (iii) identify fitting solutions.

The impact of team member isolation on coordination mechanism effectiveness also warrants investigation. To date, the majority of GVT-related studies have focused on teams where team members are geographically dispersed but still have some degree of colocation with team members based at their site. What does effective coordination look like when all team members are based in different physical locations?

The model presented in this paper represents a subset of a much larger GVT model, as it assesses the effect of GVT issues on coordination mechanisms solely. Further research is required to explore, test, and develop a comprehensive theory on the impact of GVT issues on other aspects of virtual teamwork, such as control and communication. Also, this study has looked at only five GVT issues (geographical distance, language differences, time zone differences, cultural differences and trust). A future study will explore the potential impacts of knowledge sharing, leadership, multi-teaming, and team member isolation. The set of constructs, measures, and hypotheses presented can be used by researchers as a starting point from which to begin their research on this phenomenon.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The name of the case organisation is removed pending corporate approval for disclosure.

2 To avoid perceptions of generalising to culture, references to exact project locations have been removed.