Unraveling the effects of cultural diversity in teams: A retrospective of research on multicultural work groups and an agenda for future research.

Our 2010 Journal of International Business Studies article, "Unraveling the Effects of Cultural Diversity in Teams: A Meta-analysis of Research on Multicultural Work Groups," attempted to take stock of existing research on cultural diversity in teams, to reconcile conflicting perspectives and past results, and provide a better understanding of the mechanisms and boundary conditions under which diversity affects team outcomes. To guide our analysis, we developed a theoretical framework outlining how cultural diversity leads to both process gains and losses in teams, and specifying the contextual conditions under which diversity contributes to effective team outcomes. We tested our hypotheses in a meta-analysis of research on cultural diversity in teams, encompassing 108 primary studies with a combined sample size of 10,632 work groups. The results suggested that cultural diversity does not have a direct impact on team performance but rather that the effect is indirect, mediated by process variables such as creativity, cohesion, and conflict; and is moderated by contextual influences such as team tenure, the complexity of the task, and whether the team is co-located or geographically dispersed. Unexpected findings raised important questions about the dynamics of diverse teams and underscored the need for further examination. In this Retrospective, we reflect on progress made in research on culturally diverse teams over the last decade, highlight remaining gaps and open questions, and propose an agenda for future research. © Academy of International Business 2021.

INTRODUCTION

In our original article, “Unraveling the effects of cultural diversity in teams: a meta-analysis of research on multicultural work groups,” (Stahl, Maznevski, Voigt, & Jonsen, 2010), we argued that our study added to the existing literature on diverse teams in two important ways. First, we attempted to take stock of and synthesize the findings from previous research on cultural diversity in teams, to reconcile conflicting perspectives and past results, and to propose an agenda for the next stage of research in this field. Second, and more ambitiously, another goal was to improve our understanding of the mechanisms and boundary conditions under which cultural diversity affects team outcomes both theoretically and empirically. Thus, the aim of our paper was not just to summarize what had been found in previous research on diverse teams and provide a descriptive account of empirical regularities (and irregularities) but to go a step beyond existing reviews of the literature and develop a theoretical understanding of mediating mechanisms and contextual contingencies and then test those ideas using meta-analytic techniques.

Our point of departure in the JIBS Decade Award-winning article was an idea that we had developed over more than a decade of dialogue, based on research and review of the literature on multicultural teams, as well as our experiences as management educators, consultants, and coaches working with diverse teams – namely, the now widely accepted notion that cultural diversity presents a ‘double-edged sword’ or mixed blessing for work groups. Diversity can be a source of friction and conflict – and, hence, an obstacle to effective team functioning. But it can also be a source of synergy and learning – a powerful seed for something new, depending on factors that were poorly understood and under-explored in team diversity research at the time we embarked on this study. As we set out to explore some of the yet-unidentified factors that may moderate the relationship between diversity and team performance, we realized how little was actually known about the mechanisms and contingencies underlying the effects of cultural diversity in teams. Of course, there was a rich literature on diversity in work teams (for early meta-analyses of this work, see Bowers, Pharmer, & Salas, 2000; Horwitz & Horwitz, 2007; Stewart, 2006; Webber & Donahue, 2001), including plenty of studies that had looked at the link between cultural diversity and team outcomes (for our meta-analysis, we identified 108 studies that focused on cultural diversity and its consequences for team outcomes), but the evidence was inconsistent and confusing. While some studies found significant positive correlations between diversity and team performance (e.g., Earley & Mosakowski, 2000; Thomas, Ravlin, & Wallace, 1996), others produced significant negative correlations (e.g., Jehn & Mannix, 2001; Kirkman, Tesluk, & Rosen, 2004). There were no meta-analyses available that focused specifically on cultural diversity and its effect on team performance, but existing meta-analyses of research on work group member diversity/heterogeneity had found no overall relationship with team performance (Bowers et al., 2000; Webber & Donahue, 2001), had found a small positive effect (Horwitz & Horwitz, 2007), or had found a small negative effect (Stewart, 2006) across studies. This inconsistent pattern of results led diversity scholars to conclude that “research evidence demonstrating a business case for work team diversity is by and large equivocal” (Joshi & Roh, 2007: 2), and that “much is still unknown regarding the nature of diversity, its impact on work group outcomes, and the intervening mechanisms” (Webber & Donahue, 2001: 142).

There was, however, the consensus among scholars that clarifying the mixed effects of diversity in teams would only be possible by carefully considering contextual moderators and mediating mechanisms in both theory and empirical research on diverse teams (Joshi & Roh, 2007; Mannix & Neale, 2005). As we set out to disentangle what researchers had learned from decades of research on work group diversity and reconcile conflicting perspectives and findings in this field, we developed some theoretical ideas about the mechanisms and boundary conditions under which diversity may contribute positively or negatively to team outcomes.

In this Retrospective, we begin by revisiting the original paper, highlighting the most important elements and findings, to set the foundation for the rest of the paper. Next, we comment on progress in the field in the past decade, identifying several areas of advancement. In the remainder of the paper, we reflect on three areas in need of further research: specifying the type of (cultural) diversity more carefully, opening up the black box in the relationship between cultural diversity and performance in International Business (IB) research in general, and addressing both the positive and negative aspects of cultural diversity in a more nuanced way.

A RETROSPECTIVE PERSPECTIVE ON TEAM DIVERSITY RESEARCH

Theoretical Framework: Diversity Drives Process Gains and Process Losses through Forces of Divergence and Barriers to Convergence

In our 2010 paper, we focused on one particular type of diversity: cultural differences. Most team diversity research assumed that all diversity sources (gender, ethnicity, sexual orientation, etc.) affect work groups in the same way, and there are clearly some similarities among different types of diversity (e.g., van Knippenberg & Schippers, 2007). However, there was already evidence that different diversity sources influence team outcomes in different ways (Horwitz & Horwitz, 2007; Joshi & Roh, 2009). Cultural diversity, in particular, may affect team processes and outcomes differently than other types of diversity (Lane, Maznevski, Mendenhall, & McNett, 2004; Lane & Maznevski, 2019; Taras et al., 2019). Cultural differences often operate at a subconscious level; therefore some of their effects may not be recognized or may be misattributed. At the same time, differences in culture are often a source of categorization and stereotyping (Mannix & Neale, 2005; Mortensen & Hinds, 2001), so the effects of cultural diversity may be stronger than other diversity sources.

A central tenet of our theoretical framework, as illustrated in Fig. 1, was that cultural diversity drives process gains and process losses through forces of divergence and barriers to convergence. In line with theories of social identity, social categorization, similarity attraction, and information processing, we proposed that cultural diversity increases divergent processes in teams, leading to both process gains and process losses. A divergent process that creates a process gain and may contribute positively to team performance is creativity. A divergent process that creates a process loss and could decrease team performance is conflict. Conversely, our theoretical framework suggested that cultural diversity decreases convergent processes in teams. Convergent processes are those which provide integration and cohesion, and align the team around common goals and values. Like divergent processes, some convergent processes (e.g., the emergence of group identity and mutual trust) lead to process gains and contribute positively to team performance. Some convergent processes (e.g., conformity and compliance) can hinder effective team performance, for example, by making the team closed to dissent and vulnerable to groupthink. Thus, the underlying premise of this framework was that cultural diversity does not have a direct impact on team performance but rather that the effect is indirect, mediated by process variables such as creativity, cohesion, and conflict.

Figure 1

Theoretical framework guiding the meta-analysis.

Adapted from Stahl, Maznevski, Voigt, & Jonsen (2010).

Further, we proposed that diversity affects these mediating processes or intermediate outcomes in opposing ways, i.e., cultural differences may have a positive effect on some outcomes (e.g., creativity) and an adverse effect on others (e.g., cohesion). Finally, our theoretical analysis suggested that the extent to which the process gains from diversity can be maximized, and the process losses minimized will depend on three broad sets of factors, two of which we were able to test in our meta-analysis. First, we proposed that whether team members differ in overt demographic characteristics, such as nationality, ethnicity or race (i.e., surface-level diversity), or in deep-seated psychological characteristics such as cognitions and values (i.e., deep-level diversity) would affect team outcomes in different ways; and that the hypothesized positive and negative effects of cultural diversity would depend on whether a team is composed of members from several different countries (cross-national diversity) or members from a single country (intra-national diversity). Second, we predicted that aspects of the team context, such as team tenure, the complexity of the task, and whether the team is co-located or geographically dispersed would moderate the effects of cultural diversity on outcomes. Third, our analysis suggested that the way the team process is managed (including the staffing of the team, the communication process, group decision-making, and so on) would influence team outcomes like creativity, social integration, and conflict. Since there was a paucity of research on these more process-oriented and management-related factors, we could not include them in the meta-analysis. These considerations did, however, inform our theory development.

From this framework, we developed a set of hypotheses on how intermediate team outcomes relate to cultural diversity, as well as moderator hypotheses to theorize and study how the level and type of cultural diversity and aspects of the team context might affect team outcomes. We did not formulate a specific hypothesis linking cultural diversity to overall team performance. The very idea of our theoretical framework was that process gains and process losses resulting from diversity might offset one another in their impact on team performance, and would further depend on contextual moderators. The overall impact on team performance would be difficult, if not impossible, to predict.

Meta-Analytic Results: Cultural Diversity Indeed Appears to be a ‘Double-Edged Sword’

We tested our hypotheses using a meta-analysis of existing research on cultural diversity in teams, encompassing 108 primary studies with a combined sample size of 10,632 teams. Consistent with the theoretical logic explained above, the meta-analytic results revealed that overall team performance was unrelated to diversity, with a mean effect size close to zero. Further analyses showed a number of significant effects on intermediate team outcomes. As expected, the results suggested that teams gain from increased diversity in terms of greater creativity, but diverse teams tend to suffer from process losses due to increased conflict and reduced cohesion or social integration. Diversity was positively associated with three types of conflict (task conflict, relationship conflict, process conflict) relevant to the effective functioning of a team, but only the relationship with task conflict was significant.

Contrary to our predictions, the meta-analytic results indicated that cultural diversity was unrelated to communication effectiveness and positively associated with team member satisfaction. The hypothesis that culturally diverse teams are less prone to groupthink could not be tested due to the small number of studies that examined this question empirically. Collectively, these findings supported the conclusion that cultural diversity is associated with divergent processes. More diverse teams experience the process gain of increased creativity, but also the process loss of increased task conflict. Cultural diversity’s association with convergence was less clear. More diverse teams suffered the process loss of lower social integration and cohesion, but contrary to expectations, culturally diverse teams did not experience less effective communication, and they had higher satisfaction than homogeneous teams.

Counter to our hypotheses, but not entirely unexpected to us, we found that both the level (surface vs. deep) and type (cross-national vs. intra-national) of cultural diversity were largely unrelated to the team outcomes assessed in our meta-analysis. These “null findings” had the potential to stimulate and guide further exploration in the field of cross-cultural management in general and work on culturally diverse teams in particular. Tung and her colleagues (Tung, 1993, 2008, 2016; Tung & Stahl, 2018; Tung & Verbeke, 2010) have challenged the traditional view that the effects of cross-national diversity are stronger than for intra-national diversity, arguing that “intra-national variations can often be as significant as cross-national differences” (Tung, 2008: 41). The findings of our meta-analysis supported this view by demonstrating that neither the positive effects of cultural diversity (on creativity and satisfaction) nor the adverse effects (on social integration and conflict) were more pronounced in teams composed of members from several different countries than in teams composed of members from a single country. With regard to the level of diversity, our theoretical analysis had suggested that surface-level attributes such as race or ethnicity would be associated with similarity-attraction and social-identity effects, and deep-level indicators such as cognitions and values would be related to information-processing advantages and value-incongruence effects. Contrary to our hypotheses, the findings showed that creativity, conflict, satisfaction, and social integration were all unrelated to the level (deep vs. surface) of cultural diversity.

These results highlighted the importance of specifying both the type and level of cultural diversity more carefully in research on diverse teams. Rather than assuming that cross-national and intra-national diversity, and surface-level and deep-level diversity, affect team processes and outcomes in similar ways just because the patterns of correlations are similar, we needed a more fine-grained understanding of how different diversity sources (e.g., ethnicity, language, values, cognitions) may affect the dynamics of culturally diverse teams in different ways. For example, deep-level cognitive differences such as holistic versus analytic cognitive style (e.g., Nisbett, 2003) may lead to information-processing advantages and improved group decision-making in a Sino-American work group, while deep-level value differences may simultaneously lead to social categorization processes, reduced cohesion, and increased potential for conflict in the team. We concluded that the simple categorization of surface-level and deep-level, which was ubiquitous in the diversity literature, may thus be misleading.

The use of meta-analysis to detect moderating effects not testable in the primary studies generally led to the study’s most interesting results – and substantially larger effect sizes than those obtained in the main effect analysis. The results of sub-group analyses revealed that the effects of cultural diversity on team outcomes are moderated by various aspects of the team context, but not always in the expected direction. For example, contrary to expectations, we found that cultural diversity was associated with higher levels of conflict and less effective communication in teams that had spent more time together compared to teams with less tenure. Another unanticipated, and potentially consequential, finding was that geographically dispersed teams had less conflict and more social integration than co-located teams. This finding suggests that we needed to understand much more about how diverse teams function when they are geographically dispersed. As we argued in our 2010 paper, the lower level of conflict in dispersed teams may represent an avoidance of engaging differences in values and opinion openly, or team members may simply have fewer chances to experience conflict related to value incongruence. Another explanation is that expectations play a critical role in shaping team outcomes such as conflict and social integration. Members of dispersed multicultural teams may be more attentive to cultural differences, more inclined to resolve conflicts constructively, and less prone to behavior that is detrimental to the effective functioning of a team, such as cultural stereotyping, the formation of sub-groups based on nationality, or social exclusion of team members who do not conform with the dominant values of the team.

In sum, although the meta-analytic results mostly supported the theoretical model, there were some unexpected findings that raised intriguing questions about the dynamics of culturally diverse teams and suggested promising directions for future research: Can surface-level indicators of cultural diversity, such as nationality, ethnic origin, or race, serve as a proxy for deep-level indicators such as cultural values? Does the potential for intercultural miscommunication and conflict increase, rather than decrease, in teams with longer tenure, maybe because process losses due to diversity’s effects accumulate over time, especially in teams experiencing significant pressure and strain? And could it be that a remote or virtual team context (as opposed to working in a co-located team) makes it easier, not more difficult, to avoid dysfunctional cultural clashes and create a climate of inclusion in a team? If so, what are the mechanisms that allow members of geographically dispersed teams to build trust and achieve high levels of social integration while minimizing cultural friction? And what role does communication technology play in this process? These questions are even more relevant today, in a post-Covid-19 environment and increasingly digitalized world, than they were 10 years ago when our article came out.

REFLECTING ON A DECADE OF PROGRESS

Significant progress has been made in multicultural teams research since we published our meta-analysis, of course, not only building on our research but also on other studies.

Our main recommendation was that future research should move beyond the question of whether cultural diversity has a positive or negative effect on team performance, and instead move on to fine-tune our understanding of processes as mediators and how these relationships are affected by moderators. Research now takes as a starting point that cultural diversity has both positive and negative effects, and particularly that diversity increases divergent processes and makes convergent processes more challenging. At the same time, the field of teams in management studies in general has also progressed, articulating more specific relationships between processes and emergent states (Waller, Okhuysen, & Saghafian, 2016). The impact of this precision can also be seen in international management studies. In this section, we review and comment on the progress in research on mediating processes, contextual and individual moderators, complex field studies that combine mediators and moderators, and the role of X-Culture in studying multicultural teams.

Intervening Processes that Mediate the Diversity–Performance Linkage: Creativity and Communication

Research has progressed significantly on two sets of important processes that link cultural diversity and team performance: the divergent process of creativity, and the convergent process of communication.

Culturally diverse teams are more creative

The largest effect size in our meta-analysis indicated that teams with high cultural diversity were more creative than teams with low cultural diversity. Because creativity is part of the process of innovating for new solutions (and often the two are even measured interchangeably in teams research), it is a very important intermediate team outcome. When we conducted our meta-analysis, we were disappointed not to be able to test any moderators for this relationship as we had only five studies that included creativity. Research on the relationship between diversity (including cultural diversity) and creativity in teams has almost exploded in the past 10 years (e.g., Crotty & Brett, 2012; Jang, 2017; Leung & Wang, 2015). The cultural studies were summarized well in a recent meta-analysis by Wang et al. (2019). Based on 44 studies conducted between 1985 and early 2018 (most of them after 2010), they reinforced that deep-level diversity is associated with more creativity due to its relationship with higher information diversity. This effect is moderated to be (more) positive when the team is collocated or is engaging in a task with high interdependence. Surface-level diversity, which can raise social identity threats, was negatively related to creativity and innovation for simple tasks and unrelated for complex tasks (other moderators were not significant). Further research will certainly explore more nuances around these relationships.

Language differences raise barriers to communication, overcoming them requires paying attention to other team dynamics

With respect to communication, there has been a long-overdue increase in research on the role of language differences in multicultural teams’ communication. We were not able to test the role of language differences at all in our meta-analysis, as there were not enough studies. The Journal of International Business Studies special issue on language (Brannen, Piekkari, & Tietze, 2014) set a landmark in this field. Two papers in that special issue focused on the relationship between language, communication, and other processes in multicultural teams. Tenzer, Pudelko, and Harzing (2014) looked at the relationship between language barriers and trust, showing that language differences and cultural differences are related in complex ways in their activation, perception, and impact. Hinds, Neeley, and Cramton (2014) showed the processes through which language differences and fluency levels influence power dynamics in teams, and the moderation of managing those dynamics through emotional regulation. These studies showed that language and culture are clearly related to each other, but the language and culture diversity activate and affect communication and other processes in multicultural teams. They have set a strong foundation for examining how communication barriers can be overcome.

Contextual Moderators: Geographic Configuration and Virtual Teams

Probably the most important, and certainly the most-studied, contextual moderator affecting multicultural teams is the extent to which they are geographically dispersed. Global Virtual Teams (GVTs) are internationally distributed teams working on joint tasks with international components or implications (Gibson & Gibbs, 2006; Maznevski & Chudoba, 2000), and they usually conduct most or all of their work together using communication technologies rather than face-to-face. As demonstrated by a large body of research on GVTs and virtual work, working together over technology increases the challenges faced by teams of any composition (Jimenez, Boehe, Taras, & Caprar, 2017; Reiche, Bird, Mendenhall, & Osland, 2017; Zander, Mockaitis, & Butler, 2012). In our meta-analysis, we were surprised with the finding that virtual teams had less conflict and more social integration than co-located teams. We suggested possible explanations, but the already-published research did not help us resolve the finding. Since then the research on GVTs has flourished. As noted by Zander and colleagues (2012), in GVTs it is not always easy to tease apart the extent to which dynamics are influenced by diverse composition, dispersed configuration (and virtual communication), or their combination. The effects on mediators of working virtually are similar to the effects of cultural diversity: both provide opportunities for divergence but raise barriers to convergence. The research patterns suggest that their effects usually multiply each other’s impact, especially in increasing barriers (Jimenez et al., 2017). Working virtually decreases the amount of non-verbal communication, and, therefore, decreases the amount of information available to develop judgements about trust, cohesion, and commitment (Reiche et al., 2017; Scott & Wildman, 2015). Working across time zones and across infrastructures increases the coordination costs of a project dramatically (Jonsen, Maznevski, & Davison, 2012). On the other hand, by being able to work across time zones, GVTs have the opportunity to work around the clock and access different networks and institutional resources (Wildman & Griffith, 2015).

In the past decade, more fine-grained research has been conducted on the relationship between cultural diversity and team mediators in dispersed teams. For example, Zakaria (2017) examined a large team’s e-mail communications through different stages of a decision-making process. She discovered that people from high-context cultures tended to use more indirect structures in their written communication, compared with people from low-context cultures. However, she also saw that people changed their communication style with different receivers and through different decision stages, suggesting that culture is not a simple determiner of virtual communication style. Klitmøller, Schneider, and Jonsen (2015) found that when team members from different cultures communicated using verbal media (e.g., telephone) rather than written media (e.g., e-mails), they were more likely to engage in social categorization leading to negative stereotypes and reduced trust.

Some recent research has begun to uncover the mechanisms through which working virtually can lead to positive dynamics and outcomes in GVTs. Nurmi and Hinds (2016), in a qualitative study, discovered that people engaged in global virtual work appreciated the opportunities to learn and innovate, and this in turn increased their satisfaction and engagement. The effect was enhanced when team members had the opportunity for off-job recovery, and did not need to be available all the time. Magnusson, Schuster, and Taras (2014) discovered that GVTs with a high average level of motivational cultural intelligence (CQ) put in extra effort to overcome perceived cultural differences, and increased their communication and conflict management effectiveness. Zakaria and Mohd Yusof (2020) examined the development of trust in GVTs over successive stages of a project. They saw that responsiveness was key to signaling reliability and adaptation, and therefore initiating cycles of increased trust-building. Responsiveness, though, was often affected by internet availability and time zone differences, and teams that were not mindful about these factors could ignore each other and spiral into cycles of decreased trust. Teams with members who were willing to engage in constructive conflict and revisit issues overcame the barriers to responsiveness and therefore trust-building. The common thread across these recent studies is the team members’ willingness to put in extra effort to overcome barriers, and future research will examine the antecedents of motivation further.

Individual-Level Factors: Individual Team Members Can Make a Big Difference

Recent research has addressed the effects of three important categories of individual-level moderators: individual knowledge and skills for working across cultures, individuals who are bi- or multi-cultural themselves, and team leader behaviors.

Cross-cultural competences and cultural intelligence

An extensive stream of research has examined cross-cultural competences and cultural intelligence (CQ) (Bird, Mendenhall, Stevens, & Oddou, 2010; Szkudlarek, Romani, Caprar, & Osland, 2020; Thomas et al., 2015), the set of knowledge, skills, and abilities required to interact effectively with people from different cultural backgrounds. This research has demonstrated clearly that increased cross-cultural competences lead to more effective communication, conflict management, trust building, and other processes and emergent states that mediate high team multicultural performance. Recent research has linked these competences more specifically to the team context, affirming their importance in teams. For example, followers’ increased cultural intelligence led to better individual task performance in global virtual teams, especially when contact intensity with the leader increased (Presbitero, 2020a). Further, when perceived cultural distance is high, a leader is more likely to rate a team member’s performance as high if the leader also perceives that the team member has high cultural intelligence (Presbitero, 2020b). Another study showed that team members are more likely to voice disagreement or new information if they or the leader has high cultural intelligence (Ng, Van Dyne, & Ang, 2019). Team dynamics over time are characterized by reciprocal causation, so it is also important that individuals develop cross-cultural competences through their experiences in multicultural teams (Caligiuri, 2015).

Individuals who are multicultural

Another important stream of research on the impact of individuals has evolved around multicultural individuals and how they affect organizational phenomena, as described recently in a comprehensive review by Vora et al. (2019). Research shows that multicultural individuals have a positive impact on several processes important in multicultural teams, including boundary-spanning communication (Yagi & Kleinberg, 2011) and social capital development (Fitzsimmons, Liao, & Thomas, 2017). In a recent study, Backmann and colleagues explored how cultural identity plurality is related to bridging processes in multinational teams (Backmann, Kanitz, Tian, Hoffmann, & Hoegl, 2020). They confirmed that individuals with more cultural identities engage in these bridging behaviors, with cultural intelligence as a mediator. We are sure to see more studies illustrating how multicultural individuals contribute to multicultural teams (Dau, 2016; Fitzsimmons, Miska, & Stahl, 2011). Different individuals have different ways of integrating (or not) their separate cultural identities, and the identities within an individual may be seen by others as more or less positive (Fitzsimmons, 2013). Because this aspect of individual-level multiculturalism is related to the intersectionality of diversities (see below), it is important that future research examine the effects of different types of multiculturalism.

Individuals as team leaders

Finally, the leader as an individual-level moderator has also been examined. The field of global leadership has progressed significantly (Mendenhall et al., 2018), and global teams are recognized as in important context both for practicing global leadership and for developing global leaders (Maznevski & Chui, 2018). Zander and colleagues (2012) applied the literature on team leadership to global teams and concluded that transformational leadership combined with cross-cultural competences provide appropriate leadership for helping multicultural teams to overcome barriers and realize opportunities. An increasing body of research focuses specifically on the role of the leader in multicultural teams. For example, Tröster and van Knippenberg (2012) found that leaders who were open to non-dominant ideas increased affective commitment, which in turn encouraged team members to voice local and contradictory ideas. Eisenberg and Mattarelli (2017) suggest that leaders who are effective cultural brokers help to bridge differences across subgroups, leading to more effective knowledge exchange. Lisak and Erez (2015) found that individuals high on all three of cultural intelligence, global identity, and openness to cultural diversity were more likely to emerge as leaders of multicultural teams. In another study, Lisak, Erez, Sui, and Lee (2016) demonstrated, in an R&D organization, that leaders with a strong global identity helped their multicultural teams focus on team-shared goals and innovate together. Studies like these, bringing together leadership, cross-cultural effectiveness, and team dynamics to explain effectiveness in multicultural teams, develop important knowledge for practitioners and we hope to see more of them.

Complex Combinations: Mediators, Moderators, and Multiple Levels of Analysis

We also note that we have a rise in more complex studies looking at combinations of mediators and moderators, including configurations of moderators at different levels of analysis simultaneously. One approach being applied more frequently in these studies is social network methodology. For example, Haas and Cummings (2015) applied social network methods within teams to examine knowledge exchange. They demonstrated that individual differences related to international configuration and different types of personal experiences affected the knowledge exchange process differently, with some flows moderated by people’s prior experience together. Vahtera et al. (2017) looked both within and across teams in an organization to uncover the antecedents of negative perceptions of social identity groups, to develop our understanding of how multicultural and globally dispersed teams link different parts of an organization together. In another study on networks of global virtual teams, Mattarelli et al. (2017) interviewed the brokers who explicitly connect the dispersed teams. They found that brokers who have a personal professional identity related to growth and learning tend to have more accurate perceptions about distant co-workers and enable teams to adapt their routines and do more challenging work.

Other studies have incorporated more analysis of the organizational environment into the multi-level complex research models. Hajro, Gibson, and Pudelko (2017) explored how a multinational organization’s climate with respect to the role of diversity affected knowledge exchange in multinational teams. They found that the more effective pattern of oscillating between assertive and cooperative knowledge exchange was more prevalent if the organization’s climate explicitly valued using cultural diversity to inform work. In the most comprehensive study examining multiple mediators and moderators over time, Gibson, Dunlop, and Cordery (2019) reported on a longitudinal field study of global teams in a multinational mining organization. They found that some formalization policies (but not others) had a positive effect on teams’ effectiveness, but only when implemented at some (but not other) points in the team’s journey. These effects were moderated by team members’ personal needs for structure.

These complex studies are challenging to undertake. They adapt multiple methodologies and often examine cause–effect relationships over time in order to parse the separate effects. We are pleased to see the field having evolved to the extent of being able to test relationships in these types of studies.

X-Culture as an Early Stage Research Lab

Finally, it is important to acknowledge the contribution of the X-Culture project to research on multicultural teams in the past decade. X-Culture is a multi-lateral initiative bringing together thousands of undergraduate and graduate students into global virtual teams each semester, to work on consulting projects for real clients (Taras, 2020). X-Culture was launched in 2010 with the dual goals of providing a relevant and valuable learning experience for students, and an opportunity for scholars to conduct research on virtual teams. In addition to delivering on its educational mandate, the initiative delivers for international management scholars the equivalent of social psychology’s student labs for studying dynamics in the early stages of research. The X-Culture research program is now bearing real fruit in providing tests of hypotheses before taking them to more complex field settings (e.g., Jang, 2017; Jimenez et al., 2017) and deconstructing the different effects of input and process variables more precisely (e.g., Taras et al., 2019). It is an important contribution and will allow us to turbo-charge more of our complex research designs.

In sum, the field of multicultural teams has made significant progress in the last decade, especially in the focus on different mediators and moderators in complex settings. The studies build on the theoretical foundations of information processing and social identity effects, and develop further richness around these perspectives by looking at how they explain performance in different empirical contexts. At the same time, the studies highlighted in this review represent “best practice” and not yet the norm for the field. In the next section, we turn to recommendations for future research, which include building on these best practices and extending them to other areas in International Business research.

LOOKING AHEAD TO THE NEXT DECADE

In this section, we focus on three areas of relevance to future research. First, we suggest there is interesting and important work to be done on the constructs of culture and cultural diversity in multicultural teams research. Second, we recommend a closer examination of the relationship between cultural diversity and performance across IB research, examining the relationships among multiple types, sources, and levels of (cultural) diversity more precisely. Finally, we advocate researching the positive effects of diversity while communicating the negative and mixed effects more carefully.

Culture and Cultural Diversity in Teams

One of the unexpected findings of our 2010 meta-analysis was that the level of cultural diversity (surface vs. deep) was unrelated to team outcomes such as creativity, conflict, satisfaction, and social integration, leading us to conclude that the simple categorization of surface-level and deep-level may be misleading and that future research on diverse teams needs to specify the level of cultural diversity more carefully. Although some studies still do not address or specify the level of cultural diversity, most are much more careful now. At the same time, though, they tend to define culture and diversity quite narrowly. Even while providing caveats that culture is not necessarily about nationality, most multicultural teams studies look only at national cultures, and within that most focus only on a few dimensions for differentiating cultures and defining multicultural team diversity. Zellmer-Bruhn and Maloney, in their recent review of cross-cultural teamwork, note that this narrow conceptualization of both culture and diversity has hindered the growth of the field, and they recommend expanding both of these constructs (Zellmer-Bruhn & Maloney, 2020).

Three recent studies have provided important perspectives for future research. The aforementioned meta-analysis by Wang et al. (2019) offers a fine-grained analysis of how surface-level and deep-level diversity attributes, in conjunction with team characteristics such as team virtuality, task complexity, and task interdependence, influence team creativity and innovation. In another study, Taras et al. (2019) abandoned the distinction between surface-level and deep-level indicators altogether and instead proposed a refined theoretical model that differentiates between the effects of personal versus contextual diversity and specifies how these distinct forms of diversity affect task outcomes and psychological outcomes in different ways. Their results revealed that contextual diversity (defined as differences in the characteristics of the environments that the team members come from, including economic development, human development, and national cultural orientations) has a positive effect on task outcomes; and personal diversity (defined as differences pertaining to the personal characteristics of the team members, including differences in age, gender, language skills, and personal values) has a negative effect on psychological outcomes. Tasheva and Hillman’s comprehensive conceptual framework (Tasheva & Hillman, 2019) suggests differentiating among three types of diversity (human capital, social capital, and demographic) at two levels of analysis (within-person and among people in a team). They argue that for tasks with different levels of required interdependence, these sources of diversity may be complementary or substitutes. These recent studies should lead the way for more research on how different diversity attributes, moderated by team characteristics and contextual conditions, affect processes and outcomes in diverse teams.

On a related note, the field must develop and adopt much more precise ways of measuring diversity. When we conducted our meta-analysis, diversity was usually measured by the proportion of the team in different categories. If there were two categories (e.g., domestic and international), the measure could simply be proportion of the minority category. If there were more than two categories, multi-proportionality was calculated to create a heterogeneity index (Blau, 1977). If the diversity variable was continuous, for example value diversity as measured by a self-assessment survey, diversity was measured by variance on those variables. Research today is still dominated by the same types of measures. Studies of psychic and perceptual distance, though, have popularized more sophisticated approaches to measuring perceptual distance between two people or two entities (Beugelsdijk, Kostova, Kunst, Spadafora, & van Essen, 2018; Zaheer, Schomaker, & Nachum, 2012), from comprehensive distance arrays to spatial dependence drawing from geography techniques (e.g., Plummer, 2010). We are beginning to see this perspective in multicultural teams, but so far it has mainly been applied to dyads within teams (e.g., Magnusson et al., 2014; Ng et al., 2019). Mathematically, diversity could be measured as distance among multiple group members. If we could measure the extent of diversity more precisely, we would be able to explore, for example, whether there are thresholds of diversity above which specific leadership interventions are essential.

In addition, it is abundantly clear that more research is needed on intersections of different types of diversity. Multicultural teams are rarely diverse only on culture. Especially for global work, they are usually diverse on gender, functional background, and possibly even organizational representation. One finding we found interesting in our meta-analysis was not in our 2010 publication. While there was no direct effect between cultural diversity and performance, teams that were high on both cultural diversity and gender diversity had significantly lower performance (effect size − 0.11, p < .001), lower satisfaction (effect size 0.31, p < .01), and lower cohesion (effect size − 0.13, p < .01) (Stahl, Maznevski, Voigt, & Jonsen, 2006). Space prohibited our exploration of that result in the article, but we have reflected on it since.

It is possible that we were picking up the effect of faultlines or other patterns of intersectionality. Most research on diverse teams focuses on one “source” of diversity at a time, seeing other sources as moderators. We followed this pattern when we published the meta-analysis. Faultline research examines the alignment of two or more types of diversity in a team, such as gender and culture or nationality and function (Lau & Murnighan, 1998, 2005). In general, research has found that faultlines tend to moderate relationships between diversity and outcomes, such that any negative impact of diversity is more negative when faultlines are “activated,” or perceived as salient by the team’s members (e.g., Antino, Rico, & Thatcher, 2019; Bezrukova, Spell, Caldwell, & Burger, 2016; Spoelma & Ellis, 2017). In other words, faultlines split groups into “us” and “them,” and it is difficult for groups to bridge these faultlines. Empirical research on faultlines has rarely examined culture (especially deep-level elements of culture, such as values) as one of the dimensions of diversity. There is a well-established stream of research on faultlines in top management teams (e.g., Orlando Curtae, Wu, Markoczy, & Chung, 2019), and another on faultlines based on information- and identity-based characteristics (Spoelma & Ellis, 2017). The findings from this research are highly relevant to multicultural teams. For example, Spoelma and Ellis found that external threats mitigated the negative effects of identity-based (gender) faultlines on team creativity.

Intersectionality acknowledges the fact that all individuals identify with and are influenced by multiple identities simultaneously, and all of these identities can be considered sources of diversity if they differ within a team. IB research focuses on cultural diversity because cultures influence shared experiences of, expectations for, and assessments of social interactions and decision-making, which in turn affect the conduct of business. Experiences of other social identities also greatly affect shared social interactions and decision-making. For example, gender, sexual orientation, race, neurological characteristics, and socio-economic status are all associated with both individual and commonly shared perspectives. Their combinations have complex effects, depending in part on whether they reinforce or complement each other with respect to the way they are perceived by others or are embedded within the society’s or an organization’s power systems.

In today’s environment, it is no longer enough to look at one source of diversity alone. In the future, we hope to see better articulation of dimensions of diversity, more precise measurement of diversity, and richer examination of configurations and intersectionalities of diversity.

Open the Black Box of the Diversity–Performance Link Beyond Multicultural Teams

As we noted at the outset of this Retrospective, research on the effects of cultural diversity in teams had yielded inconclusive, and sometimes conflicting, results. Thus, one of the aims of our study was to clarify the mixed effects of cultural diversity in teams and to reconcile conflicting perspectives and past results. It is important to note, though, that the pattern we observed was not confined to studies on culturally diverse work groups. Rather, it had been observed with respect to cultural diversity and cultural distance in a wide range of IB contexts and different subfields within IB research, including work on MNE performance, foreign market entry, FDI and expansion, international alliances and M&A, cross-border knowledge transfer and learning, and other research on culture in IB (Leung, Bhagat, Buchan, Erez, & Gibson, 2005; Tihanyi, Griffith, & Russell, 2005). For example, Tihanyi et al. (2005: 270) summarized the results of their meta-analysis of research on cultural distance as follows: “[R]esults failed to provide statistical evidence of significant relationships between cultural distance and entry mode choice, international diversification, and MNE performance. The examination of moderator effects, however, yielded important results. …Substantial additional research is needed before the role of cultural distance is fully understood.” Similarly, Cartwright and Schoenberg (2006: 3–4), in a review of research on the role of culture in cross-border M&A, concluded that “the relationship between culture and performance continues to intrigue and confuse researchers as studies examining this link in relation to international M&A have produced rather mixed, and often contradictory results. … Existing research remains incomplete in some way.” Taken together, the evidence indicated that across various domains of IB studies, effect sizes for the relationship between cultural differences (or related constructs such as cultural distance and cultural diversity) and performance outcomes were generally small and often not significantly different from zero. Effects seemed to be contingent on contextual factors, and results about the direction of effects were inconclusive and “confusing.”

In our meta-analysis of team diversity research, we found two complementary theoretical explanations for a zero-direct-effect relationship between cultural diversity and team performance and other “anomalies” identified in past research: simultaneous positive and negative effects on intermediate outcomes; and, moderated relationships with intermediate outcomes. We also found the effects of study design characteristics, which could have been related to different methods testing the same thing, or the different methodologies in fact identifying different relationships. In addition, there were alternative explanations we were unable to test in our meta-analysis, but that help explain the sometimes-puzzling results obtained in prior research. These explanations are summarized in Table 1. Although these explanations have been explored in research on multicultural teams, they have not been as systematically applied in other areas of IB research, and we suggest that research would benefit from adopting these perspectives. We look more closely at each of the explanations in turn.

Table 1

Opening up the black box of the diversity–performance link: explanations for a zero-direct-effect relationship and other “anomalies” found in previous research

Explanations for a zero-direct-effect relationship
1. Positive and negative effects on intermediate processes and outcomes, which partly or fully offset one another.
2. Moderated relationships (e.g., the impact of contextual and management-related factors).
3. Effects of study design and sample characteristics.
4. Non-linear effects on outcome variables (e.g., the “too-much-of-a-good-thing” effect).

Positive and negative effects on mediating variables

Our theoretical framework applied the input-mediator-output approach to understanding group relations (Lepine, Piccolo, Jackson, Mathieu, & Saul, 2008; Mathieu, Maynard, Rapp, & Gilson, 2008). We proposed that cultural diversity drives process gains and process losses by increasing divergence and raising barriers to convergence (Earley & Gibson, 2002; Mannix & Neale, 2005). We anticipated that these two effects on mediating variables might partially or fully cancel each other out, and the results of our meta-analysis largely supported this pattern. Overall, team performance was unrelated to diversity, and diverse teams’ gains from greater creativity and higher team member satisfaction were about evenly matched with losses from increased task conflict and lower social integration. As described above, research on multicultural teams in the past decade has studied these mediating variables with much more depth.

This mediating pattern may also explain the conflicting or zero-direct-effect relationships between cultural differences and performance found in other domains of IB research, such as work on entry mode choice, MNE performance, cross-border knowledge transfer and learning, and international alliances and M&A (Beugelsdijk et al., 2018; Stahl, Tung, Kostova, & Zellmer-Bruhn, 2016; Tihanyi et al., 2005; Tung & Verbeke, 2010). For example, Tihanyi et al. (2005: 278–279) concluded from the results of their meta-analysis that “cultural distance does not appear to be directly related to entry mode choice, international diversification, or MNE performance” and suggest that this “disturbing” pattern of results might be due to intervening processes, such as MNEs’ risk-minimization strategies and learning effects. As another example, and closely paralleling the results of our meta-analysis of team diversity research, the findings of a meta-analysis of research on cultural distance and cross-border M&A (Stahl & Voigt, 2008) revealed that post-merger performance was unrelated to cultural differences, with a mean effect size of zero. However, the effect sizes derived from the primary research studies ranged widely, from highly negative to highly positive. Further analyses revealed that cultural distance predicted outcomes such as sociocultural integration, synergy realization, and financial performance in opposing ways (e.g., the effect on synergy creation was positive, the effect on sociocultural integration was negative), canceling each other out in their impact on overall M&A performance.

Moderated relationships

Our meta-analysis clearly suggested that cultural differences mattered to the performance of diverse teams, but they indeed seemed to present a “double-edged sword” or mixed blessing – they could be positively or negatively associated with the same mediating outcome, depending on other factors. Sub-group analyses unearthed a number of contextual factors that moderated the relationships between cultural diversity and intermediate team outcomes. As reviewed above, recent research on multicultural teams has enhanced further understanding of these moderated relationships, with some studies examining them in complex combinations.

Again, there appear to be some striking similarities between the dynamics of cultural diversity within teams and the way cultural differences affect processes and outcomes in other organizational contexts, such as international R&D collaborations, joint ventures, and M&A. For example, Stahl and Voigt’s (2008) meta-analysis on M&A revealed the presence of contextual influences, such as the degree of firm relatedness and the integration approach taken by the acquirer, which moderated the relationships between cultural distance and various M&A outcomes. Tihanyi et al., in their meta-analysis on cultural distance effects in IB, also found that “[t]he examination of moderator effects…, yielded important results. The cultural distance–international diversification relationship was negative for high-technology industries, while it was positive for other industries.” (Tihanyi et al., 2005: 270).

However, research on cultural distance and cultural diversity in other parts of IB has not yet illuminated the role of moderators in sophisticated ways. For example, Beugelsdijk et al. (2018), in probably the most comprehensive meta-analysis of research on cultural distance and firm internationalization to date, found opposing cultural distance effects depending on whether home and host countries are developed or emerging markets. Their results suggest that when the home country is an emerging market, the negative effect of cultural distance on performance turns positive; alternatively, when the host country is an emerging market, the negative effect of cultural distance on performance becomes even more negative. However, due to the small number of primary studies that tested these contingencies, they were unable to explore alternative explanations of the home and host country and developed and emerging market moderator effects. Importantly, Beugelsdijk and colleagues observed that there is a paucity of work examining the role of management-related factors. The few studies that addressed management aspects such as the benefits of the transfer of practices yielded large effect sizes of cultural distance. They concluded: “Understanding when and for which aspects of the internationalization process cultural differences really matter is a necessary step in learning how to manage and possibly leverage such differences” (Beugelsdijk et al., 2018: 123).

Effects of study design and sample characteristics

This leads us to another important observation, namely how aspects of the research design might impact study results. Our meta-analysis revealed that the study setting (laboratory or field) and the location where the research was carried out (North America or elsewhere) influenced the research findings, and so did the study design (longitudinal or cross-sectional). Regarding the latter, few studies used a longitudinal design to look at intervening processes and how they unfold over time, or to test how team outcomes change over time. Whereas the large number of studies that utilized a cross-sectional design found that cultural diversity, on average, was negatively associated with team performance, the considerably smaller number of studies that utilized a longitudinal design yielded a positive mean effect size, suggesting that diverse teams outperform homogenous teams over time. Was this effect simply study characteristics, or was there more going on about longitudinal tasks providing more context for the benefits of diversity?

As another example, the results of our meta-analysis suggested that the geographic region where the research was carried out influenced the study results. While studies conducted outside of North America found that cultural diversity is positively associated with conflict, studies carried out in North America found no such relationship. Obviously, these results must be interpreted with some caution – for example, it cannot simply be concluded that the members of historically diverse societies such as Canada or the US, compared to countries like Germany or Japan, are better able to manage conflict associated with cultural diversity. It could also be that cross-country cultural diversity outside North America is greater than or works differently from cultural diversity within North America (mostly within either Canada or the United States, not cross-country). But these findings alert us to the fact that demographic attributes like race or ethnicity play a different role in different countries.

Thus, researchers need to pay more attention to how aspects of the study design, location, and sample characteristics may affect the cultural dynamics of the research being conducted. As we reviewed, there have been more complex field studies published about multicultural teams in the past decade. They provide precise theorizing and testing about elements of the study design and context (e.g., Cramton & Hinds, 2014; Gibson et al., 2019; Hajro et al., 2017; Hinds et al., 2014). Yet not all studies on multicultural teams follow this pattern, and we encourage more studies to do this.

Other areas in IB should follow suit. For example, Beugelsdijk et al. (2018), in their meta-analysis of the literature on cultural distance and firm internationalization, found that cultural distance effects are sensitive to the measurement and operationalization of cultural distance (e.g., perceptual measures or cultural distance index based on secondary data), as well as the sample structure (e.g., developed or emerging markets). Again, due to the small number of studies that addressed these contingencies, the authors were unable to fully explore the effects of study design characteristics, but the results of this meta-analysis highlight their importance and present opportunities for future research.

Non-linear effects of diversity on outcome variables

There is another possible reason for the inconsistent and sometimes confusing results obtained in past research that is methodological in nature. It seems likely that the effects of cultural differences are not always linear but may vary depending on the amount of cultural diversity present in a team or organization. Drawing on March’s (1991) exploration versus exploitation organizational learning framework, Stahl and Tung (2015) proposed that across various IB contexts and levels of analysis cultural diversity exhibits an inverse U-shaped relationship with the capacity for learning and other positive outcomes. That is, as diversity increases – across individuals in an organization, across different sub-units in an MNE, across different groups in a societal setting – the capacity for learning and the potential for synergy increases, but beyond a certain threshold level, the relationship turns downwards as the amount of diversity becomes overwhelming and hard to manage or cope with. In other words, there might be a “too-much-of-a-good-thing” effect (Grant & Schwartz, 2011) when it comes to cultural differences and cultural diversity. This is consistent with the general idea that many positive qualities have costs that at higher levels may begin to outweigh their benefits. The greater variety of ideas, perspectives and experiences that a diverse team brings to the table may thus be an asset and contribute to improved group decision-making as long as the group’s diversity is at a moderate level, but at very high levels (e.g., if team members come from very different cultures or if several diversity sources combine and their effects align or accumulate) they can be harmful and lead to miscommunication, lack of cohesion, and conflict (e.g., Earley & Mosakowski, 2000; Lau & Murnighan, 1998).

Becoming serious about opening the black box of culture’s impact

In general, research on culture in IB has suffered from overly simplistic assumptions about cause and effect relationships, specifically the assumption that cultural differences have a direct effect on performance, regardless of the process and context. Few studies have attempted to test more complex models that include various levels of culture, moderating variables, intervening processes, and multiple outcome measures. To advance research on culture in IB, we need to adopt more complex “input-process-output-context” models. Gibson, Maznevski, and Kirkman (2009) caution that such complex moderated mediation models necessitate changes in research design, particularly with regard to sampling and data analysis. For example, investigating the influence of multiple moderator variables and intervening mechanisms concurrently may require more complex field research, including the need to engage in longitudinal studies, especially when these moderators include process-oriented and management-related factors (Tung & Stahl, 2018).

More research on the Positive Effects of Diversity, and More Careful Communication of the Negative and Mixed Effects

It is now widely accepted among diversity and cross-cultural management scholars that cultural diversity has both positive and negative effects, and our 2010 meta-analysis indeed provided evidence supporting the “double-edged sword” nature of cultural diversity in teams. However, the field of team diversity research is still characterized by an over-emphasis on the negative, and researchers have expended little effort on developing new theoretical perspectives highlighting the positive dynamics of cultural diversity. In their review of the team diversity literature, Stahl, Mäkelä, Zander, and Maznevski (2010) concluded that all but one of the dominant theoretical perspectives on cultural diversity in teams are consistent with the problem-focused view of diversity. They emphasize the process losses resulting from reduced perceptions of similarity-attraction among team members; negative biases associated with social categorization processes; feelings of dislike and resentment due to incongruent values; and communication barriers resulting from differences in language and communication styles. The only “positive” exception is information-processing theory, which holds that diversity may lead to information processing advantages due to team members’ different perspectives, knowledge bases, and decision-making styles, which, if properly harnessed, can enhance creativity and decision-making quality. This positive exception has been studied extensively already in the relationship between cultural (and other types of) diversity and creativity (Wang et al., 2019), but other positive aspects of cultural diversity have not yet been researched.

Stahl and Tung (2015) have argued that current theory in IB, in general, tends to over-emphasize the “dark side” of culture by focusing primarily on the difficulties, costs, and risks associated with cultural differences and underlying drivers while paying scant attention to the potentially positive effects of cultural diversity and how they can be explained. Their content analysis of 1141 articles on culture in IB published in the Journal of International Business Studies over a 24-year time period revealed a 17:1 imbalance of negative over positive theoretical assumptions in studying the role of culture in various IB contexts, demonstrating that there is an overwhelming emphasis on the liabilities associated with cultural differences. Thus, there is a need to develop new theoretical models to examine the positive aspects of cross-cultural dynamics and account for the mixed effects of cultural differences observed in various sub-fields within the IB field, including work on culturally diverse teams.

At the same time, ironically, the management perspective on diversity tends to be overwhelmingly positive. McKinsey & Company reported again in 2020 that “The business case for inclusion and [gender, ethnic and cultural] diversity is stronger than ever. For diverse companies, the likelihood of outperforming industry peers on profitability has increased over time, while the penalties are getting steeper for lack of diversity” (Dixon-Fyle, Dolan, Hunt, & Prince, 2020). According to their research, companies in the top quartile of gender and ethnic diversity in the executive team are likely to perform better than the industry average on earnings before interest and taxes (EBIT), while companies in the bottom quartile of gender and ethnic diversity in the executive team are likely to perform worse than the industry average on EBIT. Boston Consulting Group (BCG) reported that companies with more diverse (on many dimensions, including culture and ethnicity) management teams had higher proportions of their revenue coming from innovation (Lorenzo, Voigt, Tsusaka, Krentz, & Abouzahr, 2018). The World Economic Forum declared that “The business case for diversity in the workplace is now overwhelming” (Eswaran, 2019). These studies tend to use large cross-sectional samples with public data to test correlations, although some (e.g., McKinsey) are now tracking firms over time and can examine within-firm trends. Further, these studies rarely examine mediators or moderators in any depth or with rigorous research designs. A Google search “Does diversity improve performance?” produces a plethora of results from the popular press and consulting firms “showing” that diversity increases company and board performance, and very few links questioning the notion (the few are inevitably academic studies).

We find it curious and unsettling that the well-established finding from academia, that diversity is a double-edged sword, tends not to be shared by or accepted in practice. In our own experience with leaders in organizations, the message “diversity is always good” does not fit with most people’s lived experience. Yet the message itself has the potential to shut down the very dialogue it needs to open up by denying any possibility that diversity can lead to lower performance, and dismissing the importance of contingencies (moderator variables). We suggest this is an area in which practitioner-scholar dialogue has not yet been sufficient, and we must take our share of responsibility for that. For example, our research on mediators and moderators in multicultural teams should be better integrated with research on multicultural competences development, as illustrated by the studies on cross-cultural competences and CQ in teams reviewed above. These approaches should be applied not just to international management situations but also more frequently (and with more legitimacy) to within-country situations and intersectionalities (Maznevski, 2020).

Conclusion: More Research Needed

With global tensions around immigration, refugees, prejudice, and pandemics at new heights, research on multicultural teams has never been more important. At the same time, the potential benefits from multicultural teams, such as innovation, improved decision-making, and engagement in distributed working relationships, have never been more needed. Multinational enterprises and the societies in which they operate need to know how diversity is related to performance, in which configurations and which contexts diversity matters most, and, importantly, how the potential benefits of diversity can be unleashed while the frictions arising from diversity can be mitigated. In the field of international business, in general, diversities of all kinds will likely widen in light of the rising economic power of emerging markets, the seismic changes that have occurred in the international political landscape, and the growing economic, social and environmental problems facing the world (George, Howard-Grenville, Joshi, & Tihanyi, 2016). These developments highlight the urgent need to understand diversity and its impacts in today’s world to avert mistakes and, possibly, cataclysmic consequences that cultural misunderstanding, discordance, and conflict may bring (Tung & Stahl, 2018).

By studying diverse teams and organizations with a broader sense of culture, more nuanced measurement of diversity, and more fine-grained examination of the relationships between multiple sources of diversity and combinations of moderators and mediators, we can improve our understanding of how cultural diversity can create value for global organizations and society, such as learning through cross-border knowledge-sharing and collaboration, leveraging the talent of migrants and refugees, and unleashing of creative potential in diverse teams. Addressing these research questions will also allow us to offer managers leading diverse teams and organizations much-needed guidance and the ability to build capacity in this important area.

We are pleased that the field has grown in richness in the past 10 years, enough that we can offer direction for today’s scholarship, multinational enterprises and societies. Yet we find ourselves impatient, reflecting that the progress is not nearly enough. As a field, we have not kept pace with the accelerating needs of our stakeholders. We do not have adequate answers or clear advice for the ever-more-complex situations that leaders of global organizations face. We hope that scholars in the field share our sense of urgency and that the reflections and suggestions in this Retrospective help propel this research forward.