Leadership-promoted diversity climate and group identification

Purpose – Besides diversity ’ s positive effects, groups of “ we ” against “ them ” may form in accordance with social categorization theory, showing diversity ’ s negative consequences. The authors aim to reconcile these results and examine their boundary conditions. Design/methodology/approach – The authors studied 584 working professionals from five contexts (transnational companies dealing with multicultural interactions) and analyzed data using moderated- mediation procedures. Findings – A leader-promoting diversity climate plays a crucial role in moderating the negative relationship between perceived dissimilarity and group identification, which is mediated by value dissimilarity. Originality/value – This study mainly contributes by treating dissimilarity as a multicomponent construct, emphasizing the crucial differences embodied in various conceptualizations of dissimilarity – namely visible and value dissimilarity. For dissimilarity to result in group identification, the results highlight leaders ’ crucial role, beyond that of organizations and individuals, in stimulating a diversity-embracing climate in work units.


Introduction
With globalization and demographic changes, diversity in the workplace exposes contemporary leaders to several challenges and threats (Hennekam et al., 2019;Lumby, 2009). Previous studies have aimed to determine diversity's effects on individual and team behavior and on work outcomes; however, findings remain ambiguous (see Van Knippenberg et al., 2004). On the one hand, diversity can open up a larger pool of talent (Cox, 1991) and thus enhance creativity and innovation at work (Van Knippenberg et al., 2004). On the other hand, diversity's positive effects can be quickly overshadowed by adverse behavioral and affective outcomes, such as reduced social cohesion, relational conflicts, miscommunication and higher staff turnover (Hofhuis et al., 2016;Shaban, 2016), and minorities can feel excluded (Mor Barak, 2016). Diversity's negative impact is usually explained via the social categorization theory (Turner et al., 1987), which posits that individuals in groups based on diversity (e.g. visible or value) may begin to form "we" against "them" perceptions.
Creating homogeneous working teams could reduce diversity's negative side effects, but that would exclude the many opportunities and benefits of diversity. Thus, scholars continuously search for factors and strategies to reduce diversity-related problems in the workplace without losing the potential benefits (Hofhuis et al., 2016). Developing policies and practices that engender an inclusive climate (Mor Barak, 2016) where individual dissimilarities are accepted and valued seems to be a good solution.
However, the processes through which workplace diversity enhances outcomes remain largely unknown (Dwertmann et al., 2016;Hofhuis et al., 2016). Similarly, it is unclear how much individual employees, leaders or organizations contribute to identification with coworkers and to forming a positive climate for diversity. Moreover, it is unknown whether different dimensions of dissimilarity operate independently or if they buffer or reinforce one another (cf., Sahin et al., 2019), even though the influence of specific dimensions on certain outcomes has been widely researched.
Thus, we aim to examine a negative relationship between perceived visible dissimilarity and employee group identification via the mechanism of value dissimilarity and the moderating role of a leadership-promoted diversity climate in these mediated relationships. This will contribute to leadership and diversity literature. First, we explore how individuals perceive visible and value dissimilarity (which influence a team's initial impression) and how leaders' promotion of diversity impacts groups' team identification. We provide an in-depth explanation of how team members build their own group identification and improve their understanding of one another in diverse working environments. As such, we answer the recent call from Johnson and Avolio (2018) for detailed research into how to enhance group identification.
Second, we highlight leaders' importance in shaping dissimilarity perceptions to capitalize on group identification. Such a view challenged traditional management scholarship proposing a prevalent role of individuals and organizations in shaping the identification perceptions of heterogeneous group members (cf., Bartels et al., 2006). Our view instead emphasizes immediate leaders' role in this process. Third, by distinguishing between visible and value dissimilarities, we contribute to the diversity literature by examining various dimensions of dissimilarity, their interrelationship and their association with group identification based on social categorization. In doing so, we complement existing research on the relationship between dissimilarity and group identification (cf., Hobman et al., 2004).
Our research model appears in Figure 1, followed by the hypotheses tested in our paper. We use multiple methods and locations to test our hypotheses and report data from a field study of working professionals (n 5 584) from five cultural contexts.

Theoretical background and hypotheses development 2.1 Visible dissimilarity and group identification
Regarding diversity literature, we distinguish between surface-level diversity (relatively visible attributes such as age, gender and race) and deep-level diversity (less visible or underlying attributes such as beliefs and values; Guillaume et al., 2014;Mor Barak, 2016;Sahin et al., 2019). Both diversities can be explored as perceived dissimilarity/diversity or as actual dissimilarity/diversity. Actual dissimilarity/diversity is usually measured via Blau (1977) diversity index (using within-group standard deviations to reflect, e.g. diversity in team members' ages), and perceived age dissimilarity/diversity is measured via how the group members perceive dissimilarity or difference in their work group (e.g. Harrison et al., 2002). According to Harrison and Klein (2007, p. 1,216), however, "perceived diversity within a LODJ unit may have unique and more proximal explanatory power than actual diversity." Thus, we will explore perceived visible (as part of surface-level diversity) and perceived value dissimilarities (as part of the deep-level diversity).
We define perceived visible dissimilarity as part of surface-level diversity that individual perceived based on differences in age, gender, ethnicity and more. Perceived value dissimilarity is part of deep-level diversity because it is perceived via differences in work ethic, work values, motivations, personal interests and attitudes (Hobman et al., 2004). Our research examines specific dissimilarity categories, namely visible and value dissimilarity, for two reasons. The first reason lies in previous research findings. Researchers (e.g. Guillaume et al., 2014;Pelled et al., 1999;Shore et al., 2011;Sahin et al., 2019) have shown that different dimensions of dissimilarity differently influence the outcome variables, and/or their effects depend on different moderating factors. The second reason is that visible dissimilarities seem readily apparent, and (undesirable) unique characteristics cannot be hidden. Value dissimilarity, on the other hand, may be invisible to others. Thus, we delineate between visible and value dissimilarity.
Over the past few decades, studies on the relationship between team performance and diversity have contradictory and mixed results (Stahl et al., 2010). Diversity can lead to positive outcomes and enhance business performance (such as creativity and innovation), but it also associates with negative outcomes via less fluent social interactions and reduced team cohesion due to the categorization process in a diverse working environment (Van Knippenberg et al., 2004;Bogilovi c et al., 2017). In accordance with self-categorization theory as part of social identity theory (Tajfel, 1978;Turner et al., 1987;Hogg and Terry, 2000), employees tend to self-categorize based on differences (visible or value-based) in a certain category while other people are automatically categorized as outsiders. In addition, research on the similarity-attraction paradigm (Byrne, 1971) reveals that "interpersonal similarity facilitates communication, improves trust and enhances reciprocal relations" (Mor Barak et al., 1998, p. 88  Research findings (e.g. Chattopadhyay et al., 2004) confirm that the extent of perceived dissimilarity with colleagues influences peoples' identification with their workgroup and, consequently, work-related outcomes. Our conceptualization complements research showing that employees who feel an emotional commitment to their group, feel they belong in a group and identify with their working team will more likely show positive attitudes and behaviors toward the group than will employees who do not feel the same toward their working team (see Salazar et al., 2017;Tsui et al., 1992).
The self-categorization theory views demographic characteristics as salient bases for categorization. Visible demographic variables cannot be hidden easily (e.g. age, gender and ethnic background), so they likely trigger the social categorization process (Pelled et al., 1999;Tsui et al., 1992). Specifically, individuals who are visibly dissimilar in a work team may be classified as outgroup members (visibly nonprototypical). Moreover, individuals are motivated to categorize themselves as group members via visible similarity because this reduces uncertainty about their sense of self (Hogg and Terry, 2000) and because it helps their feelings of competence (self-efficacy; Ashforth, 2001;Brickson, 2013). To feel safe and selfefficient, individuals will likely form multiple identity subgroups (see Carton and Cummings, 2012) based on visible similarities (e.g. man vs woman or young vs old). In addition, people quickly determine which subgroup they belong to in order to simplify their social world (Hoog and Terry, 2000). Researchers (Garcia, 2017;Johnson and Avolio, 2018) confirm that employees identify with coworkers based on perceived similarity to one another. As such, we predict that visible dissimilarity may stimulate individuals to self-segregate into minority subgroups. Thus, visibly dissimilar employees may have difficulty identifying with coworkers. As such, we propose the following: H1. Visible dissimilarity is negatively related to group identification.

The role of value dissimilarity and a climate of leadership-promoted diversity
In addition to individual characteristics that visibly distinguish individuals, values about one's self and about others have deeper and less visible roots that underlie assumptions about colleagues. This assertion justifies focusing on visible dissimilarity and focusing on value dissimilarity as a mechanism through which, under specific conditions, visible dissimilarity leads to group identification.
Indeed, visible and value dissimilarities are crucially connected. People can identify and interpret value dissimilarity by observing behaviors (particularly dissimilar behaviors). For example, achievements that are highly rewarded usually reflect a value, as do clues like prevalent status symbols, such as luxury goods, famous clothing brands, the promotion of assertive and ambitious behavior and more. Thus, individual values (despite being sorted into deep-level dissimilarities) can often be identified or assumed during members' interactions with one another over time (see Harrison et al., 2002). Based on clues expressed via behavior patterns, verbal and nonverbal communication and exchanges of personal information, individuals identify their values as similar or dissimilar to those of colleagues. Thus, Guillaume et al. (2012) unsurprisingly argue that group members who work together frequently have attributes that are not obviously surface-level, deep-level or both. In addition, previous findings reveal that visible and salient dissimilarities make individuals likely to use social distinctions (whether relevant or irrelevant to the task) to predict who shares deep-level task perspectives (see numerous authors in Phillips and Lewin Loyd, 2006).
As such, we take a step further and explore the impact of distinct dissimilarity categories because "it is unknown whether different dimensions of dissimilarity operate independently, buffer, or reinforce one another" (Sahin et al., 2019, p. 2). In line with Sanchez-Hucles and Davis's (2010) view on intersectionality (multiple identities in a working environment), we propose that value dissimilarly can play a crucial part in the relationship between visible LODJ dissimilarity and group identification. Value diversity, defined as deep-level diversity due to attributes such as beliefs and values (Guillaume et al., 2014), expands the social categorization process and creates multiple subgroups when visible dissimilarity emerges.
We predict thatto reduce uncertainty about their sense of self (Hogg and Terry, 2000)individuals will create subgroups based on visible dissimilarities and based on value dissimilarities. Individuals will most likely start the social categorization of others based on surface-level attributes (visible dissimilarities) and continue to categorize in-group and outgroup members based on value dissimilarity, using different categories until it no longer helps make sense of their social context (Hogg and Reid, 2006). For example, if the categorization's initial fit is poor (e.g. differences do not correspond to people's gender, and people do not behave in racially stereotypical ways), individuals will further categorize people based on other accessible categorizations (e.g. values or knowledge) until they reach an optimal level of fit with in-group members.
According to Harrison and Klein (2007), common values between individuals trigger identity-based subgroups because people feel their strongest social attachment to people with the same values. We propose that initial categorization based on visible dissimilarity might trigger further categorization based on value dissimilarity, which is also due to the depersonalization process. When group members categorize themselves and others via a higher-order identity, they will more likely depersonalize out-group members rather than perceive and value out-group members' unique individuality (Hogg et al., 1995). Sahin et al.
(2019, p. 2) explain that "the degree to which people perceive themselves to be surface-level and/or deep-level dissimilar to others can be indicated by multiple attributes they have as well as the intersection of these attributes." Therefore, we predict that an individual who already categorizes others based on visible differences will categorize and create social identity subgroups based on value diversity and on the feeling of separation.
Most situations include perceptions about visible and value attributes overlapping (e.g. see Phillips and Lewin Loyd, 2006). The basic connection between visible and value dissimilarities is intuitive and based on prior evidence. When visible dissimilarities are salient, however, individuals will likely use these social distinctions (whether relevant or irrelevant to the task) to predict who shares deep-level task perspectives with whom (e.g. Allen and Wilder, 1979;Chen and Kenrick, 2002;Diehl, 1988;Harrison et al., 2002;Holtz and Miller, 1985;Phillips, 2003;Phillips and Lewin Loyd, 2006). Sahin et al. (2019) warn that this prediction about deep-level perspectives is not necessarily true because employees who look similar may hold different values, and vice versa. Being different (unique) does not necessarily correlate with the negative experience of being excluded. When a sense of belonging is high and an employee's unique characteristics are valued within a group, the employee will more likely be treated as a part of the workgroup (Shore et al., 2011) and will identify with his/her initial work group. A climate that values diversity in a working environment enhances workgroup involvement (Hobman et al., 2004) and team identification (Luijters et al., 2008) and positively relates to feelings of inclusion among all employees, regardless of their perceived (dis)similarity (Sahin et al., 2019).
Studies (e.g. Martins, 2015) note that the backbone of any activity (including promoting climates when managing diversity) is the leader's behavior and ability to achieve operational goals. We emphasize that leaders have a crucial role in minimizing subgroups based on visual and value diversity and in enhancing group identification. Importantly, we define a leader as an individual with supervisory duties in diversity management who supports the operational implementation of diversity practices. The extent to which employees view an organization as utilizing fair personnel practices and socially integrating all employees is due to the leader's behavior (McKay and Avery, 2015) and the social relationships between the leader and followers. Leaders who engage in social relationships with their followers can provide followers with a "clear set of values [and] a means of expressing these vales within the framework of collective action" (Howell and Shamir, 2005, p. 98). As such, we argue that leaders can promote a climate of diversity by minimizing the negative side of social categorization (e.g. multiple subgroups based on different dissimilarities) and encouraging individuals to value each other's unique perspectives (Bass, 1985).
Leaders can help promote a climate of diversity by creating a common vision, serving as a role model, paying attention to followers' needs and appreciating individuals' initiatives and viewpoints (Bass, 1985;Shin et al., 2012), thus affecting the relationship between dissimilarity (visible and values) and group identification. For example, leaders can minimize negative social categorization by redirecting individuals' attention to a common vision and encouraging group members to transcend group differences ) and facilitate identification with the group. Also, leaders who promote a climate of diversity by appreciating individual points of view can decrease team members' negative reactions and behaviors toward dissimilar (i.e. visible and values) others and increase members' identification (i.e. we work together as one team; Van Knippenberg, 1999). Leaders who promote positive intergroup contact and advocate for diversity might effectively manage workplace diversity (Van Knippenberg et al., 2013). Thus, leaders who promote a climate of diversity increase the chances that individuals will stop creating multiple subgroups and will gain perspective from diverse colleagues (visible and value), which will increase group identification.
In addition, leaders who act as role models to promote diversity at work help to shape an atmosphere of cohesion and acceptance of diverse colleagues (Strauss et al., 2003). Leaders are the most influential figures in shaping workplace perceptions (Boekhorst, 2015), including those related to diversity (Nishii and Mayer, 2009), so we expect leaders to bring together visible-and value-dissimilar colleagues together and to stimulate group identification. We thus propose the moderated-mediation hypothesis: H2. A leadership-promoted diversity climate moderates the relationship between visible dissimilarity and group identification, which is mediated by value dissimilarity.

Sample and procedures
We collected cross-national and transnational culturally and cognitively diverse samples, including the data of 584 employees from five transnational contexts. To capture the most diverse sample of those who deal with multicultural interactions and collaborations daily, we included 1) two large firms in Italy (an engineering technology supplier and a global player in the food industry); 2) three organizations in India (the national postal service and vocational and public services); 3) several organizations in Slovenia (the NGO that unites and represents organizations that perform the activities of youth centers in Slovenia; the Public Agency of the Republic of Slovenia for the Promotion of Entrepreneurship, Internationalization, Foreign Investment and Technology; the Slovenian part of a multinational cigarette and tobacco manufacturing company; and a distributor for various brands in the hair cosmetics industry); 4) University Clinic Centre in the Republic of North Macedonia; and 5) a set of multinational SMEs. From October 2018 to May 2019, data were collected via web questionnaires that defined diversity and explained this study's aim to improve our understanding of how diversity (e.g. age, gender, cultural and national) in a working environment affects employees' engagement in the organization. The study focused on culturally diverse and cross-cultural work settings, so it was administered in English. The survey informed employees that participation was anonymous, confidential and voluntary. To reach employees of the firms that agreed to participate, we followed prior research and used a snowballing approach (e.g. Halbesleben LODJ and Bowler, 2007;Priesemuth et al., 2014;Tepper, 1995). We invited some participants to recommend contacts within the organizations who would be interested in the project and to share our survey with them. As such, 584 employed professionals completed the online survey; 59.60% of respondents were female and about 40.40% male, with an average age of 28.88 (SD 5 11.57). The employees in our sample were from various cultural backgrounds with at least ten different national origins (Slovenia 5 19.0%, India 5 10.0%, Portugal 5 5.9%, Italy 5 5.7%, Croatia 5 5.7%, Macedonia 5 3.4%, Hong Kong 5 2.9%, Spain 5 2.7%, France 5 2.1% and Germany 5 2.0%). The majority of participants had a bachelor's degree (46.20%), 44.20% had managerial duties, and they had about 6.37 years of work experience on average (SD 5 7.14).

Measures
We used five-point Likert-type scales ranging from 1 ("strongly disagree") to 5 ("strongly agree") to measure our constructs, unless otherwise noted. All items were self-reported by the participants.
Employees' feelings and experiences about (dis)similarity in the workplace are likely more meaningful than are objective measures of dissimilarity. Following previous studies indicating that perceived dissimilarity has stronger effects than actual (objective) dissimilarity (Hobman et al., 2004;Shemla et al., 2016;Strauss et al., 2001), our current research utilizes a subjective measure of dissimilarity.
Perceived visual dissimilarity was assessed via a two-item scale from Hobman et al. (2003). Participants were asked to assess their identification with their team members at work based on informational, visible and value dissimilarity. The items were "In terms of visible characteristics (e.g., age, gender, ethnicity) I think I am different from other group members" and "I feel I am visibly dissimilar to other group members." The scale's reliability was adequate (α 5 0.72).
Perceived value dissimilarity was assessed with a two-item scale of Hobman et al. (2003). The items were "I feel my work values and/or motivations are dissimilar to other group members" and "In terms of principles that guide my work (e.g. detail-oriented, reward-driven), I think I am different from other group members." The scale's reliability was adequate (α 5 0.70).
Group identification was measured with a five-item scale adapted from previous studies (e.g. Hogg and Hains, 1996). Participants were asked to assess their identification with their team members at work. Sample items included "I identify with my team at work very much" and "I feel strong that I fit into my team at work." The scale's reliability was α 5 0.81.
Leadership-promoted diversity climate was measured with a six-item scale adapted from Mor Barak (2016). Participants were asked to assess their leader and how they experience his promotion of diversity at work while focusing on perceived fairness and nondiscrimination based on different diversities. Namely, Brimhall et al. (2014) claim that managing diversity via fair policies, practices and procedures may yield positive perceptions of the diversity climate, which may improve perceptions of inclusion. It can even represent different (responding on different needs of diverse employees) but equitable treatment (equally fair). Thus, the items include "My supervisor has a track record of hiring and promoting employees objectively, regardless of any differences (e.g., employees' race, sex, sexual orientation, religion, nationality or age)" and "My supervisor gives feedback and evaluates employees fairly, regardless of employees' race, gender, sexual orientation, religion, age or social background." The scale's reliability was α 5 0.85.

Controls.
We controlled for which data collection context (out of five) a respondent belonged to and for the third dimension of dissimilarityinformation dissimilarity, measured with two items from Hobman et al. (2003). We also controlled for individual attitude toward Leadershippromoted diversity climate diversity (measured with a six-item scale adapted from Mor Barak et al., 1998) and for organization-promoted diversity climate (assessed with a four-item scale, adapted from Mor Barak, 2016) in the moderated-mediation models looking into the moderation of leadershippromoted diversity climate. For measuring individual attitudes toward diversity, two aspects were considered: the diversity value factor (including three items, such as "Diversity issues keep some work teams here from performing to their maximum effectiveness") and personal comfort with diversity (including three items, such as "I feel at ease with people from backgrounds different from my own"). To measure an organization-promoted diversity climate, we considered organizational fairness (including two items, such as "I feel that I have been treated differently here because of my race, gender, sexual orientation, religion, or age") and organizational inclusion aspects (with two items, such as "The company spends enough money and time on diversity awareness and related training"). Table 1 presents means, standard deviations and correlations for the key study variables. A confirmatory factor analysis with maximum likelihood estimation procedures using AMOS version 21 showed that the expected four-factor solution with focal constructs in our research model (visible dissimilarity, value dissimilarity, group identification and leadershippromoted diversity climate) displayed a good fit with the data (Chi-square [71] 5 227.68, CFI 5 0.94, RMSEA 5 0.06). Because our research is cross-sectional (our constructs' data were collected from the same respondents at a single time), we conducted two additional analyses to alleviate potential issues related to common method biases (Podsakoff et al., 2012). First, we conducted Harman's single factor testa principal component analysis of all items of our constructs that extracts only one factor and uses no rotation method. No dominant factor emerged; the overall variance explained by the extracted factor fell below the threshold of 50% (specifically at 23.19%), thus providing no evidence that common method variance might be an issue.

Results
Second, we applied Lindell and Whitney's (2001) marker variable test and used a theoretically unrelated variable (i.e. marker variable) to adjust the correlations among the model's principal constructs. The marker variable's high correlation with any of the study's other principal constructs could indicate a common method bias. For robustness, we separately repeated the marker variable test with two variables not included in the model (employee education and whether they hold any managerial duties) with little or no theoretical basis to expect a relationship with the study's principal constructs. The average correlation between the study's principal constructs for employee education (r 5 0.06) and managerial duties (r 5 0.03) was low and nonsignificant, providing no evidence of a common method bias.  (Podsakoff et al., 2012), we also conducted a common-latent factor test to account for the variance common to all items in the model. This test includes one latent common-method factor and allows all items to load on their theoretical constructs as well as on this latent factor. We estimated the common variance as the square of the common factor of each path before standardization. This test suggested a common variance of 15%, much below the commonly applied heuristic of 50%. Taken together, we conclude that common method bias was likely not important for our data.
To test our hypotheses, we first examined the direct relationship between visible dissimilarity and group identification (H1) via an ordinary least squares linear regression and then examined the moderated-mediation model, which included a mediator of value dissimilarity and a moderator of leadership-promoted diversity climate (H2). Our respondents came from various contexts, causing inherent nestedness in our data. However, a multigroup analysis revealed that 96% of our independent variable's (group identification) variance can be explained at the individual level. Thus, we conducted the analysis at the individual level and controlled for the effects of the data collection contexts. We followed standard procedures to examine this moderated-mediation model using a bootstrap approach (Preacher and Hayes, 2004). Drawing 5,000 random samples via replacement from the full sample, we constructed 95% bias-corrected confidence intervals for the hypothesized indirect mediating effects at various moderator values.
The linear regression (Column 2, Table 2) indicated that visible dissimilarity did negatively relate to group identification (β 5 À0.10, se 5 0.03, p < 0.01), supporting Hypothesis 1. Next, we expected the moderated mediation Hypothesis 2 to show that a leadership-promoted diversity climate moderates the relationship between visible dissimilarity and group identification, which is mediated by value dissimilarity. Column 3 in Table 2 presents the moderated mediation results with the conditional indirect path coefficient and 95% confidence intervals using the PROCESS macro Model 14.
The results show that a mediated relationship was supported at low and high levels of leadership-promoted diversity climate; for high levels (þ1 SD above the mean), the conditional effect size was À0.03 (unstandardized confidence intervals excluded zero; the lower bound 5 À0.08, and the upper bound 5 À0.01) with a p-value less than 0.01. Moderate levels of leadership-promoted diversity climates showed an insignificant conditional effect because the confidence interval included zero. For low levels (À1 SD below the mean), the conditional effect size was 0.06 (unstandardized confidence intervals excluded zero; the lower bound 5 0.01, and the upper bound 5 0.10) with a p-value less than 0.01 [1]. The moderated mediation index was significant (index 5 À0.06, se 5 0.02, lower level confidence interval 5 À0.10, upper level confidence interval 5 À0.02).
We further examined the interaction graph for a leadership-promoted diversity climate's moderating role in the relationship between value dissimilarity and group identification, as shown in Figure 2. As expected, leadership-promoted diversity climate lifted the line of the relationship between value dissimilarity and group identification. Simple slope analyses showed moderation for both low (gradient 5 0.50, p < 0.01) and high (gradient 5 0.42, p < 0.01) levels of leadership-promoted diversity climates. Taken together, Hypothesis 2 was supported.

Discussion
By nature, humans tend toward categorization and the establishment of an in-group (we) and out-group (they) to feel safe. The support for Hypothesis 1 also confirms a major stream of research showing negative links between dissimilarity, group functioning and, consequently, work outcomes (cf., Guillaume et al., 2012 for meta-analytic evidence). Hypothesis 2 was also supported, which aligns with other researchers (e.g. Hobman et al., 2004) who found that Leadershippromoted diversity climate interactions between diverse individuals causing either benefits or deficits for the organization depend largely on other factors: moderators or boundary conditions. Our research reveals that a leadership-promoted diversity climate plays a significant moderating role in the relationship between visible dissimilarity and group identification. More precisely, our results show that a leadership-promoted climate of diversity interacts High leadershippromoted diversity climate Table 2.
Results of the moderated-mediation analyses predicting group identification using the PROCESS macro Figure 2. The moderating role of leadership-promoted diversity climate in the relationship between value dissimilarity and group identification LODJ with the perceptions of visible dissimilarity to yield different levels of group identification. The highest group identification emerges when the leadership-promoted diversity climate and value dissimilarity are both high. This indicates that values can help with group identification. Second, it indicates that leadership represents a crucial contingency in overcoming intergroup tensions (Berg and Smith, 1990) due to diversity and in creating higher levels of group identification. A close examination of the reasons behind these results reveals a reasonable explanation.

Theoretical contributions
The research findings significantly contribute to leadership and diversity theory. First, it distinguishes between visible and value dissimilarity and contributes to diversity literature by examining various dimensions of dissimilarity and their interrelationship and their association with group identification based on social categorization. In doing so, we complement existing research on the relationship between dissimilarity and group identification (cf., Hobman et al., 2004). The majority of recent studies examined the relationships between a single visible/value characteristic and individuals' perception of belonging (e.g. gender dissimilarity and felt inclusion, Jansen et al., 2017). However, few studies (e.g. Sahin et al., 2019) focus on the interactive effects of perceptions of visible and value dissimilarity on perceived inclusion, which influences work outcomes. Thus, our current research contributes to the organizational diversity literature by examining the interrelationships of visible and value dissimilarity perceptions and their effect on group identification.
Second, we add to the diversity literature by exploring how individuals perceive visible and value dissimilarity, which influence a team's initial impressions, and how contextual factors (e.g. leadership promotion of diversity) interact with these perceptions to impact group-team identification. We thoroughly explain how team members build their own group identification and a better understanding of one another. As such, we answer the recent call from Johnson and Avolio (2018) for detailed research into enhancing group identification. Moreover, this finding supplements prior research (e.g. Drach-Zahavy and Trogan, 2013) by providing empirical evidence that diversity climate promoted by leaders strengthens the dissimilarity perceptions in group identification. In our findings, members of units with higher levels of leadership-promoted diversity climate perceived higher group identification.
In addition, our results indicate that visible dissimilarity positively relates to group identification via value dissimilarity. Dissimilar values can be identified by observing individuals' behavior during interactions and clues expressed in behavior patterns, communication and exchanges of personal information. However, value dissimilarity has deeper roots, and individual behavior patterns can be only learnt through extended interaction and information gathering. Thus, it takes time to uncover these deep-level value differences (Guillaume et al., 2012). It will also take more time to identify social identities based on value diversity. This might be why visible dissimilarity relates more positively to group identification via perceptions of value dissimilarity when leadership-promoted diversity climate is high. Our results align with previous findings from empirical studies, research reviews and meta-analyses concluding that diversity is not completely "bad" or "good" (see Joshi and Roh, 2009). Rather, the impact of diversity should be considered in light of the specific outcomein our case, group identification (Joshi and Roh, 2009) and a specific diversity climate.
Third, we add to leadership theory by answering the question, "What or who is the most influential actor in developing the positive/inclusive workplace environment?" Responding to those lacunas in the literature and in practice, we place leadership-promoted diversity climate as a key entity overlapping existing theoretical concepts, and we aim to offer insights into Leadershippromoted diversity climate who can or should use his/her potential to implement diversity management theory in day-today management. Our results clearly indicate that the leader (while promoting diversity climate) plays a crucial role in moderating the negative relation between visible dissimilarity and group identification. Our findings also indicate that a leadership-promoted diversity climate is crucial for group identification and that group identification was the highest in high leadership-promoted diversity climates and high value dissimilarity conditions. This is no surprise because leaders' commitment to a climate that promotes equal opportunities for training, diversity and a diversity-friendly work environment will reduce threats to team identity, thus creating an inclusive team categorization (Thatcher and Jhen, 1998). Our theoretical contribution to existing literature is highlighting the importance of leaders in shaping dissimilarity perceptions to capitalize on group identification, even beyond organizational effects. Namely, such a view challenges traditional management scholarship proposing the prevalent role of individuals and organizations in shaping the identification perceptions of heterogeneous group members (cf., Bartels et al., 2006). Instead, our view emphasizes the immediate role of leaders in this process.
Our results and theoretical conceptualization show that social identity, self-categorization and other theories on one side and diversity management (striving to "annul" categorization processes) on the other should both be considered parts of a complementary (not antagonistic) concept/paradigm. While respecting social identity theory ideas (that assume tensions associated with human needs for validation and similarity to others), leaders should promote a positive/inclusive diversity climate where diversity (uniqueness) and group identification (belonging) would coexist. Moreover, the leader's crucial role is enabling the implementation of theoretical ideas within management strategy and action.

Practical implications
Our research results benefit practitioners in the field, among others, because the results elucidate that diversity management concerns managing diverse (unique) employees, aiming to make them feel identified and committed (sense of belonging)not reducing dissimilarities and enjoying the fruits of homogeneity. Diversity management should always concernas Shore et al. (2011) statethe theme of belonging. It is about identifying with coworkers while working on the uniqueness and compatibility of dissimilarities. The latter seems very important for understanding and improving team effectiveness (Solanas et al., 2013).
A key practical implication of our results is the role of leaders as a key element in the workplace who contribute to group identification within diverse teams. Organizational diversity efforts often manifest in prepackaged programs seeking to instill new attitudes in employees (Beer et al., 1990). A more effective approach would be assessing the organization's needs and employees' current attitudes before training employees for diversity awareness.
Van Knippenberg et al. (2007, p. 207) found out that "work group diversity and group identification are more positively related the more individuals believe in the value of diversity". However, individual attitudes toward a diversity climate can vary. Here, leadership's attitude toward diversity policy implementation in firms is critical in supporting individuals' commitment to diversity.

Limitations and future research suggestions
The first set of limitations concerns the research design and setting. The sample's size and structure could be improved in future research to cover a larger and more diverse set of countries and representative samples of working professionals within them. Importantly, though, obtaining samples of employees who work in diverse conditions with sufficient multicultural and transnational exposure is highly difficult. Our research design was also cross-sectional and self-reported, but all variables are such that self-perceptions make sense LODJ to assess them. Still, reverse causality is possible. Future research could thus explore the reverse formulation of our current hypotheses with group identification as the independent variable and with perceived dissimilarity as the dependent variable. Branscombe et al. (1999) point out that team identification may encourage dissimilar individuals to follow team norms and conventions to gain acceptance in the team, which can mitigate the negative effects of diversity. Longitudinal or experimental studies would definitively establish the causality in the proposed relationships.
The second set of limitations within our research relates to some constructs and their conceptualizations. For example, the relationship between perceived visible dissimilarity and group identification is perhaps not negative, per se (even though numerous studies support this negative relation, as cited within our paper). Future research could explore boundary conditions where dissimilarity could yield higher group identification, such as particular industry settings that highly value diversity or virtual teams operating together from a distance. Future research should also examine specific leadership characteristics, styles and behaviors because they could act as additional boundary conditions for the moderating role of a leadership-promoted diversity climate. Particular leadership traits are important in shaping messages to employees about their working context and are relevant for individual perceptions that develop at work (cf., Cerne et al., 2013;Lumby, 2009). We also did not measure or account for team/group tenure and team size in our research model. Group tenure has been known to impact identification processes because it impacts individuals' ability to determine their teammates' values and whether they are similar to theirs (Barker and Tomkins, 1994;Shemla and Wegge, 2019). Group size also likely affects perceptions of dissimilarity because it may affect communication patterns in groups, which carry the meaning related to (dis)similarity perceptions (Zenger and Lawrence, 1989).
In addition, social identity's complexity highlights how people subjectively combine multiple social identities with room for uniqueness and similarity with other group members. Thus, individuals may be out-group members based on one social identity (e.g. gender) but simultaneously identify with a group on another social identity characteristic (e.g. race). Moreover, we should consider that visible and value dissimilarities do not necessarily overlap or act independently. Sahin et al. (2019, p. 3) note that "employees may both look different than others at work and also hold different values to them, but it is also possible that they look very similar yet hold different values or that they look very different yet hold the same values." Further research would probably further detach the separate and joint effects of visible and value characteristics in relation to desired manifestations of workplace psychological well-being (e.g. feelings of being identified, inclusiveness, belonging, etc.).

Conclusion
In summary, our study mainly contributes by treating dissimilarity as a multicomponent construct, emphasizing that crucial differences appear in different conceptualizations of dissimilarity. For heterogeneity and dissimilarity to yield group identification, the results highlighted leaders' crucial role (above and beyond that of organizations and individuals) in promoting a diversity-embracing climate in work units. The current study confirms that, in practice, improvements could be made in almost every organization. As Patrick and Kumar (2012, p. 1) put it, "Diversity management should be considered as the process aimed to both create and maintain a positive work environment." This process leads to an inclusive workplace and embraces many of the employees' dissimilarities. Until an inclusive and positive work environment that respects unique characteristics and values differences appears on paper (in theory), perceived dissimilarity will negatively impact group cohesion and worsen business results. Lastly, it further behooves us to add to the understanding of various dissimilarities at work and how to embrace and capitalize on them rather than attempt to avoid them.