Introduction

International organizations have long recognized the importance of training managers to cross cultural boundaries. Today international organizations, ranging from large MNCs to small ‘born global’ firms, are increasingly recognizing that multicultural employees can help them operate across countries and across cultures (Hong & Doz, 2013). Multicultural individuals1 are those who have internalized and identify with more than one culture (Brannen & Thomas, 2010; Lücke, Kostova, & Roth, 2014). For example, both migrants and their descendants can internalize and identify with more than one culture. With 244 million international migrants in 2015 alone, this is a fast-growing demographic (United Nations, 2016), and an under-appreciated resource for international organizations. Multicultural individuals are members of more than one cultural group, so they are more likely to see themselves as straddling cultures than crossing them. This unique perspective allows them to bring a different set of capabilities and skills to roles such as expatriates, members of multicultural or global virtual teams, or global leaders (Barner-Rasmussen, Ehrnrooth, Koveshnikov, & Mäkelä, 2014; Lücke et al., 2014; Yagi & Kleinberg, 2011). In all of these roles, they may play a pivotal role in facilitating positive outcomes related to cultural diversity in international organizations, while also helping to mitigate some of its challenges (Fitzsimmons, Miska, & Stahl, 2011; Lücke et al., 2014). However, multiculturals themselves are a diverse group, and understanding how they can best contribute to organizations must be substantiated by clear evidence about the relationship between their multiculturalism and personal, social and task outcomes.

Our approach focuses on leveraging the skills of employees along a spectrum of multicultural to monocultural, rather than starting with the outdated assumption that employees are monocultural. This approach addresses two trends in international business research: the trend to examine more positive outcomes of culture (Stahl & Tung, 2015); and the trend to recognize the prevalence of cultural diversity within countries (Dow, Duypers, & Ertug, 2016), organizations (Kulkarni, 2015), and teams (Stahl, Maznevski, Voigt, & Jonsen, 2010). In this research we emphasize that we must also recognize cultural diversity within individuals. Multiculturals have potential advantages and face challenges that are only beginning to be explored by researchers in cognitive and social psychology (Chen, Benet-Martínez, Wu, Lam, & Bond, 2013; Hong, Morris, Chiu, & Benet-Martínez, 2000; Nguyen & Benet-Martínez, 2007). This psychological research can provide a foundation for expanding our knowledge about multicultural employees and how to manage them.

By focusing on the valuable resources multicultural employees possess and the challenges they face in the global workplace, we pinpoint not only the idea that multicultural individuals can be an asset, but also how they become an asset. We demonstrate that individuals’ cultural identities exist on a continuum called identity plurality, ranging from monocultural to multicultural. We find that this continuum is especially useful for predicting social outcomes such as social capital developed through network relationships, and task outcomes such as intercultural skills. A second dimension, identity integration, refers to the extent to which individuals integrate their cultural identities versus keeping them separate. As such it can only be applied to multicultural individuals (Benet-Martínez & Haritatos, 2005). We find that it is especially useful for predicting personal outcomes such as well-being. The dimensions of identity plurality and identity integration create a map of identity patterns we use to test how personal, social and task outcomes vary among multicultural individuals, and across multicultural and monocultural individuals (Fitzsimmons, 2013). Our intent is to demonstrate how multicultural organizations can leverage both groups of employees to mutual advantage.

Mapping Multicultural Identity Patterns

Starting from a paper titled ‘Multicultural minds’ (Hong et al., Hong 2000), much of the research on multiculturals has taken a cognitive approach that focuses on how multiculturals perceive and interpret information differently from monoculturals when making decisions. For example, one stream of research uses experimental manipulations to examine how and why multicultural individuals switch among cultural frames (Chen et al., 2013; Cheng, Lee, Benet-Martínez, & Huynh, 2014). Along these same lines, Lücke et al. (2014) recently proposed that multicultural individuals develop unique skills and abilities through a social cognitive process called cognitive connectionism. However, multiculturalism is more than a cognitive phenomenon.

Rather than explaining outcomes using cognitive mechanisms alone, we draw on the underlying organization of cultural identities as the basis for explaining relationships between multiculturalism and individual outcomes, ultimately expanding the field’s theoretical arguments to include both cognitive and motivational mechanisms. Cognitive mechanisms focus on the knowledge structures and associated cognitive capabilities multiculturals have and utilize in problem-solving, whereas motivational mechanisms are more goal-oriented and explain how individuals express themselves, regulate their behavior, and manage their relationships with others. We combine these perspectives by testing a theoretical framework based on social identity theory that explicitly defines both cognitive and motivational mechanisms through which multiculturalism affects outcomes (Fitzsimmons, 2013).

When individuals identify with more than one culture they also internalize more than one cultural schema. Cultural schemas are organized knowledge structures, including cultural norms, values and beliefs, that help people interpret and select information associated with a culture (Markus & Kitayama, 1991). Cultural identities are mentally organized along the dimensions of identity plurality and identity integration in order to facilitate sense-making (Ashforth, Harrison, & Corley, 2008), by guiding relative accessibility to these underlying cultural schemas. We provide illustrative quotations for these dimensions from our pilot study, which was designed to elicit descriptions of individuals’ identity patterns.2

Identity plurality refers to the number of primary cultural identities, ranging from one to many. Primary cultural identity refers to deeply held identities based on internalized cultural schemas, not situated identities that can be tried on and discarded (Rousseau, 1998). This framework excludes those who identify with a culture they know little about, because they have no internalized schema related to that culture. Following from the definition of a multicultural individual as someone who both identifies with two or more cultures, and internalizes cultural schemas for all of their cultures (Brannen & Thomas, 2010; Fitzsimmons, 2013), it is not enough for someone to only identify with a culture, without also internalizing cultural norms, values, assumptions and the behaviors associated with that culture (Pekerti & Thomas, 2016). Deeply held identities remain part of the self across situations, although they primarily guide behavior when salient or activated (Ashforth et al., 2008; Markus, 1986). For example, in our pilot study the following 41-year-old male described himself with a series of percentages that correspond to his cultural identity, not bloodlines, indicating a high level of cultural plurality:

Anglo-Canadian 40%, Pakistani 20%, American 20%, French-Canadian 10%, Indonesian 10%.

This high level of identity plurality resulted from his global experience. He speaks four languages and moved to Canada 15 years ago, after having lived in five other countries. In contrast, another individual clearly prioritized one culture over the other, indicating low identity plurality. Although she identified herself as Chinese-Canadian, she reported, “Chinese is most important.”

Identity integration is the extent to which individuals integrate their cultural identities versus keeping them separate (Benet-Martínez & Haritatos, 2005). For example, the following quotation illustrates how a Barbadian-Canadian woman described her own highly integrated identity pattern:

I would describe my own form of biculturalism as integrated. Neither cultures are separate on their own but rather are combined to create a unique culture that draws on my identification with both the Barbadian and Canadian cultures.

Her high level of integration is especially noticeable when contrasted to the following description from a Jewish-Canadian male that exemplifies low identity integration, “I see myself as having two separate cultures that both influence who I am. There is little overlap between how these cultures influence me.” Together, identity integration and plurality dimensions produce a map that can be used to compare different identity patterns, linking multicultural and monocultural individuals within the same framework.

Both identity dimensions are easiest to understand by looking at their polar opposites. These patterns represent ideal types, not categories, because the dimensions are continuous, not categorical. We adapted Figure 1 from Fitzsimmons (2013) to illustrate the continuous nature of identity dimensions. Table 1 illustrates the nature of each ideal type with exemplary quotations from our pilot study. Although the ideal types are useful for explaining the dimension, multicultural individuals could be located anywhere on the map. As illustrated in Figure 1, a range of possible patterns can emerge from these two dimensions, which are not limited to the ideal types shown at the endpoints of each dimension.

Figure 1
figure 1

Model of multicultural identity dimensions, adapted from Fitzsimmons (2013), illustrating continuous nature of the dimensions.

Full size image
Table 1 Exemplary responses to open-ended questions asking for descriptions of participants’ multicultural patterns, by ideal type (Pilot Study)
Full size table

Illustrating the two extremes of identity plurality are prioritizing multiculturals, who have one primary cultural identity and a second cultural identity that is less important, and aggregating multiculturals, who strongly identify with three or more cultural identities. Prioritized patterns do not refer to relative weightings or objective positioning of cultures. Instead, they account for varying degrees of identification with cultures (Roccas, Sagiv, Schwartz, Halevy, & Eidelson, 2008). To count cultures accurately, we must take this variation into account by differentiating between those who fully identify with two cultures and those who identify fully with one and partially with a second. Aggregating multiculturals may be more likely to encompass their cultural identities within a broader identification that extends beyond the boundaries of their own cultural groups. The aggregating ideal type reduces differentiation between in-group and out-group members, because the in-group in this case is more heterogeneous than it is in other patterns (Park & Rothbart, 1982). Compared to aggregating, the prioritizing ideal type permits a simplified identity structure, in which most phenomena are filtered through the prioritized cultural schema, with accents of the second or third cultures.

Illustrating the end points of identity integration, compartmentalizing multiculturals see their identities as separate and identify with one or the other, depending on the context, while hybridizing multiculturals identify primarily with the intersection of the two cultures, more than with either culture individually. For example, hybridizing Chinese-Canadians will identify with other Chinese-Canadians as their in-group (more than with Canadians or Chinese). The identity integration dimension has been shown to significantly influence multiculturals’ frame switching behavior, creativity and other outcomes (Cheng, Lee, & Benet-Martínez, 2006; Cheng, Sanders, Sanchez-Burks, Molina, Lee, Darling, & Zhao, 2008).

While it may be the case that individuals who accumulate more identities are more likely to integrate them, we theorize on the basis of distinct outcomes that result from each dimension, independent of the other. In some cases, we develop hypotheses from both dimensions for the same outcome. This indicates that an outcome may be influenced by both the number of cultural identities (identity plurality) and their integration (identity integration).

How Multicultural Identity Patterns Influence Outcomes

The map of multicultural identity patterns represents different ways to mentally organize cultural identities and their associated cultural schemas. Using mechanisms drawn from social identity theory, the map can be used as a basis for predictions about individual-level outcomes important to organizations. According to social identity theory, people sort others and themselves into social groups in order to reduce uncertainty, and they positively differentiate their own in-groups from out-groups in order to enhance self-esteem (Tajfel & Turner, 1986). In the following sections, we draw on these two mechanisms as a theoretical basis for developing propositions about the relationship of identity integration and identity plurality to outcomes. While all of the outcomes we examine have multiple antecedents, including context and other individual characteristics (Johns, 2006), we isolate cultural identity to clarify its unique influence.

We theorize relationships in three categories of personal, social and task-related outcomes based on a long history of this three-part categorization in the expatriation and adjustment literature (Bhaskar-Shrinivas, Harrison, Shaffer, & Luk, 2005; Black, Mendenhall, & Oddou, 1991). Personal outcomes refer to variations in personal well-being; social outcomes are those related to interpersonal relationships; and task outcomes refer to those related to the work itself. We develop and test hypotheses for exemplary outcomes in each category. We focused on common outcomes in international management research that shared a theoretical logic. This approach allowed us to develop a more coherent picture from the disparate strains of research on multicultural individuals. For example, a common mechanism is the internal consistency of multicultural identity patterns, which influences personal outcomes, such as identity uncertainty and feeling overburdened from cultural translation work.

Personal Outcomes: Identity Uncertainty and Feeling Overburdened

We argue that identity patterns vary in their internal consistency, and that these variations result in different levels of effectiveness at reducing uncertainty. Roccas and Brewer (2002) explained that identity patterns with a single in-group are more internally consistent than patterns with multiple in-groups, because the latter has many cultural schemas, involving many sets of values, norms, assumptions and expected behaviors. Specifically, identity patterns with a primary in-group (i.e., prioritizing) are more internally consistent than patterns with multiple in-groups (i.e., aggregating), and integrated patterns (i.e., hybridizing) are more internally consistent than separated patterns (i.e., compartmentalizing). Thus the most internally consistent pattern is monocultural, followed by the top left of the diagram in Figure 1, in which one cultural identity is prioritized over the other, and both are integrated. The least internally consistent patterns fall in the bottom right corner of the diagram, in which multiple cultural identities are maintained separately. We expected consistent identity patterns to reduce uncertainty more effectively than inconsistent patterns because of the potential for inconsistent patterns to provide conflicting guidance for behavior. These conflicting cultural schemas can result in lower levels of personal well-being, including higher identity uncertainty and feeling overburdened with cultural translation work.

Identity uncertainty refers to a lack of clarity about one’s identity (Goldberg, Riordan, & Schaffer, 2010). It tends to be lower for individuals who identify with clearly defined and easily identifiable groups, especially when those groups remain consistent over time and across contexts (Goldberg et al., 2010). This implies that lower levels of identity plurality will reduce identity uncertainty. Among long-term immigrants, integrated multiculturals have been found to exhibit better psychological adjustment than those with separated patterns, based on the argument that separated identity patterns lead to ambivalence and confusion (Chen, Benet-Martínez, & Bond, 2008). Thus individuals with single, integrated identity patterns are expected to exhibit the lowest levels of identity uncertainty, while those with multiple, separated identities will exhibit the highest levels.

In addition to evidence linking multiple identities to identity uncertainty, multiple identities have been shown to cause stress through role overload (Thoits, 1983). Specific to the cultural identity domain, and using terms that emerged from our pilot study, an example was feeling overburdened with cultural translation work. This refers to the stress that resulted from an overwhelming amount of time spent on cultural translation activities, such as helping colleagues understand and interpret behavior across cultural boundaries. Too much time spent on activities related to helping colleagues work across cultures can undermine time spent other activities, ultimately causing overload and distress (Barnett & Hyde, 2001). It may be related to the internal consistency of identity patterns, as cultural translation work also involves switching among cultural frames. Individuals with many conflicting identities have been found to experience lower psychological well-being relative to those with fewer, or more harmonious, identities, because the former group felt as if they were behaving in ways that were inconsistent with the activities they ought to be undertaking (Brook, Garcia, & Fleming, 2008). Importantly, this finding was only related to highly salient identities, such as culture, and not with less salient identities, such as being a member of a work committee (Brook et al., 2008). These findings suggest that individuals may be especially susceptible to feeling overwhelmed by cultural translation work when they have internalized many cultural identities, and also when those identities are separated, because the process of doing cultural translation work likely involves more taxing identity switching for those whose identity patterns are more complex, relative to those with simpler identity patterns.

H1a:

Identity plurality will be positively related to identity uncertainty and feeling overburdened with cultural translation work.

H1b:

Identity integration will be negatively related to identity uncertainty and feeling overburdened with cultural translation work.

Social Outcomes: In-group Cultural Diversity

Social outcomes are those related to social capital – the goodwill that can be accessed through the structure and content of social relationships (Adler & Kwon, 2002; Bordieu, 1986; Nahapiet & Ghoshal, 1998). Employees who are better connected across multiple groups are valuable for their ability to span boundaries and facilitate interactions, even across unplanned cultural faultlines (Lau & Murnighan, 2005), making social network ties a valuable outcome (Barner-Rasmussen et al., 2014). When individuals seek to increase self-esteem by positively differentiating their in-groups from referent out-groups they develop more social ties with other in-group members than with out-group members (Ashforth et al., 2008). This effect occurs regardless of the number of cultural identities, and explains why individuals develop more friendships within cultural in-groups than outside of them. As a result, individuals with low identity plurality are likely to develop lower levels of social capital than individuals with high plurality, whose friendships are likely to span a wider range of cultures.

Beyond this relatively straightforward effect of increasing social capital when in-groups contain more cultures, we also expected that individuals who have internalized many cultural identities are more likely to build social connections outside their cultural in-groups. As the number of internalized cultural identities increases, social categorization based on cultural membership becomes more complex and less obvious, making cultural groups less useful as a criterion for distinguishing in-group from out-group members. That is, individuals who have internalized many cultural identities are expected to have more difficulty differentiating between cultural in-group and out-group members because the boundaries are less clearly defined, relative to individuals who have internalized fewer cultural identities (Crisp, Hewstone, & Rubin, 2001). Therefore instead of cultural backgrounds, people with multiple cultural identities may be more likely to categorize people based on alternative domains with clearer group boundaries, such as organizational or professional groups. In contrast, people with fewer cultural identities see clearer boundaries containing their cultural in-group. As a result, they may be more likely to categorize out-groups based on culture, and thus develop fewer social ties with members of other cultures.

Two studies have demonstrated similar effects. A study of intergroup bias and multiple identities found that when individuals were presented with more than two identities as a basis for judgment, they exhibited less intergroup bias than when they were presented with only one (Crisp et al., 2001). This decrease in bias was explained through a related decrease in the degree to which participants saw the situation as an us-versus-them situation, and an increase in the degree to which participants saw the targets as individuals over group members (Crisp et al., 2001). Another study found that groups with more meanings, as measured by the number of distinct group names, have less intergroup hostility relative to groups with fewer meanings (Mullen, Calogero, & Leader, 2007). This finding indicates that when in-groups are more heterogeneous, individuals are less likely to exhibit bias against out-group members, suppressing the preference for in-group over out-group members. Together, these studies indicate that as cultural group membership becomes more complex, clarity with which individuals delineate culture-based referent out-group decreases. We therefore propose that identity plurality will predict the cultural diversity of in-groups because individuals’ patterns of relationships will reflect their identity patterns. We expected that individuals with high identity plurality would have the widest variety of cultures in their in-groups, even beyond their own cultures, thereby earning them higher levels of social capital, and supporting the following hypothesis:

H2:

Identity plurality will be positively related to the cultural diversity of in-groups.

We did not expect to find that identity integration would be related to the cultural diversity of in-groups because the degree of integration does not influence whether or not individuals belong to their cultural groups. Thus the mechanism of increasing self-esteem by positively differentiating cultural in-groups from referent cultural out-groups should be the same for individuals with highly integrated or separated cultural identities. We therefore expected no relationship between identity integration and the cultural diversity of in-groups. We included our proposition that identity integration would not be related to the cultural diversity of in-groups in the summary table for consistency across the framework. However, as there is insufficient evidence to suggest a directional effect we did not test the null hypothesis.

Task Outcomes: Cultural Metacognition, Adaptability, Language Interpretation and Job Performance

Beyond the reduction in personal well-being that can result from mentally organizing inconsistent cultural identities, and the increase of social capital resources that multicultural employees can access, multicultural employees may also draw on their identity patterns to perform work-related tasks, such as solving complex problems, leading multicultural teams, and negotiating across cultures (Fitzsimmons et al., 2011; Tadmor, Galinsky, & Maddux, 2012). Indeed, two meta-analyses found statistically significant relationships between identifying with at least two cultures (versus identifying with one culture) and behavioral competence, including academic achievement and career success (Berry, Phinney, Sam, & Vedder, 2006; Nguyen & Benet-Martínez, 2013). Although the empirical evidence indicated a relationship between cultural identities and behavioral competence, theoretical mechanisms to explain this relationship have remained elusive (Nguyen & Benet-Martínez, 2013).

We expected that identity patterns would influence task outcomes through the mechanism of developing intercultural skills. Drawing on the same argument we described for personal outcomes, identity patterns that are internally inconsistent do not reduce uncertainty as effectively as internally consistent patterns. Inconsistent patterns allow for more complex cognitive schemas, containing multiple sets of values, assumptions, norms and behavioral repertoires that can be accessible simultaneously, and are sometimes in conflict. Thus task outcomes represent the flip side of personal outcomes, in that inconsistent patterns produce lower levels of personal well-being, but also produce higher levels of intercultural skills, which draw on higher levels of cognitive complexity. Overall, we expected identity patterns to influence which set of intercultural skills individuals develop, and in turn, expected these skills to influence their success at performing intercultural tasks, such as solving complex global problems by drawing on ideas from multiple sources.

More specifically, we hypothesized relationships between identity patterns and four task-related outcomes commonly examined in international management research. These are cultural metacognition, adaptability, language interpretation, and job performance. Cultural metacognition refers to the ability to attend to cultural cues, and control one’s behavior appropriately in response to those cues (Thomas et al., 2012). Thus it involves both cognitive and behavioral components. Tadmor and Tetlock (2006) found that multicultural individuals who identified strongly with two cultures exhibited more cognitive (integrative) complexity than those who identified more strongly with one culture over the other. They attributed this difference to the increased dissonance of having two equal cultures. Cognitive complexity has also been found to increase with identity separation, because perceptions of cultural conflict sharpen cultural awareness (Benet-Martínez, Lee, & Leu, 2006). Both findings are consistent with the argument that inconsistent identity patterns produce more dissonance than consistent patterns, and this pushes individuals to pay more active attention to cultural content, increasing cultural metacognition as a consequence.

Related arguments support the relationships between identity pattern inconsistency and two behavioral intercultural skills of adaptability and language interpretation. Both skills have been found to predict task achievement across cultures (Mol, Born, Willemsen, & Van der Molen, 2005; Shaffer, Harrison, Gregersen, Black, & Ferzandi, 2006). Adaptability is the ability to behave flexibly and appropriately according to different intercultural situations (Barner-Rasmussen et al., 2014; Cramton & Hinds, 2014; Molinsky, 2007). Language interpretation refers to the spoken word, not written texts (Barner-Rasmussen et al., 2014; Hinds, Neeley, & Cramton, 2013; Kulkarni, 2015). We focused on oral interpretation over written translation because the former is more likely to be done informally by non-professionals than the latter. Multicultural individuals are usually expected to be more adaptable than monocultural individuals because multiculturals have more cultural identities than monoculturals, and thus have access to a wider variety of cultural schemas to guide behavior (Bell & Harrison, 1996). It follows that multiculturals with the most inconsistent cultural identity patterns should be more adaptable than those with consistent patterns, because pattern inconsistency results in less overlap among behavioral repertoires, including linguistic resources. Multicultural individuals may also experience a demand effect, in which they sense they are expected to behave in a particular way. For example, individuals who are noticeably multicultural will likely feel more obligated to interpret than will monocultural-bilingual individuals whose language skills are less obvious. Noticeable multiculturalism does not necessarily mean someone is of mixed race, it could mean an accent, manner of dressing or stories about time spent in a parent’s country that make one’s multiculturalism noticeable to colleagues. This example of demand-driven differences in language interpretation likely holds across the identity plurality spectrum, in which people with more cultural identities are also more likely to experience subtle pressure to interpret. Adaptability and language interpretation should both be highest for individuals with the most inconsistent identity patterns. Finally, job performance is expected to be higher for individuals with the most inconsistent identity patterns, again driven by higher levels of cognitive complexity. For jobs within the cultural domain, bicultural professionals who identified highly with both of their cultures were found to be promoted more quickly (implying better performance) than those who prioritized one of their cultures over the other (Tadmor et al., 2012). The authors argued that the effect on promotions was driven by cognitive complexity. According to our definition of identity plurality, the former group had higher levels of identity plurality than the latter, indicating that identity plurality had a positive relationship to performance.

H3a:

Identity plurality will be positively related to cultural metacognition, adaptability, language interpretation and job performance.

H3b:

Identity integration will be negatively related to cultural metacognition, adaptability, language interpretation and job performance.

Together these personal, social and task outcomes illustrate what happens when cultural identity patterns influence the way people think and behave. We designed a series of studies to test these predictions with both multicultural and monocultural individuals as summarized in Table 2.

Table 2 Multicultural identity outcome hypotheses, mechanisms, and summary of results
Full size table

METHOD

The series of three studies involved 1196 participants. Study one tested hypotheses with multicultural individuals, and studies two and three included both monocultural and multicultural employees, within a hotel chain and health care organization, respectively. Table 3 reports the cultural characteristics for the samples in each study.

Table 3 Frequency of most common cultures represented in each sample
Full size table

Study one: multicultural employees

Study one was designed to test the personal and task outcomes of multicultural identity patterns, and it also provides preliminary validation for new scales used in studies two and three. Based on the rationale that internally inconsistent identity patterns reduce uncertainty less effectively than consistent patterns, we expected that individuals drawing on patterns with higher inconsistency (higher identity plurality or lower identity integration) would have lower personal well-being (h1) and higher levels of intercultural skills (h3) than individuals with more consistent patterns (lower plurality, higher integration). Thus feeling overburdened with cultural translation work and identity uncertainty (lack of personal well-being) and cultural metacognition were all expected to increase along with identity plurality, and decrease along with identity integration. Social outcomes were not tested with this sample because participants’ organizations exhibited wide variance in cultural demographics and size, making social network comparisons difficult. For example, some respondents were individual entrepreneurs, while others worked for highly culturally diverse organizations.

Participants and Procedure

This study was conducted as a web-based survey with 300 employed multicultural students (mean age = 20.79 years, SD = 2.45 years, 136 male, 164 female), using Fluidsurveys online software. Participants were undergraduate students and participated in exchange for partial course credit. They filled out the survey at scheduled times in a computer lab on campus, ensuring they were not distracted. Access was controlled to avoid dual submissions. Participants entered their cultural identities on the first page of the online survey (e.g., Canadian and Austrian). All further questions were tailored to the respondent’s own unique set of cultures (e.g., “When I`m at work, my Austrian identity is …”). All participants self-identified as multicultural, based on the following definition:

You’re multicultural if you have more than one cultural identity. A cultural identity is a culture that is so familiar to you that it becomes part of who you are. A culture can refer to a region or a country. For example, Chinese, Indian, and French-Canadian are all cultures. You can be a member of a culture even if you’ve never lived there, but it must be so deeply embedded in you that it influences your values, your behaviors and the way you see the world.

All respondents identified with at least two cultures, forty-five identified with three cultures and nine identified with four cultures, totaling 18% of the sample with more than two cultural identities. In total, forty-six different cultures were represented, as shown in Table 3. Canadian culture was reported by 289 respondents and Chinese by 192. The next most common cultures were Indian (N = 40), Taiwanese (N = 18), and Hong Konger (N = 11). Non-country-based cultures included Christian (N = 10), Sikh (N = 4), Muslim (N = 4) and Arab (N = 3). On average the sample had lived in Canada for 11.8 years (SD = 7.3), spoke 2.1 languages (SD = 0.7) and the average self-reported English language fluency was 3.1 out of 4 (SD = 0.7), indicating a high degree of fluency in English.

Measures

The measures in the study were identity plurality, identity integration, feeling overburdened with cultural translation work, identity uncertainty, cultural metacognition. Unless otherwise stated, all variables were measured on a five-point Likert-type scale (0 = strongly disagree; 4 = strongly agree), such that higher values indicate higher levels of the corresponding constructs.

Identity plurality was calculated as the sum of identification with each of the participant’s cultures [identification with culture one + identification with culture two + identification with culture three], such that people who identified strongly with more cultures had higher scores than those who identified strongly with only one culture. It was calculated additively to represent current understanding that identification with one culture does not necessitate reduced identification with another (Berry, 1997). Instead, individuals can identify strongly with two or more cultures, represented by an additive relationship. Identification with each culture was measured with Roccas et al. (2008) four-item identity importance subscale (alphas range from 0.80 to 0.91 in the current study, calculated separately for identification with each culture). It measures the centrality of a social identity to one`s self-concept, including the following item “It is important to me that I view myself as a member of culture X.” We first calculated an average score of identification with each culture (ranging from 0 to 4), and then calculated a sum score across all cultures for each participant (ranging from 1.5 to 16). For example, someone whose identification scores were 3 (culture one) and 2 (culture two) would have an identity plurality score of 5.

The identity importance subscale is more appropriate for measuring identity plurality than the other three modes of identification – commitment, deference and superiority. Identity importance most clearly taps the cognitive aspect of identification as it is most closely linked to self-construal. Consistent with social identity theory, this cognitive dimension is expected to be influenced by contextual changes such as immigration or retirement (Roccas et al., 2008). Thus it fits the theorizing in this model better than, for example, commitment, which taps the degree to which an individual wants to benefit the group.

Identity integration (α = 0.84) was measured with the nine-item blendedness subscale from the revised bicultural identification inventory (BIIS-2; Benet-Martínez, 2010; Huynh & Benet-Martínez, 2009). It measures the degree to which individuals perceive their cultural identities as fused versus dissociated. A sample item is “I feel X and Y at the same time.”

Two personal outcomes of feeling overburdened with cultural translation work, and identity uncertainty were measured. Feeling overburdened with cultural translation work (3 items; α = 0.89) was measured with three items written for this study in the style of the Global Measure of Perceived Stress (Cohen, Kamarck, & Mermelstein, 1983). The Global Measure of Perceived Stress indicates the degree to which individuals experience situations as stressful as a result of overloading (Cohen et al., 1983), and is among the most common scales used to measure general stress (Cerclé, Gadéa, Hartmann, & Lourel, 2008). This scale measured the degree to which individuals experienced their regular situations as stressful as a result of an overwhelming amount of time spent on cultural translation activities, such as helping others understand other cultures. Items were “In the past four months, how often have you felt like you were spending too much time helping others understand one of your cultures?”; “In the past four months, how often have you felt like explaining different cultures to people was a burden?” and “In the past four months, how often has your schoolwork suffered because you had to help other people understand a different culture?” Responses were collected on a scale from 0 = never to 4 = very often.

Identity uncertainty (3 items; α = 0.75) measured the degree to which individuals were unclear or uncertain about their cultural identities. The scale was adapted from the three-item uncertainty subscale of the measure of sexual identity exploration and commitment (MoSIEC) (Worthington, Navarro, & Savoy, 2008). In this adaptation sexual identity was replaced with cultural identity (e.g., “My cultural identity is not clear to me”). To examine support for our adapted scales, we conducted a confirmatory factor analysis (CFA) in which all items for each adapted scale (identity importance, identity uncertainty, feeling overburdened) were loaded onto their theoretical factors. We found good model fit, indicating items loaded as expected: χ 2 (32) = 34.51, p = 0.35, CFI = 1.0, RMSEA = 0.02, SRMR = 0.03.

Cultural metacognition (12 items; α = 0.89) was measured with a scale developed by Thomas and colleagues (2012) that measures the degree to which respondents are aware of, and have control over, their own thinking and learning activities within the cultural domain (sample item “In situations when I have interacted with people who are culturally different, I really enjoy analyzing the reasons or causes for their behavior”). Demographic questions were age (years), gender (1 = female), and English language fluency (0 = very poor to 4 = excellent).

Analysis and Results

In order to examine identity plurality against a straight count of the number of cultures reported by participants, we first conducted a one-way analysis of variance (ANOVA). Results indicated that identity plurality was significantly different across groups (two, three or four cultures) in the expected direction (F(2, 300) = 123.68, p < 0.01). Post-hoc comparisons using the Scheffé criterion indicated that identity plurality was significantly lower for individuals who identified with two cultures (M = 5.49, SD = 1.09), than it was for individuals who identified with three cultures (M = 7.74, SD = 1.52), and that identity plurality for the three-cultures group was significantly lower than it was for the group identified with four cultures (M = 10.67, SD = 3.30). Together, these results indicate that this measure of identity plurality represents the construct consistently with a simple count of each individual’s cultures.

Means, standard deviations, and correlations of key variables are presented in Table 4. Identity plurality and identity integration were significantly correlated with a modest effect size (r = 0.36, p < 0.01), indicating that multiculturals are more likely to integrate their cultures when they have internalized more of them. To test if these two dimensions are empirically distinct, we conducted CFAs with Chi square difference test using the four items capturing identity importance of the first culture and the nine items capturing identity integration. We loaded all 13 items on one latent factor in the first CFA model, and loaded the items on their respective theoretical factors in the second CFA model. We found a significant improvement in model fit with the two-factor model over the one-factor model. The two latent factors were not correlated with each other, suggesting the identity importance scale and the identity integration scale captured two distinct constructs as theorized. Full CFA results are available from the first author.

Table 4 Correlations, means and standard deviations in Study1
Full size table

In subsequent analyses we included both dimensions to control for shared variance and conducted a series of hierarchical regressions. Three separate outcome variables (feeling overburdened with cultural translation work, identity uncertainty, and cultural metacognition) were regressed separately on identity plurality and identity integration. English fluency, the only demographic variable correlated with dependent variables, was included as a control. Results are presented in Table 5. Supporting predicted relationships, identity integration was negatively related to feeling overburdened with cultural translation work (β = −0.24, p < 0.01) and identity uncertainty (β = − 0.22, p < 0.01). This indicated that feeling uncertain about one’s identity and feeling overburdened with cultural translation work both decreased along with identity integration. On the other hand, identity plurality was significantly related to cultural metacognition (β = 0.19, p < 0.01), supporting the expected relationships, and indicating that cultural metacognition increased along with the number of internalized cultures.

Table 5 Multiple regression results in Study 1
Full size table

Discussion

This study provided an initial test of relationships predicted by our theoretical framework about relationships between multicultural identity patterns and outcomes. Results supported expectations that multiculturals who separate their cultural identities tend to experience lower levels of personal well-being (as measured by higher identity uncertainty and feeling overburdened with cultural translation work) than those who integrate their cultural identities. Also, identity plurality in multiculturals was related, as expected, to increased cultural metacognition.

With respect to the objective of developing an appropriate measure for identity plurality, results indicated that this measure – a sum of identification with each culture – was a more nuanced representation of the identity plurality construct than counting cultures alone. It differentiated between those who identify with one primary culture and two secondary (less important) cultures, versus someone who identifies strongly with three cultures. Counting cultures, by itself, ignores variability in the centrality or importance of each cultural identity.

Sampling only multicultural individuals in this study precluded the possibility of comparing multicultural and monocultural individuals along the spectrum of identity plurality. Also, these participants were in their early 20s, suggesting their cultural identities might still be evolving (Phinney & Devich-Navarro, 1997). Our next step was to further examine the effect of identity plurality on outcomes within the context of organizations, including both monocultural and multicultural employees.

STUDY TWO: HOTEL EMPLOYEES

Study two was designed to accomplish three goals: (1) examine the full spectrum of identity plurality, including both monoculturals and multiculturals; (2) validate the results from study one in a different context, and (3) test social outcome hypotheses. Specifically, we expected to find that identity plurality would be positively related to feeling overburdened with cultural translation work (H1a), cultural metacognition (H3a) and supervisor-rated job performance (H3a), while identity integration would be negatively related to the same (H1b, H3b); and that identity plurality would be positively related to in-group cultural composition (H2). This study was conducted with employees at a small hotel chain located in tourist areas and large cities that experience significant numbers of international visitors. Both the employee base and hotel guests were highly culturally diverse, creating an ideal environment to test for skills and abilities in the cultural domain.

Participants and Procedure

This study was conducted with a paper-based survey matched with participant’s most recent supervisor-rated performance appraisals. Seventy-seven employees across five locations in a hotel chain in western Canada completed this survey. Participants self-identified as either multicultural (N = 40) or monocultural (N = 37), based on the same definition of multiculturalism used in study one. The sample included 44 women and 33 men, with a mean age of 36.8 years (SD = 11.3 years). They rated their English language fluency as 3.5 out of 4 (SD = 0.7) and had worked for this organization for an average of 6.0 years (SD = 6.7 years). There was no statistically significant difference between monoculturals and multiculturals, on gender, age or English language fluency. On average, the multicultural respondents had lived in Canada for significantly fewer years than the monocultural group (t = 2.34, p < 0.05), with mean residency of 25.9 years versus 33.2 years, respectively. This indicated that several may have immigrated to Canada as children. All multicultural respondents identified with at least two cultures, six respondents identified with three cultures and one identified with four cultures. Thus 18% of the multicultural respondents identified with more than two cultures. In total, 26 different cultures were represented (see Table 3). Participants were contacted through posters and announcements during weekly meetings, and offered an incentive of the chance to win a prize.

Measures

The following variables were all measured using the same scales as in study one: identity integration (α = 0.82), identity plurality (α culture1 = 0.94, α culture2 = 0.81, and α culture3 = 0.83), feeling overburdened with cultural translation work (α = 0.84), and cultural metacognition (α = 0.92). All of the response sets for these constructs were five-point Likert scales (0 = strongly disagree; 4 = strongly agree), such that higher values indicate higher levels of the corresponding constructs. For monocultural individuals, identity plurality necessarily measures identification with only one culture. Identity plurality ranged from 0 to 10.5 for the complete sample (including monoculturals and multiculturals).

Representing social outcomes, in-group cultural composition was calculated as the number of cultures in each respondent’s in-group, excluding cultures to which the participant belonged, so the measure would represent access to other cultural networks. This measure was influenced by the in-group cultural composition measure presented by Mok and colleagues (2007). Participants were asked to list their five closest friends at work, and denote each friend’s culture(s). Each friend was coded against the following criteria, representing the number of cultural networks to which respondents have access through their friendship network: 0 for no distinct access (e.g., a Chinese-Canadian respondent with a Chinese-Canadian friend); 1 for access to one distinct cultural network (e.g., a Chinese-Canadian respondent with either an Israeli or Israeli-Canadian friend); 2 for access to two distinct cultural networks (e.g., a Chinese-Canadian respondent with an Israeli-Bulgarian friend). Scores were summed across all five friends to create a measure of the number of distinct cultural networks to which participants had access through their in-groups at work. While all hotel locations were populated by a very culturally diverse workforce, some variation in this measure could have been the result of the number of distinct cultures represented at each hotel. Reported ahead, we used dummy variables to control for variance associated with hotel locations in subsequent regression analyses.

Job performance was measured by the only two items measuring overall performance on annual performance evaluations conducted by supervisors. These were: (1) “When performing the job, to what extent does this employee demonstrate the interest/enthusiasm, initiative, productivity, accuracy/quality, and safety and compliance expectations for the role?”; (2) “To what extent does this employee demonstrate the job knowledge, skills and abilities, and work habits expected in this role?” At this hotel these annual performance ratings were used to help determine benefits and promotions. Forty-six performance evaluations were collected, but only lower-level employees, including housekeeping, front desk and restaurant staff, were rated quantitatively on job performance and job competence (31 total). Of these, 19 forms evaluated multicultural employees, and 12 evaluated monoculturals. The response set ranged from 1 = does not meet expectations to 4 = exceeds expectations, such that higher values indicate higher performance. The two performance items were highly correlated (r = 0.65, p < 0.01) and were combined in subsequent analysis. Demographic questions were English language fluency (0 = very poor to 4 = excellent), gender (1 = female), age (years), work experience at this hotel chain (years), and four dummy variables (1 = located at that hotel) to account for variability across the five hotel locations.

Analysis and Results

Identity integration and identity plurality were not significantly correlated in this sample (r = 0.16, p = ns; multicultural participants only). In addition, the sample size in this study was too small to conduct the CFA test of construct distinctness reported in study one. As shown in Table 6 and consistent with expectations, identity plurality was significantly and positively related to in-group cultural composition (r = 0.43, p < 0.01) and cultural metacognition (r = 0.25, p < 0.05).

Table 6 Correlations, means and standard deviations in Study 2
Full size table

Measuring identity plurality. In order to examine identity plurality against a straight count of the number of cultures identified with, we first conducted a one-way ANOVA using the same process reported in study one. Groups (monocultural, bicultural, multicultural) had significantly different levels of identity plurality in the expected direction (F(2, 77) = 80.24, p < 0.01), indicating that this measure of identity plurality represented the construct consistently with counting cultures, but with additional variance related to identity importance. Post-hoc Scheffé comparisons indicated significant differences among each of monoculturals (M = 2.91, SD = 1.08), biculturals (M = 5.46, SD = 1.20) and multiculturals – three or four cultures (M = 7.47, SD = 1.52).

Comparing Multiculturals and Monoculturals

Before testing hypotheses, we conducted a test of mean difference to explore differences between multicultural and monocultural employees. As shown in Figure 1, monoculturals were theorized to be at the lowest point of identity plurality, so we expected the relationships proposed for identity plurality to also hold between monocultural and multicultural employees. That is, we expected multicultural employees to report lower levels of personal well-being, and higher structural social capital, higher levels of intercultural skills and higher job performance than monocultural employees. Results indicated significant differences on three out of four variables, all in the expected direction. Multiculturals scored higher than monoculturals on job performance (t(29) = 2.30, p < 0.05), and feeling overburdened with cultural translation work (t(76) = 2.09, p < 0.05), indicating that while multicultural employees may have felt more overburdened with helping colleagues understand other cultures, they were also evaluated better than monoculturals on performance evaluations. Multiculturals’ in-groups were composed of individuals from more cultures beyond their own than monoculturals (t (76) = 6.40, p < 0.01), indicating that monoculturals’ in-groups were composed of more people from their own culture(s) than those of multiculturals. There was no significant difference between monoculturals and multiculturals on cultural metacognition, indicating this may be a trainable skill.

Main Effects

Identity plurality and identity integration were not correlated in this sample, and only multicultural individuals had scores for identity integration. Therefore we examined regression results for identity plurality separately from identity integration. Results are presented in Table 7. English language fluency was the only demographic variable correlated with dependent variables, so it was entered as a control in the first step along with the four hotel dummy variables. After controlling for English language fluency and hotel location, identity plurality was significantly related to in-group cultural composition (β = 0.42, p < 0.01), supporting H2, and cultural metacognition (β = 0.28, p < 0.05), supporting H3a. There was no significant relationship between identity plurality and feeling overburdened with cultural translation work (β = 0.15, p = ns), failing to provide support for H1a.

Table 7 Multiple regression results in Study 2
Full size table

Again controlling for the effects of English language fluency and hotel location, identity integration explained significant additional variance with respect to feeling overburdened with cultural translation work (β = −0.45, p < . 01), supporting H1b. Results offered mixed support for H3b, as identity integration was significantly related to job performance (β = −0.63, p < 0.05), as expected, but not related to cultural metacognition (β = 0.02, p = ns).

Discussion

These results have implications for assessment of the overall theoretical framework. Compared to study one, this study included monoculturals to examine the full logical range of identity plurality. When the results of this study are taken as a whole, they indicate mixed support for relationships predicted by our framework. Social and task, but not personal outcomes were significantly related to identity plurality, while both personal and task outcomes were significantly related to identity integration. Results related to job performance should be interpreted with some caution. We had only 31 supervisor ratings of job performance, and there are many competing factors that also influence job performance. We do not propose that multicultural individuals will generally have higher job performance than monocultural employees. Instead, we expect that cultural identity patterns may influence performance primarily for tasks related to culture. This is consistent with earlier findings that identity integration influences cognitive complexity, but primarily for tasks related to culture, indicating that outcomes related to multicultural identity patterns may be domain-specific (Benet-Martínez et al., 2006). Even though only two participants (one monocultural, one multicultural) did not identify with Canada, some of the multicultural participants may have been the only representatives of their non-Canadian culture at their hotel locations, limiting the potential to develop in-group relationships within those cultures. For this reason, our social outcome findings should also be interpreted with caution. Our next step was to look for confirmation of the expected relationships among a larger group of employees, and test relationships with additional measures of task outcomes.

STUDY THREE: HEALTH CARE EMPLOYEES

Study three was designed to replicate findings from studies one and two by examining additional measures of personal and task outcomes with a large sample of employees. The health care organization chosen for this study is located in a large, multicultural North American city, where language interpretation is a common, but unexpected, part of the job for many health care practitioners. Consistent with studies one and two, we expected to find that feeling overburdened with cultural translation work (H1), identity uncertainty (H1), cultural metacognition (H3), adaptability (H3), and language interpretation (H3) would all increase along with identity plurality, and decrease along with identity integration. To adhere to the organization’s privacy concerns, and allow adequate space in the survey for task outcomes related to language interpretation, social outcomes were not measured in this sample. This health care organization had professional interpreters on staff, but employees often found themselves in situations where it was inconvenient to find an interpreter for a brief conversation. Many employees admitted to interpreting informally, even though this was not official company policy. Thus the ability to interpret, and willingness to do so, represented a useful intercultural skill in this environment.

Participants and Procedure

This study was conducted with a web-based survey of 819 employees at a Canadian health care organization using Fluidsurveys software. As in study one, participants’ self-identified cultures were dropped into items so participants could clearly see which culture was being referenced.

Participants self-identified as either multicultural (N = 340) or monocultural (N = 479), based on the same definition of multiculturalism used in studies one and two. The sample included 710 females and 109 males, consistent with the female-dominated employee population in this organization. The mean age was 42.9 years (SD = 10.3 years) and participants had worked for this health care organization for an average of 10.44 years (SD = 8.57). Multicultural respondents rated their English language fluency as 3.87 out of 4 (SD = 0.3), indicating a very high level of English language ability. Monocultural respondents (M = 43.7, SD = 10.4) were significantly older than multicultural respondents (M = 41.5, SD = 10.1) (t = 3.03, p < 0.01), had more work experience (M monocultural = 11.1, M multicultural = 9.5, t = 3.36, p < 0.01) and better English language skills (M monocultural = 3.94, M multicultural = 3.87, t = 3.03, p < 0.01), so these variables were controlled in subsequent analyses. There was no gender difference between monoculturals and multiculturals. All multicultural respondents identified with at least two cultures, 87 respondents identified with three cultures and 20 identified with four cultures. Thus 31.5% of the multicultural respondents identified with three or more cultures. The sample included 66 different cultures, presenting a wider range of cultures than the first two studies (see Table 3). Participants were contacted through an email sent from the Diversity Officer of this health care organization, and were offered an incentive of the chance to win a prize.

Measures

The following variables were all measured using the same scales as in previous studies: identity integration (α = 0.81), identity plurality (α = 0.88 to 0.91; range of 0 to 13.3), identity uncertainty (α = 0.71), feeling overburdened with cultural translation work (α = 0.83), and cultural metacognition (α = 0.90). We also added two additional intercultural skill variables that were important in this organizational context (adaptability and language interpretation). Adaptability (5 items, α = 0.81) is a scale that measures behavioral flexibility across intercultural situations (sample item, “I tend to show different sides of myself to people from different cultures”) (Thomas et al., 2012). All of these response sets were five-point Likert scales (0 = strongly disagree; 4 = strongly agree), such that higher values indicate higher levels of the corresponding constructs. We again conducted a CFA with four factors (identity importance, identity uncertainty, feeling overburdened, and adaptability) to provide support for the independence of our adapted scales. The model fit well (χ 2 (84) = 229.43, p < 0.01, CFI = 0.92, RMSEA = 0.07, SRMR = 0.06).

Language interpretation is one of the most common intercultural skills expected of multicultural individuals, because individuals who are multicultural are often also expected to be multilingual (Chen et al., 2008). In this health care organization, it was the most commonly discussed outcome. Employees were both proud of and concerned about the frequency with which employees translated for patients. Languages spoken (M = 1.34, SD = 0.76) was measured by asking participants how many languages they knew well enough to interpret for a patient who spoke that language (including English). Responses ranged from individuals who spoke only one language (n = 731) to six languages (n = 3), including 129 participants who spoke two languages. Times interpreting (M = 1.57, SD = 1.83) was measured by how many times participants were asked to interpret in the past month. Responses ranged from 1 (n = 657) to 12 (n = 13). Demographic questions were English language fluency (0 = very poor to 4 = excellent), gender (1 = female), age (years) and work experience (years).

Analysis and Results

Consistent with studies one and two, a one-way ANOVA found a significant difference in identity plurality across groups (monoculturals, biculturals, multiculturals) in the expected direction (F(3, 799) = 525.42, p < 0.01). Post-hoc Scheffé comparisons indicated significant differences between monoculturals (M = 3.11, SD = 0.88), biculturals (M = 5.72, SD = 1.43) and multiculturals (M 3cultures = 7.73, SD = 1.92; M 4cultures = 9.22, SD = 2.42).

Identity integration and identity plurality were significantly correlated in this sample with a small effect size (r = 0.27, p < 0.01). To test if these two dimensions were empirically distinct, we conducted the same CFA analysis reported in study one. Again, we found a significant improvement of model fit in the two-factor model, in which items were loaded on their respective theoretical constructs, over the one-factor model, in which items were loaded on one single factor. This supported the independence of the two constructs. Full CFA results are available from the first author. Age (r = −0.09, p < 0.05) and years of work experience (r = −0.07, p < 0.05) were both negatively related to identity plurality, while English language fluency (r = 0.15, p < 0.01) was positively related to identity integration. Therefore all three variables were included as control variables in subsequent analyses. Means, standard deviations and correlations are reported in Table 8.

Table 8 Correlations, means and standard deviations in Study 3
Full size table

Comparing Multiculturals and Monoculturals

We next conducted tests of mean difference to examine differences between multicultural and monocultural employees. We did not hypothesize differences between groups per se. However, based on the same arguments as in H3a, we expected task outcomes to be higher for multicultural than monocultural employees along the dimension of identity plurality. Results supported all four expected task relationships. Multiculturals spoke significantly more languages than monoculturals (t (817) = −5.96, p < 0.01); were asked to interpret significantly more frequently (t (760) = −5.53, p < 0.01); were more adaptable (t (772) = −3.76, p < 0.01); and had higher cultural metacognition (t (772) = −5.17, p < 0.01). We also expected that personal well-being would be lower for multicultural than monocultural employees, based on the arguments for H1a. Results supported one out of two expected relationships. Multiculturals reported feeling significantly more overburdened with cultural translation work (t (772) = −5.09, p < 0.01), but there was no significant difference between monoculturals and multiculturals on identity uncertainty (t (772) = 0.12, p = ns).

Main Effects

We next conducted regression analyses to examine relationships between multicultural identity patterns and outcomes. Results are shown in Tables 9 and 10. To retain monocultural individuals (who do not have a score for identity integration) in the identity plurality analyses, we conducted analyses separately for identity plurality and identity integration. After controlling for the effects of age, English language fluency and years of work experience, identity plurality explained significant additional variance in personal and task outcomes, as expected. Individuals who internalized more cultural identities felt more overburdened with cultural translation work (β = 0.16, p < 0.01), spoke more languages (β = 0.19, p < 0.01), were asked to interpret more often (β = 0.16, p < 0.01), had higher levels of cultural metacognition (β = 0.24, p < 0.01), and higher adaptability (β = 0.13, p < 0.01). However, contrary to expectations, individuals with lower identity plurality were more uncertain about their identities (β = −0.11, p < 0.01). After demographic controls, identity integration also explained significant additional variance in outcomes. As expected, individuals who separated their identities were more uncertain about their identities (β = −0.17, p < 0.01), while, counter to expectations, those who integrated their identities were asked to interpret more often (β = 0.15, p < 0.01). In sum, individuals with integrated cultural identities felt less uncertain about their identities and were asked to interpret more often. Those who internalized more cultural identities had higher levels of cultural metacognition and adaptability, spoke more languages, interpreted more frequently for patients, felt more certain about their identities and yet, felt more overburdened by cultural translation activities.

Table 9 Multiple regression results on identity plurality in Study 3
Full size table
Table 10 Multiple regression results on identity integration in Study 3
Full size table

Discussion

This study offered clear support for the relationship between identity plurality and task outcomes, with all four variables in this category statistically significant (H3a). New to this study, identity plurality was found to be significantly related to language interpretation, a key concern for health care employees who work with non-English-speaking patients. Other results are more tentative, with conflicting results for the relationship between personal outcomes and identity plurality (H1a), and with one out of two personal outcomes significantly related to identity integration (H1b). Only one out of four task outcomes was significantly related to identity integration (H3b), and it was counter to the expected direction. That is, individuals reported they were asked to interpret more often when their cultural identities were more integrated. Given current concerns about replicability of scientific results (Simons, 2014), we examine these results in light of all three studies in the overall discussion.

OVERALL DISCUSSION

Individuals who have internalized multiple cultural identities are changing the focus of international management research from crossing cultural boundaries to straddling them. The three studies in this article test a general framework that clarifies relationships between cultural identity patterns and the potential of individuals to contribute to their organizations. Study one tested relationships with employed students in an online survey, while studies two and three examined additional outcomes that are particularly salient in their work contexts: social networks at a hotel chain, and language interpretation in health care.

Across all three studies, individuals were found to vary in their skills, abilities and challenges, depending on their multicultural identity patterns. Consistent results across studies, shown in Table 2, support three overall conclusions. Identity integration predicts personal outcomes (H1b), while identity plurality predicts social (H2) and task outcomes (H3a). That is, individuals who integrate their cultural identities experienced higher levels of personal well-being than those who separated them, while individuals with more cultural identities had more social capital and higher levels of intercultural skills than those with fewer cultural identities. These conclusions do not align perfectly with the theoretical model we tested, suggesting a direction for improving the framework. The pattern of findings across studies indicates that identity integration and identity plurality may not work in tandem or through the same mechanism of reducing uncertainty, as initially theorized. This indicates potential paths for developing the model, which we explore in the following section.

Theoretical Implications

Three major theoretical implications stand out from our overall pattern of results. First, each framework dimension requires a distinct set of theoretical mechanisms to explain why they resulted in a different set of outcomes. Second, identity plurality is a useful dimension for predicting outcomes across monoculturals and multicultural individuals, without relying on an artificial binary division between groups. And third, it is time to expand research on multicultural individuals beyond psychological outcomes, to also include positive social and task outcomes. Building on Fitzsimmons (2013), we had originally argued that, through the mechanism of reducing uncertainty, task outcomes would improve and personal outcomes would deteriorate with both identity plurality and identity separation. Instead, we found that personal and task outcomes did not correspond, in that personal outcomes were better with higher levels of identity integration, while task outcomes were better with higher levels of identity plurality.

Recent developments in cognitively-focused research on multiculturals (Lakshman, 2013; Lücke et al., 2014) indicates a potential way to reconcile this difference between our theoretical expectations and findings. One possible explanation is that identity integration is naturally a cognitive construct, one turned inward, such that outcomes are also internal (i.e., personal well-being). In contrast, identity plurality may draw on factors beyond the cognitive, including power dynamics, behavioral repertoires, context and social interactions, naturally resulting in more externally-oriented outcomes (i.e., social and task outcomes). Based on this distinction, the original uncertainty-reduction mechanism can be used to explain the relationship between identity integration and personal outcomes, while relationships between identity patterns and social and task outcomes require alternative explanatory mechanisms. This may begin to address an issue identified by Lücke et al. (2014: 185), “it is important to consider in which situations cultural identities matter for thought and action, and in which situations other parts of self are more salient.” There is some evidence supporting this boundary condition. For example, Brook et al. (2008) found that individuals developed higher levels of well-being when they had more facilitative identities, meaning their identities encompassed similar behaviors, while they developed worse well-being when they had more conflicting identities, meaning their identities encompassed incompatible behaviors. It may be useful for future research to attempt to reconcile the mechanisms driving different sets of outcomes by measuring socially-oriented mechanisms and power dynamics, such as those often used in research on intersectionality (Howard, 2000).

Intersectionality research examines individuals’ mutually reinforcing identities, often including gender and race (Werbner, 2013). In contrast to multiculturalism research, which usually examines multicultural individuals in a positive light, intersectionality is usually cast as a negative experience. For example, Werbner (2013) describes intersectionality as being multiply burdened by multiple spheres of oppression. Despite this negative orientation, it may indicate a way to combine the cognitive and social mechanisms associated with multicultural identity patterns by considering external mechanisms such as power dynamics and social connections. For example, outcomes stemming from Aboriginal-Australian or Arab–Israeli identities could only be understood within the context of power dynamics associated with these unique and challenging cultural combinations (Howard, 2000). This research direction parallels the development of cultural friction research, where researchers are calling for an integration of factors such as power and context, beyond cultural distance alone (Shenkar, Luo, & Yeheskel, 2008). Similarly, our results indicate that there is likely more to the relationship between identity plurality and outcomes than can be explained by cognitive mechanisms alone. Thus we encourage future research to look beyond the cognitive in examining how identity patterns influence outcomes.

Our second theoretical implication relates to the first, in that our results indicate the value of examining identity plurality as a continuous dimension, rather than comparing monoculturals and multiculturals categorically. Indeed, identity plurality seems even more predictive of social and task outcomes than identity integration. Our results held across monoculturals, biculturals and multiculturals. When we re-imagine the monocultural–multicultural dichotomy as a continuum, we can ask research questions that we could not otherwise ask. Conceptual work on ‘n-culturals’ suggests there are beneficial effects of both international exposure and international experience along a related spectrum (Pekerti & Thomas, 2016). For example, the identity plurality continuum approach allows us to consider if the differences between individuals with one versus two cultures are similar to the differences between those with two versus three, and if these are differences in kind or in degree.

Finally, we see the results of all three studies pointing toward opportunities to take research on multicultural individuals beyond its origins in psychology toward more social and task outcomes. Our research especially facilitates research into the positive outcomes of culture by exploring cultural diversity within individuals (Stahl & Tung, 2015). We think the continued exploration of social capital outcomes and linguistic resources are both particularly valuable. For example, if individuals’ social network structures reflect their identity patterns, identity integration could be related to the degree of closure in social networks. Individuals with integrated cultural identities may have networks with higher levels of closure than individuals who separate their identities, where closure refers to the degree to which an individual’s network ties are themselves connected (Adler & Kwon, 2002; Burt, 1992). We expect this to occur because individuals who separate their identities may be more likely to develop distinct groups of monocultural friends associated with each of their cultures, and these groups may not be connected to one another. For example, an Arab-American who separates her identities may have Arab friends, and also American friends, making this person valuable as someone who can bridge networks, whereas an Arab-American who integrates her identities may have more Arab-American friends, making this person less valuable as a network bridge. When an individual connects groups that would not otherwise be connected (the former example), that person develops more social capital (Adler & Kwon, 2002; Burt, 1992).

Language research is becoming increasingly prominent in international business. We conceptualized languages as an outcome variable for both theoretical and pragmatic reasons. Theoretically, it allowed us to clearly distinguish between multiculturalism and multilingualism. Pragmatically, the health care organization where we studied language outcomes cares deeply about being able to predict instances of language interpretation. They have professional interpreters on staff, and health professionals are not allowed to interpret informally, lest they do it poorly. Nonetheless, administrators were aware that this happened, and wanted to know why. As we show, identity plurality was related to the number of languages spoken and the frequency of interpretation activities, indicating one possible reason for the phenomenon of unsanctioned interpretation. However, our conceptualization of linguistic outcomes resulting from identity pattern antecedents is only one possible explanation for this complex relationship. It is entirely possible that causation instead flows from languages to social identity, where knowledge flow among those who share a language is used to construct a shared social identity (Reiche, Harzing, & Pudelko, 2015). Alternatively, they might both be independent antecedents of other outcomes, such as boundary spanning (Barner-Rasmussen et al., 2014). We see the most potential for research that examines the relationships between multiculturalism and multilingualism and personal, social and task outcomes, to identify how effects differ across these two potential antecedents.

Together, all three theoretical implications suggest a direction for future research on multicultural employees that examines monocultural and multicultural employees together along the dimension of identity plurality to predict social and performance-based outcomes, while moving beyond cognitive mechanisms alone. Similar to cultural distance in cross-cultural studies (Shenkar et al., 2008; Zaheer, Schomaker, & Nachum, 2012), identity integration has become a ‘must-have’ dimension of analysis in research on multicultural individuals. This focus is reasonable, so long as cognitive explanations are seen as part of the picture, and not the whole explanation for multicultural individuals’ outcomes. To broaden the scope of theoretical developments related to multicultural employees, we call for researchers to consider a wider range of mechanisms in future research on multicultural employees. Those mechanisms will contribute to better understanding of the roles multicultural employees play in expatriate assignments, multicultural teams, global leadership, and related IB activities. We hope this research can help international business scholars recognize that cultural diversity exists beyond teams, organizations and countries. It also exists within individuals.

Practical Implications

Multicultural employees can use this identity pattern framework to help them recognize their own contributions and challenges in their workplaces, and develop agency over their self-representation. Multicultural individuals sometimes see themselves in terms of minority status and often associate this with negative outcomes. If some individuals shift their self-representations from minority to multicultural as a result of this framework, they may place more emphasis on their potential access to intercultural skills and abilities as opposed to their potential to be the object of discrimination. One consequence of this shift in self-representation may be increased confidence in individuals’ potential to make positive contributions to their organizations. This shift of focus within individuals parallels the organization-level shift we discussed in the introduction to this article, from considering cultural diversity between individuals to also considering it within individuals.

Beyond the implications for multicultural individuals themselves, the model tested in this article offers insight for managers of multicultural employees as they help to transition this growing demographic from an unrecognized entity to a valued resource. For example, Cramton and Hinds (2014) found that multicultural individuals played a pivotal role in the informal and iterative process of cultural adaptation that occurred among binational software development teams. Among their participants, team members turned first to informal liaisons, such as multicultural individuals, before developing more formal liaison positions. Their finding supports several of the conclusions from the framework presented here. Teammates may expect their multicultural team members to act as liaisons because they are more likely to have boundary-crossing social networks and additional language skills, but when this expectation is added on top of employees’ usual work expectations it can cause multicultural individuals to feel overburdened. One of the participants in Cramton and Hinds’ (2014) study reported that acting as his team’s “bridge to India” consumed 70% of his time (p. 10). Managers working with multicultural teams should examine the amount of time their multicultural employees spend performing liaison, coordination or translation (cultural or language) activities. They should also consider whether this role is central to their performance in their organizational context, as it would be for hotel front desk staff, or peripheral, as for health care providers who may be spending their time on cultural translation tasks that would be better if done by specialists (Johns, 2006). If the time is found to be excessive, managers could either find a replacement liaison to take over some of those activities, or reduce multicultural employees’ other work expectations, compensating for these important – but often unrecognized – activities.

Limitations

Several limitations should be kept in mind while interpreting results from these studies. Across all three studies, most outcome variables were reported by participants on the same survey instrument used to measure independent variables, so common method bias and self-report inaccuracies should both be considered when interpreting results. In order to test for presence of common method variance, in each study we loaded items onto their respective theoretical latent factors, then loaded all items onto one factor representing the common method factor in a confirmatory factor analysis. If common method variance is largely responsible for the relationships among the variables, this CFA model should fit the data well (Korsgaard & Roberson, 1995; Mossholder, Bennett, Kemery, & Wesolowski, 1998; Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). CFA results in each case showed that the common factor model did not fit the data well. This test is less reliable for small sample sizes, such as that in study two. While the results of these analyses do not preclude the possibility of common method bias, they suggest that it is not the primary driver of relationships among variables and thus is less likely to confound interpretation of results. CFA results are available from the first author.

We took three purposeful steps to reduce common method bias in designing the questionnaires (Conway & Lance, 2010), First, identity plurality was computed as a sum of two, three or four scales, depending on the number of cultures each participant specified, such that the final value of identity plurality would not be highly influenced by common method bias. Second, outcome measures were separated contextually from independent variables (participants were primed to think about their experiences at a specific organization when answering outcome measures). And third, we included outcomes that were not measured with Likert-type response sets, such as feeling overburdened with cultural translation work (response set is frequency of occurrence), cultural composition of in-group (formula derived from reported cultures of five work friends), and job performance (as rated by supervisor). While these steps do not entirely remove the opportunity for common method bias they may suppress that potential.

We recognize that the operationalization of identity plurality could confound effects driven by number of identities, with those driven by strength of identification with each identity. To test this potential confound, and examine whether our measure of identity plurality was an improvement over a straight count of cultures, we recalculated the regression models in studies one, two and three, using a straight count of number of cultures instead of the current measure of identity plurality. In all three studies, variables that were originally significantly related to identity plurality were still related to the straight count of number of cultures in the same direction, although most were no longer statistically significantly. Only 18–32% of participants in each sample had more than two cultures, so variance was restricted by measuring identity plurality as a count of number of cultures. This restricted variance may be responsible for the lack of statistical significance. Our measure of identity plurality was slightly positively skewed in the first and third studies, so we re-analyzed the data using a logged version of the identity plurality variable. The significance of every value remained constant, indicating our original analyses were robust to small variations in skewness. Thus our new measure of identity plurality seemed to be an improvement over a basic count of cultures.

Finally, the cross-sectional design means that causal relationships could only be inferred, not tested longitudinally. All respondents were in Western Canada, such that almost all monocultural participants were monocultural Canadians. Future research in other locations is necessary before these results can be considered generalizable. Canadians are taught at a young age that theirs is a nation of immigrants, so people have a tendency to overstate rather than understate their cultural identities. For example, people will sometimes report identifying with ancestral cultural roots, even if these roots play only a marginal role in the respondents’ lives.

Beyond measurement issues, two theoretical limitations indicate other opportunities for future research contributions. First, alternative theoretical frameworks could be used to test relationships not covered by Fitzsimmons’ (2013) model. For example, Lücke et al. (2014) cognitive connectionism approach is not constrained by linear dimensions of identity patterns. It therefore proposes patterns not covered by our model, such as generalization, where individuals combine their cultures into something new and emergent. Second, our model represents a cross-sectional approach to multicultural identity patterns. It is possible that identity patterns fluctuate across contexts, or vary over time. Current research does not reveal if or how this occurs. Answers to these questions would help to develop a more dynamic understanding of how multicultural individuals contribute to their organizations.

CONCLUSION

This article tests a theoretical framework of the outcomes of different multicultural identity patterns. Critically, our results indicate a direction for building on the proposed theoretical framework by incorporating replicable results across studies. Individuals with more cultural identities were found to have more social capital and higher levels of intercultural skills than those with fewer cultural identities, while individuals who integrated their cultural identities were found to experience higher levels of personal well-being than those who separated them. Thus we predict a shifting research focus, from examining the degree to which multicultural individuals cognitively integrate their cultural identities, to examining identity plurality as a spectrum that includes both monoculturals and multicultural individuals. This new spectrum approach avoids an artificial dichotomy between these two groups. When researchers and practitioners understand how to draw on cultural complexity that resides within individuals, they may be better prepared to help international organizations derive value from their multicultural employees and from their multicultural workforces. Compared to monocultural managers who cross borders, multicultural employees straddle them, by belonging to two or more cultures. This cultural complexity within individuals is a resource and a source of challenge, both for the individual and for the organization. The findings from this article indicate it is possible for multicultural individuals and their organizations to benefit from their multicultural resources while seeking to cope with the challenges.

Notes

  1. 1

    We use the term multicultural instead of bicultural because bicultural individuals, who have internalized and identify with exactly two cultures, are a subset of the larger group of multiculturals.

  2. 2

    The paper-based pilot study was conducted with 46 multicultural individuals (mean age = 28.78 years, SD = 8.35 years, 23 male, 22 female, 1 missing gender) recruited from an MBA cross-cultural management class in Canada (39 participants), and snowball sample (7 participants). After individuals reported their cultures and demographic information, two open-ended items elicited descriptions of identity patterns: “How important are each of these cultures to your identity, or to how you see yourself? Why?” and “A bicultural person is anyone who has more than one culture, or who belongs to more than one culture. There are many ways to be bicultural. How would you describe your own form of biculturalism?” We used this pilot study to help us gain a better sense of how multicultural individuals make sense of their own identities, and include quotations in this manuscript to illustrate concepts.