Abstract
The number of immigrants worldwide has grown rapidly in recent years, and their integration poses challenges, such as cultural and language barriers, for organizations and societies. Securing and maintaining employment is a key challenge for immigrants, yet management research has devoted little attention to migration. We aim to contribute to the emerging literature on this topic by utilizing a multistudy approach with objective and time-lagged field data from 14,327 mail carriers nested in 737 units of a large Swiss logistics firm and an experimental audio vignette study with 262 participants from the United Kingdom. We investigate whether (in)congruence in terms of immigration background between employees and customers is linked to customer complaints. Controlling for service quality, we find that both congruence scenarios (both or neither migrants) are associated with fewer complaints, the latter suggesting that migrants identify with each other despite national and cultural differences. Results from the two incongruence scenarios show increased complaints. In Study 1, we find that units that receive more complaints experience higher rates of voluntary employee withdrawal behaviors (short-term absenteeism and voluntary turnover), highlighting how unfair customer complaints can hurt organizations twice, by increasing the risk of loss in both customers and employees. In Study 2, we replicate the immigrant identity effect at the individual level and find that social attraction mediates the (in)congruence–complaints link.
Keywords
The number of immigrants, 1 based on information on country of birth or citizenship, has grown rapidly worldwide since 2000, from 173 million to 244 million in 2015 (United Nations, 2015). While almost two-thirds of international migrants live in Europe or Asia, the United States is the country with the largest immigrant population (47 million immigrants represent about 19% of the overall population). In other Western economies, the percentage of immigrants is even larger. In Switzerland and the United Kingdom, the contexts for our studies, official data indicate that the population with foreign nationality accounted for about 25% and 9%, respectively, of the total resident population in 2018 (Office for National Statistics, 2019; Swiss Federal Statistical Office [SFSO], 2019b). If all migrants worldwide were to form a new country, it would be the fifth largest in the world by population (Connor, 2016). Immigration is a hot topic, as reflected by its relevance for elections around the globe (Agbafe, 2016; Sciurba & Tazziolo, 2018). Two prominent examples include Donald Trump’s border wall and “Brexit,” the decision of the United Kingdom to leave the European Union. Increased migration has been repeatedly cited as a leading cause for the result of the Brexit vote (Robertson, 2016).
Challenges of migration are not limited to the political arena. One of the most critical challenges for immigrants and their host countries across the world is gaining employment. It is no surprise that a majority of immigrants move to high-income countries (United Nations, 2015) with attractive employment options. Employment is even more relevant since it is often a precondition for immigration. The U.S. government, for example, has recently announced a plan to limit access for legal permanent residents to apply for citizenship if they have ever received support through social welfare programs (Ainsley, 2018). While employment is crucial, immigrants may face many employment barriers, such as supervisor and coworker stigma (Harrison, Harrison, & Shaffer, 2019), a lack of cultural understanding, and the devaluation of existing education and work experience (Akresh, 2008), to name a few. In addition to individual and organizational factors, employers are also concerned about customer reactions toward immigrant employees (Learning and Skills Council, 2006), particularly when they fill service roles. In a qualitative investigation, Tracey and Phillips (2016) found that a social enterprise in the East of England that initiated a supportive program for migrants in its community experienced spillover effects of migrant stigma. Parts of the community resented the organization and accused it of putting migrants first, highlighting an us-versus-them perception in a social identity sense (Tajfel & Turner, 1986). Such organizational stigma through stigma by association (Goffman, 1963; Kulik, Bainbridge, & Cregan, 2008) can lead to a number of detrimental effects, including loss of customers (Jensen, 2006). Interestingly and contrary to this perspective, organizations also see immigrant workers as positive and a legitimate tool to gain access to new market segments in the form of immigrant customers (Enchautegui, 2015; Fitzsimmons, Liao, & Thomas, 2017). Despite its relevance and differing perspectives on immigrant workers, current management research is relatively silent regarding challenges associated with including migrants into organizations (Dietz, 2010; Harrison et al., 2019; Jonsen, Maznevski, & Schneider, 2011). Bell, Kwesiga, and Berry (2010) even call immigrants “invisible men and women” because management researchers mostly ignore them. With this article, we join prior calls for more research on immigrants due to their importance to economies and organizations and aim to contribute to this developing field of research in three central ways.
First, in line with its recent surge in public attention and importance, we investigate congruence and incongruence between service employees and customers in terms of immigration background, a currently understudied phenomenon (Harrison et al., 2019; Jonsen et al., 2011). Existing research on demographic differences has heavily focused on race and gender through a relational demography lens (Jackson & Joshi, 2011). This is partly due to the fact that the majority of extant research has been conducted by scholars who are based in the United States and grounded in the U.S. historical, cultural, and institutional context (Jonsen et al., 2011; Roberson, Ryan, & Ragins, 2017). Yet, although it is certainly an extremely important characteristic in some contexts, race may not always serve as the most salient basis for in-group/out-group distinctions. This can be the case because of sociohistorical reasons (e.g., the tense history between Protestants and Catholics in northern Ireland or of Judaism in Germany) or because certain regions of the world have relatively low racial heterogeneity (e.g., Europe, northeast Asia, and Australia; cf. Joshi & Roh, 2007), which makes it much less likely to form social identity groups based on race. 2 A striking illustration of the latter is that in many countries and regions, census data on racial heterogeneity is not collected, in favor of other, more salient characteristics, like language or ethnicity (Alesina, Devleeschauwer, Easterly, Kurlat, & Wacziarg, 2003). Ultimately, context is critical to determine which characteristics take center stage in social and organizational life (Johns, 2006; Mor Barak, 2005). Consequently, we answer calls in the literature (e.g., Dietz, 2010; Harrison et al., 2019) and aim to expand the scope of demography research by investigating immigration background, a critically important characteristic in today’s global societies and business environment.
Second, we argue for the existence of a common migrant identity and investigate its relevance for organizations. Theories in the diversity realm predict that people who share common attributes, such as race, gender, or cultural background, develop higher levels of trust and treat each other better (Blau, 1977; Hogg, 2001; Tajfel & Turner, 1986). Relational demography and homophily research found that (in)congruence between individuals, and between an individual and the demographic composition of the work group, affects important outcomes, such as personal attraction, work group identification, conflict, closeness, leader–member exchange (LMX) relationships, and perceived effectiveness (Dwertmann & Boehm, 2016; Joshi, Liao, & Roh, 2011; Tsui & O’Reilly, 1989). Transferring this logic to the theme of immigration would lead us to predict that individuals of the same nationality identify with each other whereas immigrants from different countries do not. Contrary to this, we argue and find in a large-scale field study and an experiment that despite national and cultural differences, immigrants can form a common identity, potentially based on shared unique experiences, such as being a foreigner, navigating the immigration process, learning a new language, and adjusting to a new environment with unknown norms and traditions. As Harrison and colleagues (2019) put it convincingly, they all are “strangers in strained lands.”
Third, in addition to individual factors, such as cultural background, education in a foreign country, and language, and organizational factors, like supervisor and coworker stigma, employers are concerned about negative customer reactions when it comes to hiring migrants (Learning and Skills Council, 2006). At the same time, employers see immigrant workers as a legitimate tool to gain access to new international markets and immigrant customers (Enchautegui, 2015; Fitzsimmons et al., 2017). Lending indirect support for the prior perspective, research on women and racial minorities suggests that minorities are rated more harshly (Hekman, Aquino, Owens, Mitchell, Schilpzand, & Leavitt, 2010), particularly if they make mistakes (Brescoll, Dawson, & Uhlmann, 2010) or if raters have conflicting information about their credentials (Hodson, Dovidio, & Gaertner, 2002). We apply this logic to customer service interactions to test whether immigrants and nonimmigrants show the same pattern of complaints for service mistakes and unexpectedly discover that immigrants apply stricter performance expectations toward nonimmigrants. We detect a comparable effect for nonimmigrants in our first field study. In a final step of our model, we test if unfair treatment by customers based on incongruence relates to voluntary employee withdrawal behaviors and, thus, results in two harmful consequences for the organization: lower customer satisfaction and higher voluntary employee withdrawal in the form of short-term absenteeism and employee turnover.
We investigate these questions: does migration background matter, do migrants identify with each other, and does migration background influence customer service complaints that ultimately affect voluntary employee withdrawal behaviors, using 1) time-lagged data gathered from 14,327 mail carriers nested in 737 delivery units of a large Swiss logistics firm matched with official census data from Switzerland (see Figure 1) and 2) an experimental audio vignette study with 262 participants living in the UK.
Study 1 Research Model Without Control Variables
Theory and Hypotheses
The Influence of Immigration Background on Interactions
Social identity theory (Tajfel & Turner, 1986; Turner, 1985) predicts that individuals’ self-concept is largely influenced by their membership in social groups. Individuals classify themselves and others into groups based on various features, such as demographics and values. These classifications are the foundation for in-group/out-group differentiations that shape behavior and social interactions. As a general tendency, individuals will aim to portray their own group as superior to others to derive self-esteem from the group membership (Hogg, 2001). Therefore, they will treat in-group members more favorably and exhibit higher levels of trust, credibility, communication, and cooperation in interactions within their own group. Conversely, out-group members may face stereotyping and discrimination (Brewer & Brown, 1998). Existing relational demography research has shown positive congruence effects of different characteristics, such as race, age, gender, and disability, for a range of outcomes, including personal attraction (Tsui & O’Reilly, 1989), relationship quality (Dwertmann & Boehm, 2016), and organizational citizenship behavior (Chattopadhyay, 1999). Even though management research has mostly overlooked the phenomenon of immigration (Bell et al., 2010; Harrison et al., 2019), we argue that immigration background is a salient characteristic in interactions. Whereas it is not necessarily visually detectable (e.g., a White person moving to a majority-White country), accented language is a salient marker, and often the use of a single word (i.e., “hello”) is sufficient to detect a migration background (Purnell, Idsardi, & Baugh, 1999).
In the context of our studies, we expect that customers set different expectations and thresholds for complaints by considering their own migration status compared to that of their service worker (i.e. mail carrier). Customers will hold out-group members to higher standards. If, for example, customers without an immigration background receive a slightly damaged package, they should provide more leeway for in-group members (in this case, a mail carrier without an immigration background) but be more likely to complain when the delivery person is an out-group member. Customers will likely attribute the mistake to the mail carrier since alternative explanations are less credible if presented with an accent (Lev-Ari & Keysar, 2010) and by a person they trust less. Thus, in incongruence scenarios in which these situations occur more frequently, there should be more complaints compared to congruence scenarios. For example, the congruence scenario in which the number of immigrants is small on both the employee and customer sides should result in fewer complaints because individuals are from the same country (in Study 1, Switzerland) and cultural background (Chattopadhyay, George, & Lawrence, 2004; Hogg, 2001).
However, our arguments diverge from prior management research that has mainly focused on cross-cultural distinctions, cultural diversity, and national identity (Stahl, Maznevski, Voigt, & Jonsen, 2010) in the scenario in which the number of migrants is large on both the employee and customer sides. Tröster and van Knippenberg (2012), for example, investigate nationality dissimilarity of leaders and employees in multinational teams. One of their findings is that employees feel more comfortable to voice their opinions in congruent dyads in which both members share the same nationality. Applying the same logic and findings to situations in which both individuals have a migration background would lead us to predict low trust and relationship quality since migrants do not necessarily share the same nationality. In other words, what we refer to as a congruence scenario here would be seen as an incongruence scenario because customer service interactions between immigrants from different countries are much more common than interactions between immigrants from the same country (our partner organization employs workers of more than 140 different nationalities). Thus, based on relational demography and homophily logic (Blau, 1977; Tsui & O’Reilly, 1989), one might also expect a relatively large number of customer complaints in this scenario, based on the varying nationalities and cultural backgrounds.
Contrary to the prediction based on prior literature, we posit that interactions between immigrant employees and immigrant customers will resemble patterns expected for in-group members because both the employee and customer share a common immigrant identity. Our rationale is based on the fact that immigrants share specific experiences in the form of major life challenges (Shaffer, Harrison, Gilley, & Luk, 2001), many of them rather unique for this minority group. For example, immigrants have to manage the immigration process, which can be very complicated, expensive, and stressful. Immigrants also often have to learn a new language. Customs and norms can change dramatically from country to country. Many immigrants must find jobs, which is particularly challenging since their degrees are often devalued or simply not accepted at all (Akresh, 2008). Moving to a new country also often includes family resettlement. This means immigrants must find schools for their children and understand how the school system works. Migrants also have to create or find new social circles, often away from family. We argue that these joint experiences create a certain bond among immigrants from different national backgrounds. For example, immigrants from different countries can make sense of their new home country together. The specificity of these challenges is one reason why meet-up groups or online blogs for immigrants exist, often across nationalities. We argue that these common sensemaking experiences and the feeling of being different are a powerful source of identity.
Interactions with nonimmigrants further emphasize common feelings of otherness. Individuals use visual and auditory cues to identify out-group members, such as immigrants. Rakić, Steffens, and Mummendey (2011), for example, investigate the influence of looks and accents on ethnic categorization and find that both exhibit similar degrees of categorization. Interestingly, when combining both cues, the authors find a clear predominance of accents. Thus, even if migrants look similar to nonmigrants, they will be identified as migrants and out-group members through interaction. 3 For example, psychological research has shown that a single word (i.e., “hello”) can be enough to identify accents and cause in-group/out-group differentiations (Purnell et al., 1999). This ability to identify immigrants is important because customers and service employees (mail carriers) in our study have intermittent but repeated personal contact. Customers must provide signatures for certified deliveries and packages, for example, and we argue that these short face-to-face interactions are sufficient for both parties to assess whether the counterpart is a Swiss native or an immigrant and, therefore, an in- or out-group member (Hogg, 2001). We argue that migrants will develop a common migrant identity based on their joint experiences of cultural adaptation and otherness, which can overshadow some of the differences within this heterogeneous group (Harrison et al., 2019). Thus, based on social identity logic, we predict that congruence between the percentage of immigrant employees and customers will result in smaller numbers of complaints compared to incongruence scenarios.
Hypothesis 1: The more congruent the percentage of immigrant employees in delivery units and immigrant customers in the service area, the smaller the number of customer complaints for comparable levels of service quality.
On the basis of the preceding social identity arguments, and our rationale for immigration background as a significant identity marker, we predict that incongruence will lead to increased complaints.
Hypothesis 2: The more incongruent the percentage of immigrant employees in delivery units and immigrant customers in the service area, the larger the number of customer complaints for comparable levels of service quality.
Accounting for the Form of Incongruence: Asymmetric Effects Based on Status
Relational demography literature has highlighted the importance of status differences for members of different age groups (Chattopadhyay, 1999), races (Leslie, 2017), and disability statuses (Dwertmann & Boehm, 2016), among others. Status-based mechanisms are relevant for the current study since public attitudes toward immigrants are mostly negative (Gluszek & Dovidio, 2010), and immigrants (Chattopadhyay, George, et al., 2004) and individuals with non-native accents (Fuertes, Gottdiener, Martin, Gilbert, & Giles, 2012) typically hold lower status. Based on social identity theory (Hogg, 2001; Tajfel & Turner, 1986), high-status group members receive valuable resources, such as increased self-esteem, from their membership. It is thus in their best interest to preserve the status quo. In particular, if members of high-status groups perceive any threat to their status, they are likely to engage in a social competition strategy, characterized by a greater incidence of discriminatory behavior (Jetten, Spears, & Manstead, 1999). Because such a situation is likely to occur between immigrants and nonimmigrants (Esses, Dovidio, Jackson, & Armstrong, 2001), we expect an asymmetric effect: Service units with a higher percentage of immigrant employees than the customer base will receive more customer complaints than service units with a lower percentage of immigrant workers than the customer base.
Over the past years, immigration has generally been a salient topic. For our context, Skenderovic (2007: 155) points out that “the theme of immigration has remained in the center of the Swiss political stage” since the 1960s, and in 2014, Swiss voters decided to set strict immigration limits, partly due to the fear that foreigners would take jobs from Swiss nationals (Baghdjian & Schmieder, 2014). Native customers in our sample likely perceive threat and competition for resources by immigrants. Thus, we predict that the repeated exposure to foreign service workers will trigger perceptions of status threat for high-status group customers (Petriglieri, 2011). Consequently, high-status (native citizen) customers will likely complain if the mail carrier is an immigrant and makes a mistake. This form of social competition strategy will signal their membership in the high-status group and help them to preserve their valued status differential (Chattopadhyay, Tluchowska, & George, 2004). It should be more common in areas with higher percentages of immigrant employees.
A different picture should emerge in areas with higher percentages of migrant customers. Since individuals increase their self-esteem through identification with and membership in certain groups (Tajfel & Turner, 1986), membership in high-status groups is more valuable than membership in low-status groups. Whereas members of high-status groups should be motivated to keep their membership, members of low-status groups should alter their behavior depending on three factors: perceived permeability, legitimacy, and stability of the status hierarchy (Chattopadhyay, Tluchowska, et al., 2004). Permeability represents the “perceived potential to cross group boundaries” (Chattopadhyay, Tluchowska, et al., 2004: 189) and is of central importance because when perceived permeability is high, members of low-status groups will likely employ a social mobility strategy and aim to become part of the high-status group, regardless of the legitimacy and stability of the status hierarchy.
We argue that perceived permeability for immigrants is higher compared to race or gender, demographics that have been the focus of relational demography and diversity research. Immigration status can often be concealed at first since it is not visible, and although oftentimes challenging, it is possible to become a citizen of other countries, learn a foreign culture, assimilate to local customs and traditions, and become accepted by nonimmigrants. 4 Even though research contradicts this, immigrants also often believe that they can eliminate their foreign accent over time (Gluszek & Dovidio, 2010; Moyer, 2007). Thus, we predict that a significant percentage of immigrants will engage in a social mobility strategy, in which an individual’s efforts are targeted toward disassociating from a low-status group and gaining affiliation with a high-status group (Chattopadhyay, Tluchowska, et al., 2004). We think that this is particularly likely in Switzerland because the majority of migrants come from neighboring countries with similar languages, religion, norms, and customs (SFSO, 2019b), which should increase perceived permeability.
Following this logic, low-status customers (immigrants) will seek membership in the high-status group (native citizen) and complaining about a high-status individual could backfire. Nonmigrant service employees could, for example, talk to colleagues or friends about “that immigrant who is unfriendly/always complains.” Perhaps even worse, employees could communicate with the immigrants’ neighbors or members of their community, which would hurt the reputation of individual customers and lower the general acceptance of migrants in the community. Thus, on the basis of our arguments and meta-analytic results by Bettencourt, Dorr, Charlton, and Hume (2001), who found general support for the social mobility strategy based on perceived permeability, we predict that many immigrants will be motivated to treat high-status service workers well and complain less when encountering minor mistakes. This should lead to fewer complaints in areas in which the percentage of immigrant customers exceeds the percentage of immigrant employees.
Hypothesis 3: The effect of incongruence on customer complaints will be stronger if the percentage of immigrant employees is higher than the percentage of immigrant customers compared to when the percentage of immigrant employees is lower than the percentage of immigrant customers.
The Effect of Customer Complaints on Voluntary Employee Withdrawal
As the final step in our model, we propose that increased customer complaints for equal service quality will lead to higher voluntary employee withdrawal behaviors in the form of short-term absenteeism (Harrison & Martocchio, 1998) and voluntary turnover (Berry, Lelchook, & Clark, 2012). The underlying principle of the service industry is that organizations or their representatives provide customers with some sort of service. If the service is delivered at the stated and expected quality, customers pay the prearranged price. Many organizations have adopted a “the-customer-is-always-right” motto, which essentially provides customers with more power and puts pressure on employees to deliver levels of service which match the customers’ expected levels of quality (Grandey, Dickter, & Sin, 2004). This exchange is assumed to follow commonly accepted norms (Wang, Liao, Zhan, & Shi, 2011), such as quality assessments being made fairly. However, if customer complaints (and feedback) are not based on actual performance, but are rather consciously or unconsciously influenced by demographic group membership (e.g., being a minority-group member; Hekman et al., 2010) or incongruence between the customer and service employee, customers breach fairness norms (cf. Poon, 2004). Employees will likely notice and question such unfair treatment, particularly in repeated interactions (which is the case in our sample), and it may cause withdrawal from the organization (Li & Zhou, 2013).
Unfair complaints can also be interpreted as a form of rejection, something that should be particularly harmful for immigrants who employ a social mobility strategy and want to be accepted by the high-status group members as equals (Chattopadhyay, Tluchowska, et al., 2004). As such, injustice experiences and rejection from the group whose membership they seek represent threats to immigrants’ self-interest and social identity (Cropanzano & Rupp, 2003). Prior research has even found that social rejections stimulate areas of the brain that are similarly stimulated by the experience of physical pain (Eisenberger, Lieberman, & Williams, 2003). We propose that for service employees who experience such rejection, the threat to their social identity and resulting stress due to unfair treatment will motivate them to escape these situations and exhibit higher short-term absenteeism and voluntary turnover behavior as proximal and distal forms of voluntary employee withdrawal.
Hypothesis 4: The number of customer complaints will be positively related to short-term unit absenteeism rates.
Hypothesis 5: The number of customer complaints will be positively related to voluntary unit-level turnover rates.
Study 1
Method
Sample
Data to test this study’s hypotheses were collected from two sources. First, we were able to access data from a Swiss logistics company on all of their customer delivery units, which operate throughout the entire country due to the quasimonopoly position of the company. The delivery units have daily service tasks that involve regular interaction with customers. They deliver several hundreds of parcels and packages each day walking, biking, by car, or by transporter. While they are not required to get signatures for every delivery (only certified/insured parcels and packages), most mail carriers have a fixed route and area that they service for several years, if not decades. Consequently, they are typically well known in the community they serve.
We received data from company records (demographic information as well as information on absenteeism, turnover rates, customer complaints, and service quality) and the annual employee survey (self-assessed psychological and physical health). All information was gathered at the unit level. Overall, data from 14,327 employees nested in 1,108 units were available. Second, we accessed data from the SFSO (www.bfs.admin.ch/) on the percentage of migrants in each postal code area in Switzerland. We were able to match these official data with the company information by identifying the postal code that each delivery unit served. The final sample size was 737 units because we removed units that had support functions without direct customer interaction (n = 176) and excluded units that had missing values on study variables (n = 195).
In addition to drawing from two data sources, we also utilize data from two separate time points. Whereas all predictor and mediator variables (percentage of migrant employees, percentage of migrant customers, customer complaints, and control variables) were measured at Time 1 (T1), information on voluntary employee withdrawal was collected 12 months later, at Time 2 (T2).
Delivery units had 11 majority male (65%) members on average. Average employee age was 46 years and average company tenure 21 years. The majority of employees (90%; range 28%–100%) and customers (79%; range 27%–98%) were Swiss citizens. Only a small minority, 0.5%, were citizens of a non-European country. The majority of employees (96.5%) reported that they had knowledge of at least one of the official Swiss languages. The percentage of language proficiency in units correlates (r = –.36) with the percentage of migrants.
Context
Due to the national context and specific immigration laws in Switzerland, we wish to provide additional information about our context. In this study, we measure immigration background by utilizing information on nationality (i.e., having Swiss citizenship). Importantly, Switzerland has very restrictive regulations for obtaining citizenship. Contrary to regulations in other countries (e.g., the United States), in Switzerland, children of foreign parents are not automatically granted citizenship if they are born on Swiss soil (this is true for multiple generations). In order to be eligible for Swiss citizenship, individuals need to have been residents of Switzerland for at least 10 years (years lived in Switzerland between ages 10 and 20 count double towards this requirement, 2 to 10 years in the same municipality or canton, which is equivalent to a state in the United States) or apply for citizenship through the process of facilitated naturalization. Migrants can apply only if they did not receive social support (i.e., financial) in the past 3 years. Facilitated naturalization is possible for individuals who have been married to a Swiss citizen for at least 3 years and who have lived in Switzerland for at least 5 years or for children of at least one Swiss citizen. Upon application, the State Office for Migration “examines whether applicants are integrated in the Swiss way of life, are familiar with Swiss customs and traditions, comply with the Swiss rule of law, and do not endanger Switzerland’s internal or external security” (State Secretariat for Migration, 2013). This includes the assessment of written and verbal language skills and rigorous exams about the Swiss political system, history, and customs. Any conflict with the law prevents immigration. However, a positive decision by the State Secretariat for Migration is only the first step to naturalization. The cantons and residents of the communities where the individual lives must also agree. 5 Beyond these criteria, additional requirements, such as “being well integrated,” exist. Immigrants are not allowed to vote, cannot own real estate, and can face harsh consequences when getting in conflict with the law. They are, however, required to hold a steady job and pay taxes. From 2016 to 2018, about 2% of immigrants gained citizenship annually (SFSO, 2019a). Thus, in our sample, individuals would be considered nonmigrants only if they have resided for a significant time in Switzerland, are integrated in the Swiss way of life, and have gained approval from residents of their local community. Migration has been a contentious topic in Switzerland over the past years, including votes to restrict migration (Baghdjian & Schmieder, 2014). The conservative right-wing party, SVP, won the majority of votes in all elections for the past 20 years. It is well known for anti-immigration campaigns.
Switzerland has four official languages. The majority of the population speaks German (63.5%), followed by French (22.5%), Italian (8.1%), and Romansh (0.5%). Other languages make up the remaining percentage. Importantly, the major languages are spoken in specific dialects (e.g., Swiss-German) that differ substantially from the respective languages (Zinggeler, 2005) and are usually used in all situations in daily life. These dialects are often challenging to understand and very hard for foreigners to speak without an accent (e.g., German or French natives in Switzerland). As a result, even short verbal interactions are usually sufficient to detect whether someone grew up in Switzerland or in a neighboring country, such as Germany, Austria, or France. We highlight this because we argue, in line with prior research in language psychology, that very short verbal interactions (e.g., obtaining a signature for a package) are sufficient to trigger social identity and similarity-attraction effects (Purnell et al., 1999; Rakić et al., 2011).
Measures
Percentage of immigrant employees (T1)
The information on citizenship was gathered from the company records at the first time point. Our measure represents the percentage of immigrant employees (i.e., employees without Swiss citizenship) in each unit (M = 10.02, SD = 11.34).
Percentage of immigrant customers (T1)
Customer information was collected from the official records of the SFSO. We identified the percentage of immigrants (individuals without Swiss citizenship) in each postal code area in Switzerland at T1 (M = 20.82, SD = 10.54).
Customer complaints (T1)
The amount of customer complaints is an important quality indicator for our partner company. We received information on the percentage of customer complaints per 10,000 deliveries, which were recorded by the quality management department of the company. The complaints captured represent a wide array of potential reasons, such as late deliveries, damaged mail, rudeness of the mail carrier, and leaving a package in front of the door without receiving a signature. The measure captures the last 12 months at T1 for each delivery unit (M = 2.04, SD = 2.93). Since this variable was right-skewed, we log-transformed it.
Short-term absenteeism (T2)
We gained information on unit absenteeism from company records. The variable captures the percentage of short-term absenteeism (less than 6 days) per business unit, a better proxy for absenteeism due to motivational reasons compared to long-term absences (Harrison & Martocchio, 1998), over the last 12 months (T2; M = .41, SD = .01). As this variable was right-skewed, we log-transformed it.
Voluntary employee turnover rate (T2)
We used information in the official company records to assess turnover. Our employee turnover rate variable represents the average percentage of voluntary company turnover in each unit over the last 12 months (T2; M = 2.82, SD = 5.03). As this variable was right-skewed, we log-transformed it.
Controls
We control for 16 factors that might affect customer complaints and withdrawal behavior. First, we control for the T1 turnover rate, as high levels of turnover could affect both service quality and the T2 withdrawal rates. Second, we control for average unit size, since different social and psychological processes (e.g., affect contagion processes) that affect employee withdrawal might occur in units of different sizes. Third, we include the age mean of employees in all units, as some research shows that older employees exhibit lower turnover intentions (Ng & Feldman, 2008). We also control for age and gender diversity (operationalized as the standard deviation for age and the Blau index for gender), since diversity may affect team processes and outcomes. Fourth, we control for the percentage of female unit members, as past research shows that women often exhibit higher turnover rates than men (Weisberg & Kirschenbaum, 1993). Fifth, we include the T1 absenteeism rate (percentage of work hours missed by unit members due to short term absence), as this might predict absenteeism and turnover at T2 (Harrison & Martocchio, 1998). Units with high absence rates may also have difficulties keeping up service quality. Sixth, we include one-item measures for physiological and psychological health from the employee survey at T1 (“In general, my physiological/psychological health is . . .”; 1 = bad to 5 = excellent), as health predicts turnover (Farrell & Stamm, 1988). Seventh, we control for pay satisfaction (“I receive a satisfying payment for my task”; 1 = totally disagree to 5 = fully agree), as prior research has shown that compensation structure can drive service quality and voluntary turnover (Carnahan, Agarwal, & Campbell, 2012). Eight, we include the number of customers in each district, since larger areas might be more difficult to serve and, therefore, have a higher base rate of customer complaints. Relatedly, we also control for area (i.e., rural vs. city), as it might be likelier to make mistakes in larger apartment buildings with lots of mailboxes (i.e., more complexity). City deliveries may also be characterized by less direct customer contact and more turnover within the customer base compared to rural areas with mostly single-family houses. Ninth, we use a measure of service quality that was provided by the company to ensure that the customer complaints are not merely a product of poor service. For this measure, the company regularly mails test parcels to customers and monitors their appropriate delivery through digital processing. Units receive a value of 1 if the service quality in terms of speed and accuracy is satisfactory and 0 if it is not. Tenth, we control for the productivity of each delivery unit at T1, by taking a measure provided by the company that divided the parcel weight in kilograms by delivery time that can be followed through the digital infrastructure of the company. Eleventh, we include the percentage of female customers since they may be more likely to identify with immigrants. Finally, we control for the percentage of employees with citizenships from non-European countries to account for cultural distance and a higher likelihood of different ethnicities.
Data Analysis Strategy
All data were analyzed at the unit level of analysis in two steps. First, we applied polynomial and response surface techniques (Edwards, 1994) to assess the proposed incongruence effects in Hypotheses 1 to 3. This technique enables testing of asymmetrical demographic congruence (or incongruence) effects and offers the ability to detect both linear and nonlinear effects.
A polynomial regression model includes the main effects of both factors (percentage of immigrant employees [PIE] and percentage of immigrant customers [PIC]), the interactive term of both factors (PIE × PIC), and the quadratic term of both measures. This results in the following regression equation:
We used a stepwise approach for entering the variables with adding the control variables in Step 1, followed by the main effects in Step 2, concluding with the interactive and squared terms in Step 3. We standardized all variables. To assess the nature of our effects, we inspected both the slopes and curvatures along the congruence line (PIE = PIC) and the incongruence line (PIE = –PIC), which resulted in four different test values: (a) the slope along the congruence line (PIE = PIC), (b) the curvature along the congruence line (PIE = PIC), (c) the slope along the incongruence line (PIE = –PIC), and (d) the curvature along the incongruence line (PIE = –PIC). For Hypotheses 1 and 2, we would expect significant differences of the curvature on the congruence line (b) and the incongruence line (d), while Hypothesis 3 would suggest slope differences only for the incongruence scenarios (c).
In a second step, we used a conventional ordinary least squares (OLS) regression analysis to inspect the effect of perceived customer complaints at T1 on the short-term absenteeism rate at T2 and the turnover rate at T2. We included all control variables.
Results
Correlations and Descriptive Statistics
Table 1 shows the zero-order correlations and descriptives for study variables. The percentage of immigrant employees is not related to the amount of customer complaints (r = –.032, p = .384), nor is the percentage of immigrant customers (r = –.056, p = .126). The percentage of customer complaints is significantly related to employee turnover at T2 (r = .127, p = .001) and marginally related to short-term absenteeism behavior (r = .072, p = .051).
Study 1 Correlations and Descriptive Statistics
Note: N = 737 service units.
p < .05
p < .01
p < .001 (two sided)
Polynomial Regression With Response Surface Analysis and OLS Regression
Table 2 shows the results of the polynomial regression analyses with customer complaints as the dependent variable. In Hypothesis 1, we proposed that high congruence in terms of percentage of immigrant employees and immigrant customers should relate to fewer customer complaints. As displayed in Model 3 of Table 2, the addition of the second-order polynomial terms significantly increased model fit (∆R² = .034, 95% confidence interval [CI] = [.001, .060], p < .001) and the interactive term of PIE and PIC was significant (B = −.149, CI = [−.257, −.042]; t = −2.723, p = .007). Additionally, the curvature along the congruence line (B = −.152, CI = [−.391, .078]; t = −4.269, p < .001) was negative and significant. Figure 2 illustrates the shape of this curvature. In line with Hypothesis 1, the number of customer complaints is decreasing for conditions of low percentage of immigrant employees and low percentage of immigrant customers, and for conditions of high percentage of immigrant employees and high percentage of immigrant customers. This is indicated by a change of complaints across the congruence line (PIE = PIC) at both ends running from the front to the rear corner. Interestingly, the significant curvature slope also indicates that this effect is not linear but inverted-u shaped.
Study 1 Regression Results on Customer Complaints
Note: N = 737 service units.
p < .05
p < .01
p < .001 (two sided)
Study 1 Surface-Level Effects of Employee–Customer Immigrant Status Incongruence Versus Congruence on the Number of Customer Complaints
In Hypothesis 2, we posited that incongruence in terms of employee and customer immigrant percentages is linked with higher levels of customer complaints. In line with this hypothesis, the curvature along the incongruence line was significant and positive (B = .132, CI = [.123, .029]; t = 1.981, p = .048). Figure 2 shows that the number of customer complaints increase for both high–low and low–high combinations of percentage of immigrant employees and customers, shown by a change of complaints across the incongruence line at both ends (PIE = –PIC) running from the left corner to the right corner of the plane. This incongruence effect is not linear but u shaped.
To test Hypothesis 3, we inspected the slope that connects the two incongruence scenarios (high–low, low–high). Following the argumentation of Hypothesis 3, the incongruence slope should have been significant to show a difference between the two incongruence scenarios. Contrary to Hypothesis 3, the slope in Table 2 was not significant (B = .010, t = .111, p = .911). In line with this, inspection of Figure 2 indicates that both incongruence scenarios show similar levels of customer complaints.
Finally, in Hypotheses 4 and 5, we predicted that the number of customer complaints related positively to short-term absenteeism behavior as well as voluntary turnover within the delivery units in following year. The results, as displayed in Tables 3 and 4, mainly support our prediction. For the relationship between customer complaints (T1) and short-term absenteeism (T2), we see a positive effect that is marginally significant when using two-sided testing (B = .052, CI = [–.004, .108]; t = 1.823, p = .069). The number of customer complaints is significantly related to employee turnover in the following 12 months (B = .050, CI = [.000, .100]; t = 1.969, p = .050).
Study 1 Regression Results for Employee Short-Term Absenteeism
Note: N = 737 service units. The short-term absenteeism measure is log-transformed.
p < .05
p < .01
p < .001 (two sided)
Study 1 Regression Results for Voluntary Employee Turnover Rate
Note: N = 737 service units. The voluntary employee turnover measure is log-transformed.
p < .05
p < .01
p < .001 (two sided).
Robustness Tests
We ran additional tests to inspect the robustness of our results. 6 First, models without control variables (see OS 1 in the online supplement) and only with potent control variables (those that correlated significantly with the respective dependent variable; see OS 2 in the online supplement) to account for a potential biasing effect (Becker, 2005) revealed that all central effects remained significant. The short-term absenteeism models even showed significant effects at the .05 level (two sided). Second, we calculated both models with non-log-transformed outcome variables, which led to similar results for all main findings (see OS 3 in the online supplement). Third, we utilized our longitudinal data to run an autoregressive structural equation model based on the application of Volmer, Niessen, Spurk, Linz, and Abele (2011) and found, in line with prior results, a positive effect between T1 complaints and T2 turnover (B = .072, t = 2.100, p = .036) as well as a marginally significant effect with T2 absenteeism (B = .053, t = 1.874, p = .061).
In a fourth step, we tested the interactive effect of the percentage of immigrant employees and immigrant customers on customer complaints in a time-lagged model, with T2 complaints as the outcome, while controlling for complaints at T1. Interestingly, this time-lagged model explained more variance while still indicating a significant interaction effect of the percentage of immigrant workers and customers (see OS 4 in the online supplement).
Fifth, we inspected if the customer-complaints measure taps actual bias. For this purpose, we specified two three-way interaction models in which we tested if either the service quality measure or the unit productivity measure moderates the observed interaction between percentage of immigrant employees and customers on customer complaints. Neither three-way interaction term was significant, indicating that in line with our argumentation, the observed immigrant congruence and incongruence effects on customer complaints are not dependent on unit performance (see OS 5 in the online supplement).
Sixth, we specified a three-way interaction in which percentages of immigrant employees and immigrant customers further interacted with the percentage of non-European employees. We used non-European as a proxy variable because this group is more likely to be non-White and might thus be prone to racial discrimination beyond immigrant identity. The interaction was nonsignificant, indicating that our findings are consistent for different immigrant groups (see OS 6 in the online supplement).
Discussion
Apart from Hypothesis 3, Study 1 results are in line with our core predictions (Hypotheses 1, 2, 4, and 5) and suggest the existence of a common immigrant identity. However, Study 1 has two major shortcomings. First, we were not able to directly test the theorized social identity mechanism. Second, while most of our arguments regarding the processes underlying these effects focus on the individual and dyadic levels, we utilized aggregated unit-level data. This results in the potential of an ecological fallacy since links between variables may vary across levels (Ostroff, 1993; Ostroff & Harrison, 1999; Robinson, 1950). To alleviate these concerns and test the robustness of the central part of our model—the link between (in)congruence in terms of immigration background and customer complaints—we conducted an experimental study at the individual level.
Study 2
Theory and Hypotheses
In line with Study 1 results, but adjusted to the individual level, we expect nonimmigrant customers to show a lower complaint likelihood when interacting with nonimmigrant service workers and immigrant customers to complain less when interacting with immigrant workers.
The Mediating Effect of Social Attraction
In Study 1, we argued for fewer complaints in congruence scenarios based on social identity effects (Tajfel & Turner, 1986; Turner, 1985). Due to the aggregate nature of the data, we were unable to test this mechanism directly. Based on the similarity-attraction paradigm (Byrne, 1971), people prefer to interact with similar others since they receive more affirmative feedback from them (Hinds, Carley, Krackhardt, & Wholey, 2000). Immigrants, for example, might discuss their adjustment to the culture of their host country or the other major life challenges associated with immigration. Such better understanding and shared experiences should increase social attraction and interpersonal liking in interactions between in-group members. They will treat each other more favorably in interactions and exhibit higher levels of trust, credibility, communication, and cooperation. Explanations for service errors are more credible from individuals who are liked and trusted and do not have an accent (Lev-Ari & Keysar, 2010). Thus, comparable interactions and levels of service quality between members of the same social group (migrants vs. nonmigrants) should result in higher levels of social attraction and fewer customer complaints for comparable service quality. We thus propose the following additional hypotheses (see Figure 3).
Hypothesis 6: Social attraction is higher in congruence scenarios in which two immigrants or two nonimmigrants interact compared to incongruence scenarios.
Hypothesis 7: Higher levels of social attraction lead to fewer customer complaints for comparable levels of service quality.
Hypothesis 8: Social attraction mediates the link between (in)congruence in immigration background and the number of customer complaints for comparable levels of service quality.
Study 2 Research Model
Methods
Sample and Procedure
We used audio vignettes in an experimental design to test Hypotheses 1 and 2 and the new Hypotheses 6, 7, and 8 at the individual level with 333 participants. We chose the U.K. online platform Prolific (www.prolific.ac) for multiple reasons. The United Kingdom is comparable to Switzerland in key aspects (e.g., high recent influx of immigrants, contention about immigration) but also different (e.g., lower percentage of migrants in the population, less homogeneous immigrant population, one common language) and thus offers a chance to test the generalizability and robustness of our results. Prolific participants have been shown to be more diverse, naive, and less dishonest compared to other online platforms, such as MTurk (Peer, Brandimarte, Samat, & Acquisti, 2017). Participants received a payment of £0.70. Of the total sample, 16 participants failed the manipulation check (“Please state the last word of the audio vignette”), and we excluded 55 participants because their self-reported nationality differed from Prolific’s system information that we used for our random assignment and/or they had dual citizenship. The final sample size was 262 participants. Immigrants in our sample had a broad range of national backgrounds (European, 79.60%; African, 6.60%; Asian, 10.30%; American/Australian, 12.50%), enabling us to test an overall immigrant identity effect.
To test whether migration background is an important identity basis and if language serves as a mechanism to identify in-group and out-group members, we designed a 2 (native vs. immigrant accent of the mail carrier) by 2 (native vs. immigrant customer [participant]) experiment. We recruited equal numbers of participants with U.K. (native condition) and non-U.K. citizenship (migrant condition) to mirror the conditions from Study 1. First, participants read a scenario describing a delivery error (i.e., “An important package arrives late and damaged and you ask your [English/Polish/Italian] mail carrier why that happened”). Next, we randomly assigned participants from both groups (native vs. immigrant) to listen to one of four voice recordings. All recordings were equivalent in terms of content (“I am not responsible for the delayed delivery. I got the package late from the delivery center and this is where the package must have been damaged. I try to deliver all packages and mail on time and in good condition. If you don’t believe me, just use our hotline to complain. I have to continue my route. Goodbye”) and length and were spoken by men ages 30 to 40. Two professional voice actors each read them in an English accent (native conditions) or in a Polish or Italian accent (immigrant condition). We chose two different English speakers and accents because we wanted to lower a potential bias against a specific voice/accent. For the migrant accents, we choose relatively lower-status (Poland) and higher-status (Italy) migrants from Europe to mirror a similar situation as in Study 1 (where also the vast majority of migrants came from European countries) and keep the effect of ethnic differences constant (i.e., we assume that participants would imagine a White speaker in both scenarios). Polish and Italian citizens constituted the largest migrant populations in the United Kingdom that match these criteria for the most recent data from 2017 (Migration Observatory, 2018).
Following the respective voice recording, participants answered a number of questions. First, they responded to three social attraction items from McCroskey and McCain (1974) (“I would like to have a friendly chat with the postman”; “The postman and I could establish a personal friendship with each other”; “The postman would fit into my circle of friends”; 1 = no to 5 = yes; α = 84). Next, they indicated how likely it is that they would file a complaint (“How likely would you be to file a complaint? Please slide the bar on a 0-to-100 scale to indicate your response”).
Results and Discussion
Table 5 shows descriptives and correlations of the focal variables. We used the process macro by Hayes (2017) to analyze our data. In line with our expectations, the interaction between mail carrier and customer/participant immigration background had a significant effect on social attractiveness of the employee (B = .628, t = 2.764, p = .006; F = 7.639, ΔR2 = .029). This effect is shown in Figure 4. Simple slope tests revealed that the slope for immigrant participants was significant and positive (B = .523, t = 2.983, p = .003), whereas the slope for the native participants was negative but not significant (B = –.105, t = –.726, p = .468). These results provide partial support for Hypothesis 6. Furthermore, as expected, social attractiveness of the mail carrier is negatively linked to customer complaints (B = −5.510, t = −3.260, p = .001; F = 150.209 ΔR2 = .366), providing support for Hypothesis 7. This indicates a mediated conditional-indirect effect. We directly tested this moderated mediation via bootstrapping procedures. Results indicated that the indirect effect of mail carrier accent on customer complaints via social attractiveness is significant if the customers/participants are immigrants (B = −2.883, 95% bias-corrected confidence interval [BCCI] = −5.972, −.698]) but not if they are natives (B = .578, BCCI = [−1.008, 2.206]). These results provide partial support for Hypothesis 8.
Study 2 Descriptives and Correlations
Note: N = 262 individuals. *p <. 05.
Study 2 Interaction Effect Between Accent Manipulation and Participant Citizenship on Social Attractiveness of the Service Employee
We further inspected the formation of a common migrant identity by testing if the Polish-speaking mail carrier is still rated more favorably if we exclude all participants from Slavic-language and culturally similar countries and by testing if the Italian-speaking mail carrier is still rated more favorably if we exclude all participants from southern European Romanic-language countries. Applying two-sided testing, both analyses showed marginally significant effects (Slavic accent: B = .585, t = 1.943, p = .054, F = 3.774, ΔR2 = .021; Italian accent: B = .537, t = 1.897, p = .059, F = 3.600, ΔR2 = .019), indicating that the results are driven not only by immigrants with similar backgrounds.
Discussion
Individual-level results are mostly in line with Study 1 and speak against a potential ecological fallacy for our Study 1 results (Ostroff, 1993). They also provide an extension by identifying social attraction as a mediating mechanism between (in)congruence in terms of immigration background and customer complaints for comparable service quality. This is in line with our social identity and similarity-attraction logic and provides an indication for a common immigrant identity in a different national context. We discuss all findings and implications next.
General Discussion
In this article, we investigate immigration background in organizational settings, a characteristic that has not received attention commensurate with its significance to theory and practice alike (Harrison et al., 2019). Analyses of our time-lagged data from 14,327 employees in 737 delivery units in Study 1 indicate that incongruence compared to congruence between the percentage of immigrant employees in mail delivery units and customers leads to more customer complaints at comparable levels of service quality. Interestingly, we find that this effect works in both congruence directions (i.e., low–low and high–high percentage of immigrant employees and customers), indicating that migrants identify with each other. Contrary to our predictions, incongruence between employee and customer generally leads to more complaints. This indicates that migrants express biases against nonmigrants, as well, which is contrary to our social mobility arguments based on Chattopadhyay, Tluchowska, et al. (2004). Finally, our results show that workers in units that receive increased complaints withdraw from the organization by exhibiting higher short-term absenteeism and voluntary turnover rates (Cropanzano & Rupp, 2003; Li & Zhou, 2013). This finding shows that biased customer satisfaction ratings can harm organizations twice (i.e., lower customer satisfaction and higher voluntary employee withdrawal) and calls into question the management practice of linking customer satisfaction to employee compensation and promotion. The results of our experimental Study 2 with 262 participants support the existence of a common migrant identity. Immigrants exhibit higher levels of social attraction and a lower likelihood to complain when interacting with (listening to) service immigrant workers. This study also emphasizes the importance of language or accent as a marker of immigration background.
Theoretical Contributions
Our results add to the literature in several important ways. First, our findings regarding the effects of congruence and incongruence between customers and employees in terms of immigration background highlight the importance of immigration as a topic for management research. Visible characteristics, such as gender and race, are the focus of relational demography and diversity research (Jackson & Joshi, 2011), but other demographics, like migration background and religion, appear overlooked. This is problematic because such characteristics take center stage in many regions of the world (Jonsen et al., 2011; Mor Barak, 2005). As Joshi and Roh (2007: 42) state (information in Italics added), The choice of diversity attributes and outcome variables would vary in other [than U.S.] national contexts. . . . With the growing internationalization of the management research community, taking into account the unique aspects and outcomes of diversity in non-US settings may change the nature of the debate around the pros and cons of diversity that has evolved in the US.
Based on existing relational demography research that has mostly utilized U.S.-based samples, we could expect race to be the most salient characteristic for social identity processes. We might, for example, predict Black French men to become out-group members in a group of White French men, Black French men, and White German men. Yet, particularly based on our Study 2 results, it seems possible that the common language of the French men would lead the White German men to become the out-group. This is in line with psychology and cognition research from the United States and Europe that emphasizes the importance of language and shows that non-native accents can be a stronger negative cue than visible characteristics, such as race (Rakić et al., 2011). Huang, Frideger, and Pearce (2013), for example, report that non-native speakers are significantly less likely to be recommended for managerial positions regardless of whether they are White or Asian. Race, communication, and collaborative skill were nonsignificant predictors. Particularly relevant for the context of our studies, Fox, Moroşanu, and Szilassy (2012) conclude in their analysis of Eastern European immigration to the United Kingdom that shared Whiteness offers few protections against discrimination. While our results from Study 2 do not indicate a general bias of U.K. citizens against migrants (maybe due to social desirability effects, which are more likely in a controlled experiment compared to a real-life field setting), our findings from Study 1 point in the same direction and underscore the importance of better integrating immigration background into the literature. It is important to identify its significance not only for customer service interactions but also team processes and leader–employee interactions. This will allow research to better reflect the globalized business world and result in a much-needed, more diverse approach to studying individual differences in workplaces (Jonsen et al., 2011; Roberson et al., 2017).
Second, we find a general congruence effect in Study 1 such that teams with many and few immigrants that work in areas with commensurate immigrant customers receive similarly low numbers of complaints. This finding was replicated at the individual level in Study 2, where migrants expressed higher social attraction to and a lower propensity to complain about other migrants. These findings indicate the existence of a common, overarching migrant identity beyond the national differences of immigrants (Dietz, 2010). Such a joint identity could not have been detected by the majority of existing management research, which has focused on and operationalized variables as cultural and national identity (Stahl et al., 2010). The results of studies of this type would have led us to predict that individuals from different countries engage in discriminatory behavior against each other, resulting in increased customer complaints. Contrary to this, we have argued for and found support that immigration background can act as a powerful source of social attraction and identity, potentially based on common experiences of being different and challenges such as the navigation of the legal aspects of the immigration process and a foreign language and culture (Shaffer et al., 2001). The effects seem to be strong enough to overshadow some differences within the group of immigrants (Harrison et al., 2019). We are not aware of studies that argue for and find support for an overarching migrant identity. However, a qualitative investigation by Massey and Sanchez (2010) indicates that immigrants from Latin and South America developed a common Hispanic identity despite different races, cultures, and languages (Spanish, Portuguese, and indigenous languages). Our findings go beyond this and indicate a superordinate immigrant identity (Dietz, 2010), which is particularly relevant given today’s globalized economy, in which working and living abroad are more common than ever. One important question that arises from our results is under which conditions such an overarching migrant identity develops. We would speculate that in some situations, particularly if subgroups are salient, lower-level identities come to the fore (e.g., common European or Asian identity). This interplay between different abstraction levels of identity formation is interesting from a theoretical and practical standpoint because it may offer the chance for organizational interventions to form more coherent identities among employees. Future quantitative and qualitative research should identify these conditions.
Finally, there are different views on the effects of immigration for organizations and societies. Some employers express concerns about customer reactions toward migrant workers when it comes to hiring migrants (Learning and Skills Council, 2006), while others may hope to gain access to international markets and new market segments in the form of immigrant customers by better understanding their preferences and needs (Enchautegui, 2015; Fitzsimmons et al., 2017). Complementing research on women and racial minorities (Hekman et al., 2010), we find in Study 1 that nonimmigrant customers are more critical toward immigrant workers in customer service interactions. However, we find in both studies that immigrants appear to apply stricter performance expectations when interacting with incongruent others (nonimmigrants), as well. This pattern is different from findings on gender and race that found discrimination only against traditional low-status groups (Brescoll et al., 2010; Hodson et al., 2002). The inconsistency between the findings from the two studies for nonimmigrants might be due to social desirability effects in Study 2. We can only speculate about explanations for the immigrant finding, but it might be the case that some migrants have little hope of being accepted and gaining access to the higher-status group. Thus, they perceive permeability as low and do not engage in a social mobility strategy (Chattopadhyay, Tluchowska, et al., 2004). Another possibility is that migrants perceive permeability as high but do not think of interactions with mail carriers as important enough to modify their behavior such that they give them more leeway. It would be important to investigate whether migrants see other interactions with officials in high-power positions (e.g., police officers) as important enough to adjust their behavior. The fact that the result is consistent across both studies increases confidence in its validity, and we suggest that future research investigate this direction. Importantly, the discriminatory pattern of customer complaints we discover is relevant for organizations, since we find that unfair treatment by customers increases voluntary employee withdrawal and, thus, results in multiple harmful consequences for the organization (i.e., potential loss of customers and employees).
Practical Contributions
The recent influx of refugees and migrants poses many critical integration challenges to many countries and their economies, and the representation of right-wing parties has increased in most countries (Aisch, Pearce, & Rousseau, 2016). Integrating immigrants in the labor market is a main challenge. Migrants struggle to find employment, and if they do, they often find low-paying jobs in the low-skill sector (Migration Policy Institute, 2014). Organizations operating in this uncertain environment may be concerned that hiring migrants may lead to negative consequences, such as increased costs and a worse public image (Tracey & Phillips, 2016). Yet, we do not find increased customer complaints for units with a higher percentage of immigrants. To the contrary, these units may fare better in areas with a high percentage of migrant customers. The idea of matching employees to the customer base is not new, but we want to acknowledge that such a “matching strategy” would lead to segmenting workers on the basis of immigration background and could potentially lead to poorer integration of immigrants, not to mention engendering possible legal consequences. Thus, we caution the interpretation of our findings as a rationale for segmenting workers according to community context and instead call for greater sensitization within organizations regarding the potential detrimental effects of incongruence. Organizations should consider preventive measures, such as deliberate staffing, to counter these effects by ensuring contact between immigrant and nonimmigrant workers to foster sensitivity.
Our finding that customers are more likely to complain for comparable service quality if the employee is different from them calls into question the managerial practice of linking benefits to customer satisfaction. Maybe even more damaging, ride-share companies (e.g., Lyft, Uber) expect their drivers to reach certain customer satisfaction ratings. If drivers fail to meet these, they may be banned from driving (Uber, 2017). Our results indicate that workers have harder times when interacting with incongruent others. Given the higher likelihood of incongruence, minorities are at a particular risk of receiving insufficient ratings. Organizations need to monitor satisfaction ratings and potentially account for these effects. This is even more relevant since we find that units that receive larger numbers of biased complaints show increased short-term absenteeism and voluntary turnover, both harmful proximal and distal forms of voluntary employee withdraw.
Limitation and Future Research
While our studies utilize time-lagged and objective data combined with an experiment, they are not without limitations. The context of Study 1, Switzerland, might have influenced our results. Close to 25% of the population has an immigration background (SFSO, 2019b). Residents in countries with more migrants (e.g., Middle Eastern countries, Australia, Germany, United States) should be more accustomed to migrants, which could affect their perceptions, stereotypes, and behavior toward them (Pettigrew & Tropp, 2006). Thus, discrimination against migrant workers might be stronger and a common immigrant identity less evident in countries with fewer migrants (e.g., China, India, Brazil). This calls for similar studies in such contexts. In addition, most immigrants in Switzerland are from other European countries (SFSO, 2019b) with similar cultural values, religion, and so on. Future research should test whether more diverse immigrant populations face more customer complaints than nonmigrants. Future studies could also directly test for the effects of other identities (e.g., racial-ethnic, gender, age, and religion) versus immigrant identities on social attractiveness and customer complaints. For this purpose, our experiment in Study 2 could be extended by measuring the strength of different types of identities as covariates weeks prior to the manipulation and testing if the effects of the immigrant identity manipulation are robust when controlling for other identities. Including additional identities would also allow for testing of potential double-whammy effects in service interactions (e.g., Wu, Han, & Mattila, 2016). The aggregated nature of our data does not allow us to directly test for this.
A traditional team diversity perspective on our results from Study 1 would consider increased conflict in more diverse units as a potential explanation. This is unlikely since we find increased complaints for units with high and low percentages of migrants in incongruent areas. We also tested this possibility with employee data on perceived social support in units, an antonym to conflict. Results indicate that higher percentages of migrants in teams are not associated with lower social support (see OS 7 in the online supplement). Despite this result, future research should focus on better identifying the specific theoretical mechanisms that drive our findings. Doing so would also be interesting because it would allow researchers to test whether the underlying processes for migration differ from other, more traditional diversity dimensions, such as race, gender, and age.
In Study 1, we measure migration via citizenship. Thus, migrants who have lived in Switzerland for a long time and gained citizenship would be coded as Swiss. Yet, they might still be identified as immigrants through their non-native accents. Although the influence of these cases on our results is likely small, it could affect them. This would make it less likely for us to find significant results because these individuals should increase the average number of unfair complaints received by majority-Swiss units. The fact that we still find effects of immigration background speaks to the robustness of our results. Based on this, future research should consider migration background and as an important social identity marker.
Finally, one main reason why we conducted Study 2 was to test for a potential ecological fallacy. We focused on customer complaints as the outcome because we see it as the most central to our model and contribution. However, similar concerns could be voiced for voluntary employee withdrawal. While we were able to control for a number of important predictors of withdrawal, we were not able to account for all potential antecedents (Carpenter & Berry, 2017).
Conclusion
There are 244 million immigrants worldwide, and the number continues to grow (United Nations, 2015). Questions around migration impact many peoples’ lives and make it a central topic in debates worldwide. Gaining meaningful employment represents a central challenge for immigrants, their host countries, and organizations. With these studies, we contribute to the discussion by showing that relational similarities and differences between service employees and customers in terms of migration background matter and that common major life events might create an overarching migrant identity.
Supplemental Material
Online_supplemental_material_FINAL – Supplemental material for More Than Meets the Eye: The Role of Immigration Background for Social Identity Effects
Supplemental material, Online_supplemental_material_FINAL for More Than Meets the Eye: The Role of Immigration Background for Social Identity Effects by David J. G. Dwertmann and Florian Kunze in Journal of Management
Footnotes
Acknowledgements
Both authors contributed equally to this article. We would like to acknowledge the helpful comments and suggestions by editor Bianca Beersma, two anonymous reviewers, Kristie McAlpine, and Max Reinwald. For a prior version of this article, we received the 2019 Best Paper with International Implications Award from the Organizational Behavior division of the Academy of Management. The paper was also one of the five finalists for the all-academy-wide Carolyn Dexter Award. Finally, a prior version of this article was published in an abbreviated, six-page version as part of the 2019 best-paper proceedings of the Academy of Management conference in Boston.
Supplemental material for this article is available with the manuscript on the JOM website.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
