Stereotype threat and regional far-right political support

Stereotype threat occurs when the minority identity of an individual is activated and this can be elicited by various cues [31]. Social interactions can be a source of such cues [40], but exposure to devaluing content in the media can also impair negatively stereotyped groups [6]. Immigrants who reside in regions where devaluing content is more prevalent, such as in regions with higher far-right political support, are therefore more likely to experience stereotype threat-eliciting cues when engaging with locals or consuming various forms of local media (e.g., social media, television, newspapers, public radio). In line with this, previous research has found that newspaper coverage of professional athletes is biased depending on players’ nationality [41] and ethnic group membership [42]. Even without direct insight into locals’ anti-immigrant attitudes, subtle yet symbolic cues such as apparel (e.g., the red MAGA hats supporting Donald Trump) [43], election posters, and locally individualized ads and discussions in television and social media may inform immigrants whether the social environment of the region where they reside is potentially threatening or not. Taken together, we expect that sustained exposure to negative cues reflective of regional far-right political support in the region where immigrants reside will lead immigrants to experience chronic stereotype threat, increasing their work-related performance errors. In contrast, immigrants living in regions where these cues are less common (i.e., regions with less far-right political support), should be less affected. Therefore, we propose the following:

Hypothesis 1: Regional far-right political support is positively associated with performance errors in immigrant employees.

Stereotype lift and regional far-right political support

Social interactions, media, and symbolic cues devaluing immigrants may not only affect immigrants’ performance errors but could also have the potential to affect native employees. Previous research has shown that non-stereotyped individuals (i.e., members of the ingroup) demonstrate increased performance when an outgroup is negatively stereotyped [44−47]. This phenomenon is known as stereotype lift and it has been theorized to boost performance (i.e., decrease performance errors) by increasing individuals’ downward social comparisons with the denigrated outgroup [47], in turn increasing their sense of self-efficacy and self-worth [48]. As a result, ingroup members (i.e., natives) may be less prone to “self-doubt, anxiety, and fear of rejection that could otherwise hamper performance” [47, p. 457]. Empirical evidence for this comes from studies showing that stereotype lift effects on a motor task were mediated by increased self-confidence and task involvement [45], as well as decreased anxiety [49]. Taking this previous work into consideration, we expect that sustained exposure to cues reflective of far-right political support in the region where natives reside will lead natives to experience stereotype lift, decreasing their work-related performance errors. In contrast, natives residing in regions where these cues are less common (i.e., regions where far-right political support is lower), should be less affected. We therefore hypothesize the following:

Hypothesis 2: Regional far-right political support is negatively associated with performance errors in native employees.

Data and sample

The sample size was determined by the number of basketball players playing in the 2020–2021 regular season of the NBA. We specifically selected the 2020–2021 season for this study as it spans the transfer of presidential power in the U.S. following the 2020 election and thus covers a time in which political affiliation and support for Donald Trump and his Republican party were both visible and polarizing. Of note, the 2020–2021 regular season of the NBA went from December 22^nd, 2020 to May 16^th, 2021 and supporters of Donald Trump stormed the nation’s capital on January 6^th, 2021 in an attempt to overturn the results of the presidential election in which he lost.

Data were collected for all five hundred and forty (N = 540; 100% male) players from the official NBA website (www.nba.com). Furthermore, we matched the NBA data with presidential election data at the state level openly available from the Harvard University dataverse [50]. Matching of players to state election data was based on the home location of their team and 522 players were matched to 23 regions. One team consisting of 18 players could not be matched to the presidential election data because the team was located in Toronto, Canada. Teams consisted on average of 18.2 players (SD = 1.9). Players in our final sample were on average 26.0 years old (SD = 4.1) and came from 41 different countries, the most coming from the U.S. (77%), Canada (3%), and France (2%). Of those with immigrant status, 48.3% were born in Europe, 5.9% in Asia, 13.6% in Africa, 5.9% in South America, 18.6% in North America, and 7.6% in Australia/Oceania.

Measures

Covariates.

To account for the fact that performance errors (as operationalized above) can only occur when a player has possession of the ball, we accounted for players’ total number of ball possessions. Furthermore, because some forms of performance error result from having possession of the ball for too long (e.g., a “shot clock” violation), we controlled for the total amount of time (in minutes) players had possession of the ball. We also accounted for players’ age and position played as their basketball experience and role within the team may determine their number of performance errors. In addition, and as recommended in the literature on stereotype threat [38], we accounted for factors reflecting how players are treated by their organizations. These covariates included players’ salary and the total number of minutes they were allowed to play. Players’ skin color was also included as a covariate to show that immigrant status accounts for variance in performance errors above that of skin color. To measure player’s skin color, the profile photo of each player on the official NBA website was coded by two research assistants blind to the focus of the study using the single-item Fitzpatrick Skin Phototype Scale [58]. This measure involves a 6-point Likert scale (1 = ‘light, pale white’, 6 = ‘black, very dark brown to black’) accompanying a colored image with an example skin color reflecting each scale option. As recommended [59], interrater reliability was tested using two-way random-effects intraclass correlation coefficients (ICCs). Interrater reliability for the variable skin color was found to be high ICC(A,1) = .97, validating its use in our analysis. We also controlled for team-level variables that were considered likely to affect players’ number of performance errors. These included a measure of how successful the team was during the season (as indicated by the number of games won in the regular season) and a binary variable indicating whether the team started the season with a new head coach (nine teams did). We included this as previous research has shown that coaching changes can affect team performance [60,61]. Lastly, and has been recommended [62], we included the cluster (i.e., level-2) means of our individual-level (level-1) predictors (i.e., immigrant status, number of ball possessions, time with ball possession, age, salary, minutes played, and skin color) to account for these variables at the team level.

All variables related to game-play represent statistics totaled per player across the entire 2020–2021 regular season in which each team played 72 games. The means, standard deviations, and zero-order correlations among all study variables are presented in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Means, standard deviations, and zero-order correlations among variables.

https://doi.org/10.1371/journal.pone.0289019.t001

Robustness checks and supplemental analysis

To establish the robustness of our findings, we retested Hypothesis 1 and 2 using a simplified model. Players’ total number of ball possessions was kept in our robustness check analysis because, as previously noted, players cannot demonstrate performance errors without possession of the ball. Furthermore, and as in our previous analysis, we included the cluster (i.e., level-2) means of our individual-level (level-1) predictors (i.e., immigrant status and number of ball possessions) to account for these variables at the team level.

Using the same procedure as in our main analysis, we again found that regional far-right political support interacted with immigrant status to predict players’ performance errors, B = 0.423, p < .001 (see Table 2: Model 2 for details). Furthermore, and similar to the results of our main analysis, follow-up simple slopes analyses show that regional far-right political support is positively related to performance errors in immigrant players, B = 0.222, p = .002, and negatively related to performance errors in native players, B = -0.202, p < .001. These findings provide additional support for Hypotheses 1 and 2. To further test the robustness of our findings, we re-ran our main analysis and robustness check analysis while excluding the 10% of players with the fewest number of ball possessions throughout the season (i.e., less than 115). This was to rule out that our findings were driven by players with few opportunities for performance errors. Exclusion of these players (n = 53) did not significantly affect our results or our interpretation of them and, therefore, they are not discussed further.

In an attempt to better understand the negative relationship between regional far-right political support and immigrants’ performance errors, we tested a model in which skin color was allowed to interact with regional far-right political support and immigrant status. This was done to determine whether immigrants with darker skin color exhibit greater or fewer performance errors relative to those with lighter skin color in regions with higher regional far-right political support. Using the same procedure as in our main analysis while including the 3-way interaction between regional far-right political support, immigrant status, and skin color, we failed to find statistically significant differences between indviduals based on their performance errors, B = -0.055, p = .485 (see Table 2: Model 3 for details). This lack of a finding suggests that immigrants residing in regions with higher far-right political support exhibit more performance errors, regardless of their skin color.

Theoretical contributions

Our study makes several theoretical contributions to research on immigrants in organizations. First, although immigrants make up an ever-growing subgroup of the workforce, they have received little attention in the organizational behavior and management literatures [7]. Furthermore, the limited research that has been conducted on the individual-level work-related outcomes of immigrants has focused largely on blue-collar workers [17,18]. By focusing on immigrants widely recognized as among the best in their profession, we show under what circumstances, and to what extent, even the most skilled and successful immigrants demonstrate increased work-related performance errors. In so doing, we extend the current literature on what has been considered an “invisible” group [67] by highlighting the experiences of its most visible members.

Second, our work complements research on immigrants at the macro-level that has found associations between regional far-right political support and psychological variables such as lower life satisfaction [4] and higher perceived discrimination [5]. Specifically, we extend these prior findings by showing that regional far-right political support is positively associated with immigrants’ number of work-related performance errors, thus implying that their chances of attaining (and maintaining) the financial stability that employment offers might be negatively affected. This stability is further threatened by our finding that natives, in contrast, exhibit fewer performance errors in regions with higher far-right political support. To our knowledge, this study is the first demonstrating relationships between regional far-right political support and performance-related work outcomes.

Finally, this work introduces professional sports as a viable context in which to study immigrants in real-world, organizational settings using objective measures of performance-related work outcomes (e.g., performance errors). Although there are likely relatively few immigrants working in the majority of organizations [7], professional sports teams are much more diverse as it relates to athletes’ immigration status– 23% of basketball players in our sample were immigrants–and likely to become increasingly diverse with time [68]. With this study, we hope to encourage fellow researchers investigating immigrants as a subgroup of the working population to consider the unique benefits that the professional sports context has to offer (e.g., large, openly-available datasets of individuals nested within teams) when proposing and testing theoretical models.

Practical implications

We demonstrate that the political environment in which immigrants reside is negatively associated with the performance-related work outcomes of even the most successful and capable immigrants in their field. Our findings are therefore practically relevant for organizations as they could be interpreted as surmising that there is no level of success that immigrants can attain that will protect them from the corrosive influence of a threatening, external political environment. Thus, organizational leaders would do well to try and insulate their organization from the potentially divisive influence of the external political situation by, for example, prohibiting employees from wearing discriminatory or politically-laced apparel (e.g., MAGA hats) and creating inclusive climates [69] as an internal buffer against immigrant stereotype threats. Concurrently, team leaders could publicly recognize the unique added value of each of their (immigrant) employees to reassure them of their belongingness within the organization and profession as a whole.

While our research took place in a specific professional sport setting, the main implications might also be transferable to more traditional jobs. For example, similar to professional athletes, professional drivers of busses, taxis and limousines are often immigrants [70] and must perform their job tasks under the watchful eyes of community members. Furthermore, as driving also involves well-learned and automatic tasks, immigrant drivers may be at risk of the explicit monitoring associated with stereotype threat. Moreover, and similar to professional sports teams, leading (IT) startups in the U.S. rely heavily on recruiting the best talents worldwide. In 2018, for example, a study showed that 55 percent of the startup companies in the U.S. valued $1 billion or more were founded by immigrants [71]. If executives and other top leaders make more errors as a result of anti-immigrant threats in their environment, that could seriously harm the organizational performance and even the overall economic performance in the U.S. Finally, our findings likely also pertain to international companies who send their employees for assignments abroad, as the political environment may explain in part why these employees frequently do not perform to their supervisors’ expectations [72].

Limitations and future directions

Despite the strengths of our study, it is not without limitations. First, the research context of our study is the NBA, an elite professional organization comprised of athletes who are not representative of the population as a whole. In particular, and relevant to the psychological mechanism we propose drives our results, NBA players receive status and public exposure (including scrutiny) far above what the average immigrant (or non-immigrant) does. These factors are important to consider given that perceptions of status and exposure to outgroup members can affect one’s experience of stereotype threat [46] and performance [73]. Immigrant workers in less public roles may, therefore, be at lower risk of stereotype threat arising from the region in which they work; though research has also shown stereotype threat to negatively affect performance when alone [74]. Future research can build on our model, as well as address the limited external validity of our unique study sample, by extending our findings to organizational contexts that better represent the general population of employed individuals and their exposure to public scrutiny (or lack thereof) while working. Examples of this could include studying performance differences between immigrant and native workers in roles with varying amounts of public exposure (e.g., office workers vs. postal workers vs. news reporters) across regions with varying far-right political support. Such work would help provide external validity to our findings by replicating them in more representative samples of workers. That being said, the fact that we nonetheless find our results in an elite group of immigrant professionals underlines the impact that regional far-right political support can have both on significantly less-famous immigrants (e.g., expatriates) and significantly more vulnerable immigrants (e.g., refugees). Second, our findings are limited by the fact that they are drawn from a young sample of employees. Although our results suggest that younger immigrant employees can be negatively affected by regional far-right political support, older immigrant employees may become desensitized to it due to inurement (i.e., habituation) effects [75] or personality changes over time (e.g., self-confidence and independence increase in middle age) [76]. On the other hand, older native employees may experience greater stereotype lift effects than their younger counterparts simply from having been exposed to more anti-immigrant stereotypes over their lifetime. To address this limitation, future research could build on our theoretical model by testing age as a potential moderator of interest in an age-diverse sample of employees. Third, our operationalization of regional far-right political support is based on state-level election data. Although this state-level measure may be considered coarse, prior research has found inter-state differences in personality (e.g., openness to experience, liberal values) to vary systematically [77,78] and that people can accurately report differences in residents’ personality characteristics across regions [79,80]. Specific to our study, Republican-leaning states are often referred to as conservative and close-minded [78], which is in line with findings that Republicans self-report more negative perceptions of immigrants (e.g., they are not to be trusted) than do Democrats or Independents [81]. This prior work supports our assumption that differences in attitudes towards immigrants exist at the state level and that these attitudes are salient to immigrants. Furthermore, and lending credence to our methodology, U.S. state-level election results have recently been shown to predict employees’ integration into their organization [82] as well as organizations’ corporate performance [83]. Finally, although we propose chronic stereotype threat and stereotype lift as the psychological mechanisms responsible for players’ performance errors, we could not directly test this with our dataset. Moreover, the observational nature of the data prevent causal conclusons from being made and experimental studies are needed to determine if, and how, immigrant status causes work-related performance errors in regions with high far-right political support. In the future, researchers could employ mixed method designs matching basketball statistics with player interviews to better understand players’ personal experiences of stereotype threat and stereotype lift before games. This could help elucidate not only the underyling mechanism(s) driving these individual effects (e.g., social categorization salience resulting from political rhetoric on immigration), but also the role of potential moderators such as perceived inter-group conflict and, relatedly, prejudice [84]. Qualitative studies building on our theoretical model could thereby provide more nuanced insight into which individuals are particularly susceptible to stereotype threat/lift effects, and why. Although mixed method designs increases the complexity of one’s study procedures, initial work has demonstrated the predictive validity of such approaches [85].