Considerable research points to racial discrimination as a persistent source of labor market disadvantage for racial and ethnic minorities. Audit studies show that when applying for jobs, African Americans and Hispanics are less likely than whites to receive an interview or job offer (Reference Bertrand and MullainathanBertrand & Mullainathan 2004; Reference Fix and StruykFix & Struyk 1993; Reference PagerPager 2003); employers openly discuss their reluctance to hire African Americans and Hispanics (Reference Kirschenman, Neckerman, Jencks and PetersonKirschenman & Neckerman 1990; Reference Moss and TillyMoss & Tilly 2001); and workers of color remain concentrated in the lower rungs of the occupational hierarchy (Reference Stainback and Tomaskovic-DeveyStainback & Tomaskovic-Devey 2009; Reference Tomaskovic-DeveyTomaskovic-Devey et al. 2006). Alongside empirical evidence of continued racial discrimination, there is a growing consensus among social scientists and legal scholars that racial discrimination in the post–civil rights era has taken on a new form. Contemporary discrimination rarely involves concrete actions that intentionally exclude racial minorities from employment opportunities (Reference BielbyBielby 2000; Reference Bisom-RappBisom-Rapp 1999; Reference DeitchDeitch et al. 2003; Reference GreenGreen 2003; Reference ReskinReskin 2000; Reference SchultzSchultz 2003; Reference SturmSturm 2001). Rather, as Reference GreenGreen (2003) notes, contemporary discrimination is a “fluid process” embedded in everyday interactions and workplace structures.
This subtle and diffuse nature of discrimination complicates the process by which workers perceive discrimination on the job and attempt to remedy it. As Reference FelstinerFelstiner et al. (1980–81) point out, to seek legal redress for potential racial employment discrimination, individuals must “name” the act as discrimination, “blame” their employer, and formally “claim” the behavior by seeking redress within the regulatory framework. Although a substantial body of research focuses on formal racial discrimination claims and their outcomes (Reference HirshHirsh 2008; Reference Hirsh and KornrichHirsh & Kornrich 2008; Reference Nielsen, Nelson, Nielsen and NelsonNielsen & Nelson 2005; Reference RoscignoRoscigno 2007), surprisingly little research examines the antecedents of claims—that is, the process by which workers name injurious experiences on the job as discrimination. To the extent that perceptions of discrimination constitute the first step of remedying discrimination, naming is arguably the most critical stage in the dispute framework because the extent and nature of subsequent legal challenges depend on what behaviors workers perceive as injurious and subject to legal intervention (Reference FelstinerFelstiner et al. 1980–81:635).
To better understand the early stage of discrimination disputes, in this article we consider the factors—both personal and contextual—that shape individuals' perceptions of workplace experiences as race discrimination. Our work integrates sociolegal perspectives on the importance of legal consciousness in dispute formation (and in particular the naming process), social-psychological research on perceptions of discrimination, and organizational perspectives on the relationship between workplace contexts and discrimination. Drawing on this framework, we expect individual and social status characteristics to shape perceptions of discrimination in important ways, yet job traits and workplace conditions should provide key contextual cues for interpreting employment experiences above and beyond the effects of individual characteristics. In what follows, we first discuss the mechanisms by which social status, job characteristics, and workplace context affect the process by which workers perceive racial discrimination. Then, using multilevel data on workers and employers that permit systematic examination of perceptions across multiple workplace environments, we assess the effect of these factors on the odds that workers perceive discrimination on the job. We conclude by stressing the importance of a contextual approach to understanding the antecedents of disputes and the application of legal frames, such as discrimination, to workplace experiences.
Legal Consciousness, Subjectivity, and “Naming” Discrimination
The passage of Title VII of the Civil Rights Act of 1964 outlawed racial discrimination in employment, making it illegal for employers to hire, fire, compensate, classify, or deprive workers of opportunities based on their race, color, or national origin. Like many employment laws, the burden of enforcing Title VII law falls on potential victims—via workers' claims of discrimination. If workers perceive discrimination on the job, they can invoke their rights under Title VII by filing formal discrimination claims with federal or local fair employment agencies. Thus, mobilization of the antidiscrimination regulatory framework requires that workers identify negative experiences, attribute them to race discrimination, and bring them to the attention of regulatory agents. However, two workers subject to similar treatment may not be equally likely to identify it as discrimination and subject to legal redress.
A critical factor in this attribution process is the idea of “legal consciousness”—the degree to which individuals invoke legal concepts, such as racial discrimination, to define everyday experiences (Reference EwickEwick & Silbey 1998). Here, we use the notion of legal consciousness broadly to describe the importation of legal principles into everyday life and the transformation that occurs as individuals move toward an understanding of events or experiences as injurious and deserving of redress. This transformation may also involve moral consciousness or the application of an “injustice frame,” both of which recognize that a particular experience or condition violates some moral—if not legal—principle (Reference GamsonGamson 1992; Reference LevitskyLevitsky 2008). For employment discrimination, while all workers may not interpret their experience as unequivocally unlawful, the identification and naming of an experience as race discrimination involves both the recognition that a standard has been violated and the application of a legal principle—even if in a moralistic sense—to workplace relations.
Once exposed to a negative incident at work, individuals engage in a process of causal attribution. This search for explanations can be quite ambiguous, as actions often have various plausible causes. Legally, race discrimination involves the disparate treatment of similarly situated workers on the basis of their race. However, when confronted with allegations of disparate treatment, employers can cite any number of reasons—such as performance, effort, or experience—for treating workers differently, each of which may have nothing to do with race. Indeed employers refute workers' interpretations of race discrimination in the vast majority of formal discrimination claims (Reference James and WootenJames & Wooten 2006; Reference RoscignoRoscigno 2007). Accordingly, in the absence of objective evidence of discrimination, workers' attributions of personal discrimination are “uncertain, subjective, susceptible to human error, and prone to dispute” (Reference Major, Kaiser, Nielsen and NelsonMajor & Kaiser 2005:285).
The study of legal consciousness therefore is tied inextricably to the study of subjectivity (Reference McCann, Fleury-Steiner and NielsenMcCann 2006:xiv). The degree to which individuals draw on the legal concept of discrimination to describe a potentially injurious experience depends on their subjective interpretation of their experience and their knowledge of the law. This is especially the case for race discrimination in the post–civil rights era, given its relatively diffuse form and the difficulty associated with finding hard and fast evidence of disparate treatment. As a result, legal consciousness and the naming of acts as discrimination may vary dramatically from one individual to the next, given the same injurious act or set of circumstances.
Such variation in subjective interpretation is due in part to differences in individuals' social environments. As Reference Fleury-Steiner, Nielsen, Fleury-Steiner and NielsenFleury-Steiner and Nielsen (2006:7) suggest, “How ordinary citizens understand the particular problem they face and the law's role (or lack of role) in solving it is a situational accomplishment that is dependent on the symbolic, organizational, institutional, and spatial factors that are unique to each setting.” Drawing on qualitative studies of a limited set of cases, a number of scholars situate legal consciousness by providing rich accounts of how spatial and institutional contexts and personal identities shape knowledge and use of the law (e.g., Reference AlbistonAlbiston 2005; Reference Ewick and SilbeyEwick & Silbey 1992; Reference McCannMcCann 1994; Reference NielsenNielsen 2000; Reference SaratSarat 1990). This research demonstrates that social contexts affect individuals' knowledge of discrimination and their ability to name injurious experiences on the job as such. Both individual workers and workplace interactions are embedded in larger social environments and status hierarchies that influence the meaning of discrimination and the emergence of disputes.
We build on this qualitative research by integrating sociolegal perspectives on legal consciousness with insights from organizational sociology to provide a quantitative analysis of the effects of personal and contextual factors on perceptions of discrimination. At the personal level, we contend that workers' ascriptive status, knowledge of the law, and the degree to which they feel entitled to fair treatment will affect the extent to which workers name injurious experiences on the job as discrimination. At the contextual level, we argue that workplace environments that encourage standardized, objective decisionmaking and positive intergroup relations should minimize workers' perceptions of discrimination, net of personal factors. We outline these expectations below.
Situating Legal Consciousness
Ascriptive Status
At the most fundamental level, legal consciousness of employment discrimination is a product of “the social location of subjects, and the experiences that arise from that location” (Reference NielsenNielsen 2000:1087). Specifically, membership in a lower-status ascriptive group may shape both objective and subjective workplace experiences. Most social-psychological research expects that lower-status individuals are more likely than higher-status individuals to perceive any given incident as discriminatory (Reference AllportAllport 1954; Reference CohenCohen et al. 1999; but see Reference Major, Kaiser, Nielsen and NelsonMajor & Kaiser 2005; Reference Ruggiero and MajorRuggiero & Major 1998). Regardless of whether lower-status workers experience more objective incidents of discrimination, the perception that they are more often targets of prejudice may situate discrimination as a “more accessible construct” (Reference Ruggiero and MajorRuggiero & Major 1998). The increased vigilance of lower-status individuals to discrimination would suggest that they are more likely to perceive employment discrimination as compared to individuals of higher-status ascriptive groups.
Racial and ethnic identification, foreign-born status, gender, marital status, and parental status are individual-level characteristics that, either independently or in combination, signify social status in the workplace. As a cultural construct denoting unfair treatment based on race, employment discrimination may be especially accessible to racial and ethnic minorities, as most research indicates that racial and ethnic minorities experience more discrimination than whites in various realms of their lives, including the workplace (Reference BumillerBumiller 1988; Reference Nielsen, Nelson, Nielsen and NelsonNielsen & Nelson 2005; Reference RoscignoRoscigno 2007; Reference SmithSmith 2002). Self-identified racial and ethnic minorities are not a monolithic group of workers, of course, and “race” and “ethnicity” represent broad categories with important between- and within-group variations in racialized experiences. Although we expect minorities to perceive more racial employment discrimination than whites due to increased vigilance and increased exposure to negative experiences, we also expect some differences in perceptions of discrimination reported among African Americans, Hispanics, and Asians. Survey data suggest that among racial minorities, African Americans report the most experience with discrimination, whites report the least, and Hispanics and Asians fall in between (Reference SmithSmith 2002). Data on claiming behavior follow a similar pattern; among complaints of race discrimination filed with the federal Equal Employment Opportunity Commission (EEOC), African Americans file the majority of race charges (Reference HirshHirsh 2009; Reference LancasterLancaster et al. 2006).
Other research points to the salience of gender for perceptions of workplace discrimination and suggests possible interactions between race, gender, and parental status. Given that women lag behind men in terms of pay, workplace authority, and occupational status (see Reference Reskin and BielbyReskin & Bielby 2005, for a review), we might expect women, especially women of color, to be more vigilant in identifying discrimination than men. Discrimination may also be a more accessible social construct for married women, particularly those with children, in light of the fact that women still shoulder the majority of unpaid household labor and caregiving responsibilities and thus are more likely to experience work/family conflict (Reference BianchiBianchi 2000; Reference MaumeMaume 2008; Reference StoneStone 2007) and discrimination on the basis of parental status (Reference CorrellCorrell et al. 2007). Given the inequities that women and mothers face at work, they may be more likely to perceive negative treatment as stemming from discrimination as compared to their male and childless counterparts. In these ways, multiple status characteristics may combine to influence perceptions of workplace discrimination.
Entitlement
Employees vary in terms of the degree to which they feel entitled to better treatment and their knowledge of legal entitlements. Workers who feel entitled to fair treatment may be more likely to invoke “discrimination” language to explain negative experiences. Although they are difficult to observe directly, job characteristics and the institutional contexts in which workers are embedded may shape feelings of entitlement, leading workers to be more or less aware of potentially discriminatory behavior. In particular, incumbents of “good jobs,” characterized by high wages, benefits, and authority over others, may be accustomed to better treatment and be more assertive of their rights. They may therefore be more sensitive to negative treatment and potential discrimination, whereas workers stuck in less desirable jobs may simply accept racially charged experiences as another “bad job” characteristic and be unlikely to consider them extraordinary or discriminatory.
In a similar way, promotion opportunities may influence how incumbents interpret potentially discriminatory experiences. Upwardly mobile workers and those located in positions with advancement opportunities tend to have higher aspirations and job commitment compared to workers with few prospects for upward mobility (Reference Cassirer and ReskinCassirer & Reskin 2000; Reference HarlanHarlan 1989; Reference KanterKanter 1977; Reference MuellerMueller et al. 2001). As a result, upwardly mobile workers may feel an increased sense of entitlement and thus be more vigilant in perceiving discriminatory treatment.
Jobs also differ in terms of opportunities for collective representation. Unionization, in particular, provides opportunities for workers to collectively organize to protect their rights and voice dissatisfaction with workplace structures (Reference AbrahamAbraham et al. 2008; Reference Martinez and FioritoMartinez & Fiorito 2009). Recent research on the antecedents of union support indicates that employees who are concerned with procedural justice within the workplace and who identify with the situations of coworkers are more likely to vote in favor of unionization (Reference BladerBlader 2007). Thus, union members may be particularly aware of their employment and legal rights and more likely to identify unjust experiences as racial discrimination, due either to the consciousness-raising efforts of unions or the selection of politically conscious workers into unions.
An employee's knowledge of the law should also influence a sense of legal entitlement. Certain employee demographic characteristics may serve as proxies that correlate with greater knowledge of legal entitlement and experience with legal employment regulations. Specifically, more educated and older individuals may be relatively more knowledgeable of employment discrimination law, reluctant to overlook potentially discriminatory incidents, and empowered to claim discrimination and mobilize the law. Furthermore, insofar as education and age correlate with knowledge of legal structures and sense of entitlement, these individual characteristics may condition the relationship between other status characteristics (e.g., race, ethnicity, and gender) and the likelihood of perceiving discrimination. Specifically, we would expect more educated and older minority group members to be more likely to perceive employment discrimination than less educated or younger minority workers.
Workplace Contexts
Research in organizational sociology highlights the ways in which workplace contexts can exacerbate or minimize inequality and discrimination (Reference BaronBaron et al. 1991; Reference Hirsh and KornrichHirsh & Kornrich 2008; Reference KalevKalev et al. 2006; Reference Reskin and McBrierReskin & McBrier 2000). In the discussion that follows, we elaborate on two mechanisms by which workplace context can sensitize workers to unfair employment practices and discriminatory practices, rendering them more (or less) likely to perceive race discrimination. First, workplace policies and structures can serve as important situational cues that signal to workers that employment practices are fair, lawful, and race-neutral as opposed to discriminatory. Second, the work environment can affect the extent and nature of social interaction between racial groups and the salience of racial group membership in the workplace, both of which may make race discrimination a more or less likely frame for interpreting workplace interactions and behavior.
Signaling Fairness
In the years since the passage of Title VII of the Civil Rights Act of 1964—the legislation that outlaws racial discrimination in employment—employers have instituted a range of practices to standardize employment processes. Such practices were thought to remove managerial discretion—and thus potential race biases—from personnel decisions and promote fair employment practices. While the research evidence is mixed as to whether standardized practices actually minimize discrimination (Reference Edelman and PettersonEdelman & Petterson 1999; Reference KalevKalev et al. 2006; Reference Konrad and LinnehanKonrad & Linnehan 1995), most scholars recognize their symbolic capacity to give the appearance of fairness and minimize legal liability for discrimination in the event of litigation—regardless of whether or not they eliminate ascriptive biases (Reference Bisom-RappBisom-Rapp 1999). In applying this lens to the naming stage of legal disputes, we argue that the presence of formalized and diversity-oriented personnel practices can have important effects on perceptions of discrimination by signaling to workers that practices are fair, objective, and race-neutral.
First, from workers' viewpoints, the use of formal recruitment and screening practices, such as publicly advertising job openings rather than relying on personal referrals, should indicate that hiring procedures are free of ascriptive biases, making workers less likely to perceive discrimination on the job. Indeed, in a study of a multi-establishment telecommunications company, Reference MuellerMueller et al. (2001) found that formalization was negatively associated with women's and men's perceptions of sexual harassment victimization. In keeping with this finding, we expect formalized recruitment and screening methods to minimize perceptions of racial discrimination.
Second, the use of affirmative action programs in personnel decisions, whether mandated or voluntary, can communicate to workers that management is committed to fair employment practices. Affirmative action policies also hold managers and supervisors accountable to diversity goals when making hiring and promotion decisions. Such commitment and accountability on the part of employers should communicate to workers that, to the extent that race affects employment decisions, it does so in positive ways. Thus we expect a negative relationship between employers' use of affirmative action programs and workers' perceptions of race discrimination.
Third, employers' skill requirements and evaluation criteria may affect workers' perceptions of discrimination. Employers who use soft skills such as attitude, personality, and communication style rather than hard or technical skills to hire and evaluate workers may invite perceptions of bias given the ambiguous nature of assessing soft skills. In their study of employer practices using data from the Multi-City Study of Urban Inequality (MSCUI), Reference Moss and TillyMoss and Tilly (2001) found that employers often couched negative views of racial-minority workers in terms of soft or behavioral skill deficits, suggesting that racial minority workers were lacking in personality, communication skills, presentation, and attitude. Insofar as soft skill assessments are subjective, discretionary, and often linked to racial biases, workers may be more apt to perceive discrimination in settings where employers evaluate workers on the basis of soft rather than measurable technical or hard skills.
Finally, the extent to which employment practices are formalized varies according to the size and sector of the workplace. Previous research has found a positive relationship between size and formalization (Reference DobbinDobbin et al. 1993:397; Reference MarsdenMarsden et al. 1994:917; Reference Sutton and DobbinSutton & Dobbin 1996:807) and more equitable practices in the highly bureaucratized public sector, as opposed to the private sector (Reference DiPrete and SouleDiPrete & Soule 1986; Reference Grodsky and PagerGrodsky & Pager 2001; Reference MoultonMoulton 1990). To the extent that formalization paves the way for objective decisionmaking, we expect reduced perceptions of discrimination in larger establishments and among government as opposed to small and private workplaces, respectively.
Intergroup Relations
Workplace contexts can also affect perceptions of discrimination by structuring relationships between racial groups. The racial and ethnic composition of the workplace provides perhaps the most important situational cue for employees' perceptions of race, race relations, and behavior. Workplace composition determines, in part, the visibility of race and ethnicity in the work environment as well as the likelihood that workers have contact with persons of other race/ethnicities at work (Reference ReskinReskin et al. 1999:33).
From a contact perspective (Reference BlauBlau 1977; Reference KanterKanter 1977), as a racial minority group's representation in the workplace increases, workers will have more opportunities to interact with members of other racial and ethnic groups. Such interaction allows workers from all racial groups to acquire individuating information about each other, making them less likely to indulge race stereotypes and biases. Thus, contact allows individuals to forge positive interpersonal ties, leading to improved intergroup relations. Workers in such settings may be less likely to attribute negative incidents to race discrimination compared to their counterparts in predominantly white settings. The contact perspective also implies that the relationship between workplace racial composition and perceptions of discrimination may vary by the racial/ethnic identity of employees, such that minority workers in mostly white settings may be particularly sensitive to racial discrimination.
From the perspective of social identification theories (e.g., Reference TajfelTajfel 1982), however, a different relationship between racial composition and perceptions of discrimination is possible. In particular, larger minority group presence in the workplace may correspond with increased visibility and salience of group membership, making workers more likely to identify as a racial group. As social psychologists point out, identification with a group increases a group's cohesion and members' commitment to it (Reference TajfelTajfel 1982). Such heightened racial group consciousness may render workers more cognizant of potentially racially biased behavior. Working with primarily minority workers may thus correlate with greater perceptions of racial discrimination, especially for minority (as opposed to white) workers.
The race of one's supervisor similarly provides important contextual cues for race relations and workplace equality, with nonwhite supervisors cultivating an environment of positive race relations and signaling opportunities for mobility among minority workers. This reasoning suggests that workers with nonwhite supervisors will be less likely to perceive discrimination.
Racial Biases
Thus far, we have discussed the impact of workplace context on perceptions of discrimination, detailing the importance of workplace structures and intergroup relations for situating workers' attributions of their experiences as discrimination. However, the extent of actual racial bias in the workplace environment will obviously affect workers' perceptions of discrimination. Because bias is largely subjective, it is difficult to empirically observe “actual” racial bias in any given workplace. However, we include a number of measurable indicators to serve as proxies for the extent of racial bias present in the work environment. In this way, we can examine the impact of workplace policies, practices, and environment on perceptions of discrimination while controlling for proxies for racial bias. These include assessments of the racial preferences of employees, employers, and customers and subjective measures of workplace racial tension. To the degree that racial biases lead to discriminatory behavior (e.g., Reference AllportAllport 1954), we expect workers exposed to racial antipathy on the part of fellow workers, employers, customers, and supervisors to be more likely to perceive race discrimination.
In sum, a key question for scholars of legal consciousness is the degree to which personal and contextual factors shape the first, and arguably most important, stage of dispute formation—the perception, or naming, of workplace racial discrimination. Despite theoretical expectations that variation in ascriptive status, legal consciousness, feelings of entitlement, and work environment matter for naming events as discrimination, very little research examines systematically, across multiple contexts, how these factors jointly influence the likelihood of perceiving racial discrimination on the job. We extend existing research by examining these questions with multilevel survey data that link workplace characteristics to employees' perceptions of discrimination.Footnote 1
Data and Methods
Data
We addressed questions about the relationship between personal, job, and workplace characteristics and perceptions of racial discrimination with survey data collected as part of the MCSUI (see Reference BoboBobo et al. 2000; Reference HolzerHolzer et al. 2000). Specifically, we used individual-level data from the MCSUI household survey linked to establishment-level data from the Multi-City Telephone Employer Survey (MCTES). Between 1992 and 1994, MCSUI administrators conducted face-to-face household interviews with a clustered, area-probability sample of adults living in Atlanta, Boston, Detroit, and Los Angeles. The household survey asked respondents about their demographic characteristics, work experience, workplace setting, racial attitudes, and perceptions of employment discrimination. As part of the household survey, interviewers also asked a probability sample of the respondents for the name of their current or most recent employer. These names were among the employers included in the MCTES. For the employer survey, interviewers queried the person responsible for hiring at the firm about characteristics of the work environment, including the firm's labor force, recruitment and hiring practices, workplace policies, customer or clientele base, and various aspects of the work environment. Combining the household and employer surveys produced an individual-employer linked data set of 1,179 records with individual-level information on household respondents matched to establishment-level information on their employers.
Because the Detroit survey did not ask about employment discrimination in the last year (or the last year of employment), we focused on survey data from Atlanta, Boston, and Los Angeles. After excluding respondents from Detroit as well as observations missing information on the question regarding race discrimination, we analyzed data on 830 linked household-employer records. To allow comparable analysis along racial lines, the MCSUI study oversampled African American, Hispanic (in Los Angeles and Boston), and Asian (in Los Angeles) households. We weighted all analysis to correct for the under- or overrepresentation of respondents from particular racial groups.
Measures
Dependent Variable
The dependent variable in our analysis was a dichotomous measure coded 1 if the respondents reported being discriminated against at work because of their race or ethnicity in the previous year or in the last year that they worked. Among the 830 respondents in our sample, roughly 62 workers, or 7 percent of the sample, reported perceiving racial discrimination at work in the previous/last year of employment. As expected, perceived experience of racial discrimination varied considerably by race and ethnicity. Whereas only 3 percent of whites reported being targets of discrimination at work, 20 percent of African Americans, 11 percent of Hispanics, and 10 percent of Asians perceived race-based discrimination.
Independent Variables
To explore how perceptions of discrimination may vary according to ascriptive and social status, we included a set of dummy variables to identify respondents' race/ethnic group membership. These included dichotomous measures coded 1 if the respondent was non-Hispanic African American, Hispanic, or Asian, using non-Hispanic whites as the referent category.Footnote 2 We also included dummy variables to identify respondents who were married as opposed to single, female as opposed to male, foreign versus native born, and those who had children under age 18 as compared to those without young children.
To examine how knowledge of the law and feelings of entitlement to fair treatment might affect perceptions of discrimination, we employed several demographic characteristics of employees and measures of job quality and skills thought to correlate with greater knowledge of the law and entitlement. For demographic characteristics, we included continuous measures of respondents' age and years of education completed. We captured several dimensions of job quality: We included a continuous measure of hourly wage, a dichotomous measure coded 1 if the respondent's job involved authority over others (i.e., supervised, hired, and/or set the pay of other employees), a dichotomous measure coded 1 if the job offered benefits, a dichotomous measure coded 1 if the respondent was a member of a labor union or collective bargaining arrangement, and a dichotomous measure coded 1 if the respondent had ever experienced a promotion.Footnote 3
Next we employed several measures of workplace context to assess how characteristics of the work environment might affect perceptions of discrimination. To assess workplace formalization, we employed indicators of employers' job recruitment and screening methods as reported on the employer survey. First, following Reference Moss and TillyMoss and Tilly (2001), we included a dummy variable coded 1 if the employer used formalized recruitment methods, including newspaper advertisements or referrals from institutions, such as employment agencies, community agencies, schools, or unions; the referent included the use of referrals from a current employee or other personal referrals, help-wanted signs posted on-site, or walk-ins. Second, we included a series of dummy variables coded 1 if the employer reported that the job screening process required a written application, skills test, or personal interview. Third, we assessed the effect of affirmative action policies on perceptions of discrimination with a dummy variable coded 1 if the employer used affirmative action in recruiting workers and a dummy variable coded 1 if affirmative action played a role in hiring workers. Fourth, a dichotomous measure was coded 1 if the respondent's job required soft skills, including daily face-to-face or phone interaction with customers or clients as reported by the employer, to capture how the specific skill requirements of the job affected perceptions of discrimination. Fifth, because establishment size and employment sector might be associated with more formal practices, we controlled for the number of employees in the workforce and included a dichotomous measure coded 1 if the firm was a government agency as opposed to a private establishment.
To assess if the racial composition of the workplace affected intergroup relations and perceptions of discrimination, we included a set of dummy variables to identify racial composition. First, we included a dummy variable coded 1 if most of the respondents' coworkers were nonwhite, including African American, Hispanic, Asian, or “other”; having mostly white (or racially mixedFootnote 4) coworkers was the referent category. Because this measure did not indicate the difference between the race of the respondents and that of their coworkers (i.e., whether it was an African American or white worker in a mostly nonwhite setting), we included interaction terms between the respondents' race and working in a predominantly nonwhite setting. Second, we measured the racial composition of workplace leadership with a dummy variable coded 1 if the respondent's supervisor was nonwhite. The referent category included respondents with white supervisors and those without direct supervisors. These composition measures were based on subjective assessments of coworkers' and supervisors' race as reported by workers on the household survey.
Finally, to account for racial biases present in the workplace, we included a series of dummy variables identifying the racial preferences of workers, employers, customers, and supervisors. These included a dummy variable coded 1 if employers reported that employees at the firm generally preferred to work with coworkers of the same race or ethnic group; a dummy variable coded 1 if the informant reported that customers or clients of the firm preferred working with employees of their same race or ethnic group; and a dummy variable coded 1 if employers in their line of business preferred working with people of the same race or ethnic group, as reported on the employer survey. We also included a dichotomous measure coded 1 if respondents reported that their supervisors used racial slurs in the previous year, among respondents who had direct supervisors, and a dichotomous measure coded 1 if there had been noticeable racial or ethnic tension in the firm in the previous five years, as indicated on the employer survey.
Table 1 provides descriptive statistics (weighted) and data sources for all dependent and independent variables.Footnote 5
Table 1. Descriptive Statistics
Models
Using our dichotomous measure of racial discrimination as the dependent variable and personal, job, and workplace characteristics as independent variables, we used logistic regression to estimate the effect of social status, job characteristics, and workplace context on the likelihood that workers perceived racial discrimination at work.Footnote 6 We employed multiple imputation techniques (see Reference AllisonAllison 2001) to deal with incomplete data on the independent variables. Multiple imputation produced statistical inferences for missing data points by multiply imputing a set of plausible values using Markov Chain Monte Carlo–based methods and adjusting the corresponding variance and covariance matrixes to reflect the uncertainty resulting from the imputations.
Results
Table 2 presents logistic regression models predicting perceptions of racial discrimination during the last year of employment. Model 1 includes measures of ascriptive status; demographic, job, and workplace characteristics; and indicators of racial biases and tension in the workplace. Model 2 presents interactions between the employee's personal characteristics (especially racial and ethnic ascriptive status), job characteristics, and workplace context. Although we tested a number of possible interaction effects, we report in Model 2 only statistically significant interactions that improved model fit.Footnote 7, Footnote 8
Table 2. Logistic Regressions: The Effect of Personal, Job, and Workplace Characteristics on Perceptions of Racial Discrimination
* p<0.1;
** p<0.05;
*** p<0.001 (two-tailed)
1 Coefficients and standard errors multiplied by 100.
Looking first at ascriptive status characteristics, Model 1 confirmed the common expectation that racial and ethnic minorities are more likely to perceive racial discrimination on the job. Compared to white respondents, African Americans and Hispanics were considerably more likely to perceive discrimination; the odds of reporting discrimination increased by 897 percent for African Americans and 447 percent for Hispanics.Footnote 9 Although the estimate for Asian employees was also positive, it was not distinguishable from whites at traditional levels of significance.
Furthermore, gender, marital status, and having children affected the likelihood of perceiving race discrimination. Model 1 indicated that, on average, females perceived more racial discrimination than males, married employees perceived as much discrimination as single employees, and employees with children under 18 reported fewer experiences of racial discrimination on the job than workers without children. However, as shown in Model 2, these patterns were complicated by interactions between these characteristics. Predicted probabilities calculated from the interaction terms for female, marital status, and children under 18 suggested that married women without children were the most likely to report race discrimination, followed by single women with children.Footnote 10 Married men without children were the least likely group to report discrimination.
We also found some evidence that gender and race interacted to influence perceptions of discrimination (Model 2), although the only statistically significant difference existed between African American and white women, and the direction of the effect was unanticipated. Although African Americans and women were, on average, both more likely to perceive discrimination on the job (Model 1), African American women were less likely to perceive workplace discrimination than white women.
According to Model 1, employees' education and age, which we suggest are potential correlates of legal knowledge and sense of entitlement, were not significantly related to perceptions of racial discrimination on average. However, Model 2 indicated that the effects of education and age varied by race and gender. Specifically, education increased the odds that African Americans and women perceived discrimination, whereas age increased the odds of perceiving discrimination only among Asians in our sample.
Net of personal status, several job characteristics significantly related to perceptions of discrimination. First, as expected, union members were significantly more likely to report discrimination compared to nonmembers. Second, upwardly mobile workers who had experienced a recent promotion were more likely to perceive discrimination. Other indicators of job quality—hourly wage and benefits—were unrelated to perceptions of discrimination. Furthermore, the effect of job authority was unrelated to perceptions of discrimination in the baseline model—but was statistically significant in Model 2. Although we found no evidence of racial/ethnic variation in the effects of other workplace characteristics, Model 2 indicated that race/ethnicity moderated the effect of job authority on perceptions of discrimination. Whereas wielding authority over coworkers increased perceptions of discrimination for whites, job authority minimized perceptions of racial discrimination among African American and Hispanic workers. For instance, the predicted probability of perceiving discrimination for African Americans without authority was 19.5 percent, while the probability fell to 2.4 percent for African Americans in positions of authority.
The third panel of variables in Table 2 includes indicators of recruitment and hiring policies thought to formalize personnel practices and thus minimize evaluators' discretion in decisionmaking and promote race-neutral employment practices. As shown in Model 1, the use of a written application reduced the likelihood of workers perceiving racial discrimination. The use of a written application decreased the odds of workers reporting discrimination by 69 percent. By contrast, affirmative action policies in recruitment and hiring, the scaled measure of formalized recruitment procedures, and personal interviews appeared to have little effect on perceptions of racial discrimination, net of other factors. Furthermore, we found no evidence that workers in jobs that required skill tests or soft skills perceived more discrimination than workers in otherwise similar jobs.
With regard to workplace context, the controls for workplace size and sector indicated that, counter to expectations, workers in larger establishments were more likely to perceive discrimination whereas those employed by government agencies were equally likely to perceive discrimination, as compared to workers in the private sector.
Among the subjective indicators of workplace racial and ethnic composition, the estimates in Model 1 indicated that respondents working with mostly nonwhite coworkers perceived more racial discrimination compared to those who worked with mostly white or mixed coworkers. This may be due to the fact that working with predominantly nonwhite coworkers heightens sensitivity to racial biases, making workers more likely to identify racial discrimination on the job. Yet Model 2 suggested that these compositional effects vary by workers' race/ethnicity. Specifically, the positive effect of working in a predominantly nonwhite setting pertained to all groups except Hispanics; for Hispanics, working in a predominantly nonwhite setting minimized perceptions of discrimination. For instance, working among predominantly nonwhite coworkers decreased the odds of perceiving discrimination by 27 percent for Hispanics, while it increased the odds of perceiving discrimination by more than 300 percent for all other racial/ethnic groups. Although having mostly nonwhite coworkers was generally associated with greater odds of perceiving discrimination, the estimate for having a nonwhite supervisor was negative in both models yet statistically significant only in Model 2, suggesting that nonwhite leadership decreases the odds of perceiving racial discrimination in the workplace.
Finally, the last panel of Models 1 and 2 includes five measures that tapped racial biases and tension in the workplace—including whether one's supervisor used racial slurs on the job, respondents' perceptions of racial tension in the workplace, and employer-reported indicators of customer, employer and worker racial preferences. We employed these measures as proxies for “objective” exposure to racial biases and to gauge the general racial climate in the workplace. Because racial biases and preferences are mental states that are not directly observed (Reference ReskinReskin 2003), it is nearly impossible to measure the racial preferences of customers, employers, and workers. Thus our measures are admittedly imperfect indicators of objective conditions; however, they provided the best approximation to actual racialized experiences on the job available in the data.Footnote 11 As Model 1 revealed, among the racial preference measures reported by the employers, only customer preferences were significantly related to perceptions of discrimination; however, this effect was not robust at traditional levels of significance to the addition of the interaction terms in Model 2. The most important bias measure was the use of racial slurs by supervisors. As expected, those with supervisors who used racial slurs in the last year were much more likely to perceive being a victim of racial discrimination. We also found no evidence that personal status or workplace conditions interacted with racial biases to affect perceptions of employment discrimination.
Discussion
The analysis identified several individual and workplace characteristics that affected workers' perceptions of race discrimination. Consistent with expectations, we found that members of low-status ascriptive groups were more likely to identify employment discrimination. Specifically, African American and Hispanic workers perceived much more racial discrimination on the job than did white workers; however, we found no statistical difference between the odds of perceiving discrimination for Asian and white workers net of other variables. This finding implies that race discrimination may be a more accessible social construct for understanding negative workplace experiences for lower-status racial/ethnic minority workers, yet also confirms important variation in racialized experiences among minority groups.
Exceptions to this pattern emerged when we examined the joint influence of multiple statuses via interaction effects. For example, while we found that women, net of racial group members, were more likely to perceive discrimination than men, we did not find that women of color perceived more racial discrimination than white women. By contrast, the opposite held true. Furthermore, we expected married women with children to perceive more discrimination than their counterparts, but the analysis suggested different patterns. Although marital status, gender, and the presence of children did interact in significant ways to influence perceptions of discrimination, we found that married women without children perceived the most racial discrimination on the job, whereas married men without children were least likely to perceive discrimination. We did find, however, that among women, marital status determined whether having children led to greater perceptions of discrimination. Having children exacerbated perceptions of discrimination for single women, yet minimized perceptions of discrimination among married women. This may be due in part to the increased work-life conflict that single mothers likely face given the lack of spousal support.
We hypothesized that education and age, as correlates of knowledge of legal entitlements, would be positively associated with perceptions of racial discrimination. Although we found no support for this hypothesis averaged across our whole sample of workers, we did find variation in the effect of education and age on perceptions of discrimination. Education increased perceptions of racial discrimination among women and African Americans relative to whites (but not other racial/ethnic groups), and age increased perceptions of discrimination for Asians. These results suggest that knowledge of legal entitlements may render workers of lower ascriptive status more likely to perceive race discrimination.
Several job and workplace characteristics influenced perceptions of racial discrimination above and beyond the effects of individual identities and ascribed status. For one, we found that workers who belonged to unions were more likely to report discrimination than nonunion members. Insofar as unions strive to improve employment conditions and employment rights for their members, union membership should make workers more aware of their workplace environment and legal rights. We caution, however, that we cannot adjudicate between alternative explanations for this effect, such as selection into union jobs.
Some measures of job quality, namely promotion opportunities and job authority, also structured perceptions of differential treatment. Promotion opportunities were positively associated with perceptions of discrimination for all racial groups. This suggests that workers who experience promotion opportunities may have increased rights awareness and, consequently, are more likely to perceive race discrimination. The effect of job authority on perceptions of discrimination was positive for whites and Asians yet negative for African Americans and Hispanics. Thus, authority appears to protect African Americans and Hispanics from perceived discrimination.
We identified several hiring methods that affected workers' perceptions of race discrimination. Notably, in establishments where employers used formalized screening methods, particularly written applications, workers were less likely to report race discrimination. This finding confirms that policies designed to remove managerial discretion from hiring and screening processes can have important consequences for workers' interpretations of their workplace experiences.
Among the most important workplace contextual factors for identifying race discrimination was the race and ethnicity of coworkers and supervisors. We found that working with predominantly minority coworkers increased perceptions of discrimination for most racial groups. One interpretation of this positive effect is that working among predominantly minority workforces increases sensitivity to racial discrimination on the part of many employees and increases the disposition to perceive potential incidents as discriminatory. According to our analyses, racial/ethnic minority groups perceived much more racial discrimination in employment than whites, and working in a context with mostly racial minority coworkers may lead to heightened awareness of racial issues in employment. The negative interaction between predominantly minority workforces and Hispanic workers suggests an exception to this pattern that warrants further investigation. It is possible that working in a setting with mostly nonwhite coworkers does not increase group cohesion and consciousness of racial bias among Hispanics, particularly if coworkers are mostly African American or other racial minorities of non-Hispanic origin. However, we are reluctant to draw definitive conclusions regarding this relationship given that the parameter estimate was marginally significant (p<0.09) and the cell size of the interaction was relatively small. For supervisors, our results showed that having a nonwhite supervisor minimized perceptions of discrimination for all employees, regardless of race or ethnicity. Nonwhite supervisors may promote positive race relations, and having a nonwhite supervisor may also indicate to workers that employment practices are race-neutral, thereby minimizing perceptions of discrimination.
Finally, the results indicated that some racial biases and tensions in the workplace influenced perceptions of discrimination. As expected, we found that employees with supervisors who had used racial slurs were more likely to report experiencing discrimination on the job. However, the racial biases of employers and employees, as reported by the employers, did not seem to influence perceptions. The lack of effects for these racial bias and tension measures may reflect the complexity associated with measuring racial preferences. Because racial biases can be subtle, unconscious, and difficult to observe, these measures may underestimate the influence of bias in the workplace. Nevertheless, we found that the effects of ascriptive status, job quality, hiring practices, and race composition persisted even after accounting for variation in racial biases and tensions. This finding confirms that characteristics of the work environment influence how workers perceive discrimination regardless of their exposure to overt racial bias.
Conclusion
A central question in the study of dispute formation concerns the conditions under which individuals come to name workplace experiences as racial discrimination. Although the perception of discrimination is arguably the most important stage in dispute formation in that it determines the likelihood of subsequent mobilization of the law (Reference FelstinerFelstiner et al. 1980–81), relatively few studies examine the antecedents of perceptions of workplace discrimination across multiple workplace contexts. Sociolegal research increasingly recognizes that legal consciousness—i.e., the degree to which individuals invoke legal concepts, such as race discrimination, to define everyday experiences—is a situational accomplishment influenced by both personal and contextual variables. We build on previous research by analyzing a unique multilevel data set that permits assessing workers' perceptions of discrimination across multiple employment settings while controlling for exposure to overt racial biases. Our study demonstrates the importance of individual and contextual factors for attributions of discrimination and points to a number of theoretical mechanisms through which these factors may translate into perceptions of discrimination.
First, results suggest that racial employment discrimination may be a more accessible construct among lower ascriptive status groups (African Americans, Hispanics, and women). Consistent with social-psychological research that stresses the role of status hierarchies in structuring attribution processes, our results support the idea that ascribed status structures legal consciousness of employment discrimination in important ways, either via more frequent exposure to racialized experiences or heightened vigilance to potentially unfair treatment based on race/ethnicity. Although this pattern is consistent with related research that demonstrates a tendency for lower ascriptive status groups to report more discrimination (e.g., Reference HirshHirsh 2009; Reference RoscignoRoscigno 2007; Reference SmithSmith 2002), our results advance previous studies in several ways. First, by controlling for job and workplace characteristics, we show that African Americans, Hispanics, and women more readily attribute negative experiences at work to race discrimination regardless of the quality of their jobs or location in the hierarchical structure. Thus racial differences in reporting discrimination cannot be explained away by differences in job quality. In addition, we find that lower-status ascriptive groups are more likely to report discrimination net of measures of overt racial bias in the workplace. Finally, our findings point to the importance of exploring the intersection of multiple status dimensions for workers' experiences. When we consider the joint effect of race and gender, or gender, marital status, and the presence of children, for example, we find patterns that deviate somewhat from the general rule that lower ascriptive status correlates with greater odds of perceiving discrimination.
Next we find support for the claim that the likelihood of invoking discrimination language to explain negative workplace experiences varies with a sense of entitlement and knowledge of the law. Although our data do not allow us to observe these processes directly, we argued that employees' personal and workplace characteristics may structure feelings of entitlement or legal knowledge, and our results generally support these expectations. Education and age translate into greater odds of perceiving discrimination for some lower ascriptive status groups. Furthermore, workers with job authority, promotion experience, and union members are most likely to perceive workplace racial discrimination.
Our findings also have implications for organizational theories of inequality and discrimination. An extensive literature in organizational sociology documents employers' efforts to minimize managerial discretion in personnel practices by adopting formalized and diversity-conscious procedures (see Reference BielbyBielby 2000; Reference DobbinDobbin et al. 1993; Reference KalevKalev et al. 2006; Reference KmecKmec 2006; Reference ReskinReskin 2000). Although research examines the substantive impact of such structures on workforce diversity, very little research examines their effects from the perspective of workers. We argued that the presence of formalized practices can affect perceptions of discrimination by signaling to workers that practices are fair, objective, and race-neutral. Our findings with respect to formalization are somewhat mixed. On the one hand, many of the measures of formalized practices do not impact workers' attributions of discrimination, including diversity-conscious policies such as affirmative action in recruitment and hiring. On the other hand, one cornerstone of formal hiring procedures—a written application—reduces the odds of perceiving discrimination for all groups. This finding suggests that, consistent with neo-institutional research that demonstrates the symbolic value of formal organizational structures (Reference EdelmanEdelman 1992; Reference Edelman and PettersonEdelman & Petterson 1999), some personnel practices can hold important symbolic meaning for workers that affects their perceptions of racial treatment on the job, regardless of their substantive impact on workforce diversity or discrimination. Thus, although we find some evidence that formalization can limit perceptions of racial discrimination on the job, more research is needed to explore why some policies may be more consequential than others. The lack of effects for affirmative action policies may reflect countervailing effects on workers' perceptions of discrimination. Given the often-controversial nature of affirmative action policies, some workers may view their presence and enforcement as a sign of fair practices, whereas others may view them as expressions of “reverse discrimination.” We note, however, that we did test for interactions between affirmative action policies and employees' race/ethnicity and found no variation in effects.
We also suggested that workplace contexts could affect attributions of discrimination by structuring the extent and nature of intergroup contact and relations. In keeping with social identification approaches to group cohesion (i.e., Reference TajfelTajfel 1982), we find that workers embedded in majority-minority settings are, on average, more likely to report discrimination compared to those in predominantly white settings. The increase in racial group identification and cohesion afforded by numbers makes discrimination a more accessible social construct for understanding negative events.
As these results demonstrate, the process of identifying experiences as discrimination is embedded in a larger workplace context and set of relations that influence the likelihood of naming, and subsequently reporting, discrimination. Thus theoretical and empirical accounts of the emergence of discrimination disputes must examine not only the potential acts of discrimination that give rise to disputes but also the employment environments in which they occur. The corresponding policy implication of these findings is that mobilization of the law and law enforcement efforts will vary in their frequency and effectiveness across workplace contexts. Because the likelihood of naming discrimination varies across workplace environments and naming is the necessary antecedent of formal legal claims, some workplaces may be more likely to generate discrimination claims than others simply due to their structures and practices. For instance, a potentially discriminatory act or behavior may go unnoticed in a predominantly white environment yet lead to a legal dispute in a predominantly minority workplace. Thus, regulatory agents and the courts must be mindful of the ways in which the naming and claiming of discrimination are produced situationally.
Our results should also remind regulatory agents that not all workers will interpret workplace conditions and policies in the same way. We find, for example, significant racial/ethnic variation in the relationship between job characteristics (job authority), workplace conditions (workforce racial composition), and perceptions of discrimination, further highlighting the importance of exploring perceptions of employment discrimination with a contextual lens. Such variation in how groups interpret workplace experiences and environments may complicate not only the process of dispute formation but also the success of remedying racial and ethnic tensions in the workplace. For instance, remedies that target specific racial, ethnic, and gender groups may not work for others. Future research should explore in more detail the potential reasons underlying differences in the interpretive frameworks of workers, especially along racial, ethnic, and gender lines.
Our interest in the “naming” stage of workplace discrimination necessitates a focus on the point of view of employees. We thus emphasize the individual and contextual factors associated with employees' perceptions of discrimination. This emphasis on the employees' point of view broadens traditional legal and sociological conceptualizations of employment discrimination that focus exclusively on behavior and practices on the part of employers. However, we acknowledge that employee perceptions—and legal consciousness—are at least partially a function of employer actions as well. Like other studies of discrimination, our data do not permit us to compare perceptions of discrimination with employer behaviors, and thus we are unable to disentangle employees' subjective interpretations of discrimination from employers' potentially discriminatory behavior. Nonetheless, our research demonstrates the need to broaden research on employment discrimination to include workers' points of view.
A comprehensive theoretical account of the naming stage of employment discrimination should consider both employee perceptions and employer actions, and the way that supply- and demand-side processes interact to influence legal consciousness and dispute formation. Understanding workers' perceptions of negative events as discrimination is crucial for effective antidiscrimination law enforcement. Yet subjective definitions may partly reflect perceptual biases, imperfect information, and error, and they thus may not necessarily satisfy objective or legal standards. Future research should undertake research designs (e.g., Reference BlackstoneBlackstone et al. 2009; Reference Uggen and BlackstoneUggen & Blackstone 2004) that facilitate unpacking objective experience and subjective interpretation, to the extent that researchers can distinguish objective workplace events from individuals' interpretations of the events. In doing so, researchers can provide a more thorough understanding of how workers apply legal frames, such as discrimination, to their experiences and how social environments affect this attribution process.
, where α is the intercept, β is the vector of regression coefficients, and x is the vector of covariate values. By varying the values of the dummy variables included in the interactions and holding all other variables in the model at their means or modes, we obtained predicted probabilities for each group implied by the interaction.
