Written Testimony of Rachel D. Godsil Professor, Seton Hall University School of Law

Meeting of July 1, 2015 - EEOC at 50: Progress and Continuing Challenges in Eradicating Employment Discrimination

Good morning Madam Chair and distinguished members of the Commission. Thank you for the opportunity to provide testimony for today's meeting to mark the 50th anniversary of the Equal Employment Opportunity Commission. My name is Rachel Godsil, and I am the Eleanor Bontecou Professor of Law at Seton Hall Law School, and co-founder and director of research for the Perception Institute, a national consortium of social scientists, law professors, and advocates focusing on the role of the mind sciences in law, policy, and institutional practices.

The focus of my testimony today will be the obstacles to equal employment opportunity posed by implicit bias and racial anxiety.¹ While recent events make clear that explicit racism, sexism, and other forms of express hostility based upon ethnicity and religion, continue to exist, our country has made significant strides addressing the role of explicit sexism, racism, and other express forms of bias. The vast majority of employers have express policies prohibiting such discrimination and make active efforts to prevent those forms of bias from disrupting their workplaces. The rejection of explicit bias has not had the effect we might have hoped of eliminating the salience of identity differences.

Even when most people hold egalitarian goals, identity differences continue to matter in the work place because of the inter-relationship of implicit bias and racial anxiety. Social science research shows that the vast majority of people hold implicit biases -automatic attitudes or stereotypes about people based upon their race, ethnicity, or gender that affect decision-making, and behavior, without our even realizing it. Less recognized in the popular discourse, but well-established in the literature, is the phenomenon of racial anxiety - the discomfort people feel in anticipation of or during interracial interactions. Racial anxiety leads to behaviors that appear similar to bias - maintaining less eye contact and warmth, greater physical distance, and even avoiding interactions with people of other races altogether. Such behaviors can have significant consequences for perpetuating racial inequities in the workplace - for instance, white employers conducting shorter interviews with non-white job applicants, supervisors engaging in less interaction and mentoring. The combination of implicit bias which can result in harsher critiques and assumptions about competence and racial anxiety can be substantial in creating significant differences in opportunities and experiences in the workplace.

In this testimony, I will first describe the research identifying how implicit bias and racial anxiety have the potential to undermine fairness in the workplace and then, as crucially, the range of interventions that have been found to minimize or eliminate them - which offer the promise of truly achieving equal opportunity in the workplace in the years to come.

I. Implicit Bias

Implicit bias refers to the automatic stereotypes and attitudes that result from repeated exposures to cultural stereotypes of different groups based upon race, gender, ethnicity, religion, age, and other categories that pervade society.² Over the past two decades, social cognitive psychologists have discovered novel ways to measure the existence and impact of implicit biases. Using experimental methods in laboratory and field studies, researchers have provided convincing evidence that implicit biases exist, are pervasive, are large in magnitude, and have real-world effects.³ These discoveries have migrated from the science journals and law reviews into the national discourse on race,⁴ and are now reshaping the law's fundamental understandings of discrimination and fairness.

Implicit bias is receiving attention in light of recent tragedies, but crucial to the workplace is how implicit biases are replicated in everyday micro-behaviors demonstrating that race, ethnicity, gender, and other identity differences still matter - the black lawyer who is assumed to work in the mail room, the female investment banker who is whipsawed between claims that she is too timid and too aggressive, the query about whether a Latino or Asian American speaks English, or the question "Where are you really from?" asked of fellow citizens from different racial and ethnic groups. People can consciously reject negative stereotypes or attitudes in relation to different groups, but those negative stereotypes or attitudes can still be triggered automatically or "implicitly."

Implicit bias is a result of the automatic, unconscious association of attributes with different groups, but at an explicit or implicit level, bias can also manifest as a result of comparatively positive preferences for one group over another. Social scientists refer to this phenomenon as "in-group" bias or preference. ⁵ In-group bias is more likely to be explicit than is animus, but it can often be implicit as well. Whites who hold explicit in-group preference will rarely interpret their feelings as "racist" or "sexist" if they do not involve active animus. Yet, when biases and preferences become translated into behavior, the result is the same: members of one group benefit relative to members of another. Although we tend to think of discrimination primarily as treating a person or a group worse, treating a favored racial or ethnic group better or presuming men are more competent than women rather than women as incompetent results in the same outcome.⁶ For example, studies have shown that whites generally will not overtly rate blacks negatively - they will simply rate similarly situated whites more positively. ⁷ When people experience in-group bias, they tend to be more "comfortable with, have more trust in, hold more positive views of, and feel more obligated to members of their own group."⁸ In the context of in-group bias linked to race, researchers have found that people may try to avoid out-group members-an avoidance which often leads to distortions in perception and bias in evaluation of in-group and out-group members.⁹ The combination of biased evaluations and preferences and differential levels of trust are particularly pernicious in a work-place setting.

A recent study by a consulting firm working with law firms provides a vivid example of the potential harms of implicit bias. In this study, 60 partners were given an identical memorandum written by "Thomas Meyer," identified as a summer associate from New York University Law School (a top 10 school), that contained 22 different errors, 7 of which were minor spelling/grammar errors, 6 of which were substantive technical writing errors, 5 of which were errors in fact, and 4 of which were errors in analysis.¹⁰ Half of the partners were led to believe that Meyer was white (Caucasian) and the other half that Meyer was African American. The results quoted below from the study's report are telling:

• An average of 2.9/7.0 spelling/grammar errors were found in "Caucasian" Thomas Meyer's memo in comparison to 5.8/7.0 spelling/grammar errors found in "African American" Thomas Meyer's memo.
• An average of 4.1/6.0 technical writing errors were found in "Caucasian" Thomas Meyer's memo in comparison to 4.9/6.0 technical writing errors found in "African American" Thomas Meyer's memo.
• An average of 3.2/5.0 errors in facts were found in "Caucasian" Thomas Meyer's memo in comparison to 3.9/5.0 errors in facts were found in "African American" Thomas Meyer's memo.

The commentary associated with the assessments of the memo are telling. "Caucasian" Thomas Meyer is seen as "having potential," while one partner stated about "African American" Thomas Meyer, "can't believe he went to NYU."¹¹ These partners undoubtedly reject explicit bias and are not at all hostile to the idea of a black associate; rather, the first spelling error likely resulted in what is known as "confirmation bias" in which we interpret information in a way that confirms preconceptions or stereotypes.

Research linking implicit bias to behavior relevant to the work-place is robust, ranging from perceiving facial expressions differently to offering job callbacks at different rates. Specifically, in studies of facial expressions, whites with stronger implicit racial bias perceive black faces as angrier than whites with weaker levels of bias; similarly, those with stronger implicit bias are apt to consider an expression happy or neutral if displayed by a white person, but neutral or angry if displayed by a black person.¹²

While much of the national conversation around implicit bias has focused on black-white relations, implicit bias research has shown broad implications of such bias against a wide range of groups. Implicit negative associations toward Asian Americans, for example, has been linked to less positive assessments of the competence of Asian Americans as litigators,¹³ resistance to hiring Asian American candidates for national security jobs, and rejecting progressive immigration policies if proposed by Asian Americans.¹⁴ Field studies demonstrate that Latino, as well as black, job applicants are significantly less likely to receive callbacks than are equally qualified white applicants.¹⁵ In addition, in a summary by Mark Bendick and Ana Nunes, there have been "several dozen testing studies" in the past two decades, in multiple countries, focusing on discrimination against various demographic groups (including women, the elderly, and racial minorities) consistently revealing typical "net rates of discrimination" that range from 20-40 percent.¹⁶ In other words, in 20-40 percent of cases, employers treat subordinated groups worse than privileged groups even though the testers were carefully controlled to be identically qualified.¹⁷ Laurie Rudman and Peter Glick demonstrated that when job descriptions required an employee to be cooperative and to work well with others, women who were self-promoting and competent (researchers termed them "agentic") were rated less hirable than equally "agentic" men.¹⁸

Social scientists have found that people often engage in "motivated reasoning," in which they justify assessments of who is qualified or what work is meritorious by changing criteria as they go, sometimes referred to as the "malleability of merit."¹⁹ People are also apt to judge actions differently depending upon the race or ethnicity of the actor. If the action of a minority group member has a positive result, it will be seen as a result of luck, but if the action is by a dominant group member, the result is seen as proof of their aptitude. By contrast, negative acts understood as consistent with stereotypes will be generalized to the person when the person is a minority (she is a 'tardy person'), while the same act will be deemed situation specific (she was 'delayed') if the person is white. Social scientists call this tendency the "ultimate attribution error," and have found it to be particularly prevalent in the context of racial groups.²⁰ Along with "confirmation bias," attribution error helps explain why finding a typographical error in a memorandum by a minority associate may inadvertently lead a partner to presume the associate is not worth effort, while the partner would not have the same response to the typographical error by a white associate. And it is important to recall that the partner generally would not be conscious of the different response.

Eric Luis Uhlmann and Geoffrey Cohen identified "motivated reasoning" or the "malleability of merit" when they asked participants to evaluate two finalists for police chief-one male, the other female. One candidate's profile signaled book smart, the other's profile signaled streetwise, and the experimental design varied which profile attached to the woman and which to the man. Regardless of which attributes the male candidate featured, participants favored the male candidate and articulated their hiring criteria accordingly.

Addressing implicit bias is clearly a crucial step. Yet researchers warn that focusing only on the need to reduce bias and inhibit the automatic activation of negative attitudes and stereotypes creates the potential for "rebound effects."²¹ Racial anxiety must simultaneously be addressed to ensure that workplaces offer equal opportunities regardless of group membership.

II. Racial Anxiety

Racial anxiety is a common but often unrecognized dynamic that has significant implications for the workplace. People experiencing racial anxiety during a cross-racial interaction often distance themselves from the other person, are less apt to share eye contact, and use a less friendly and engaging verbal tone. These behaviors may seem fairly benign, but can undermine a job interview when the interviewer is white and the person being interviewed is minority, or create tension between supervisors and employees of different races.²² Racial anxiety in one interaction can also translate into avoidance of both the particular individual and others of the same race.²³

A related concern is that racial anxiety may lead white supervisors to refrain from providing honest feedback because of a worry that they will be perceived as prejudiced. While praise rather than critical feedback may seem either benign or even helpful, in the education setting, researchers have found that false praise undermines rather than encourages a student's growth.²⁴ If given skewed feedback in the workplace, employees will be uninformed about the quality of their work and will be deprived of the necessary tools to learn and improve. Again, if we extrapolate research from the education context to the workplace, it is likely that employees are often aware when praise is unwarranted²⁵ and would view critical feedback as a sign of care when it is conveyed supportively and shows the supervisor's belief that the employee is capable of improvement.²⁶ The experience of receiving unwarranted praise may lead minority or female employees to discount genuine praise as a sign of "intergroup politeness,"²⁷ and distrust or cynicism flowing from the experience can lead to "attributional ambiguity" in which the employee is unsure whether criticism is a result of bias and praise a result of condescension.

III. Interventions

The EEOC has a Congressional mandate and the institutional authority to serve as a catalyst for institutions to adopt practices that will best eliminate or reduce implicit bias and racial anxiety. Individuals hold implicit associations and attitudes and experience racial anxiety because unconscious processes absorb both biased cultural messages and deeply held norms of racial fairness. Yet broad cultural messages and noxious stereotypes can be defused by contexts that reduce bias and anxiety. The catalysts for institutions and individuals to undertake interventions to achieve these effects will vary. Some will embrace the opportunity to create conditions that are consistent with equality ideals. Others may respond only under pressure from institutions such as the EEOC which will have greater likelihood of success in light of the robust evidence that race is the proximate cause of harmful behavior.

A. Interventions to Address Implicit Bias

Social science research focusing on addressing the effects of implicit bias can be divided into two broad categories: interventions seeking to "debias" (that is, to reduce implicit bias) and those directed toward mitigating the effects of bias and preventing implicit biases from affecting behavior. All agree that generic admonitions about race are unhelpful; the premise of this literature is that the vast majority of people already hope to adhere to racial equality norms.

Patricia Devine and colleagues have found success in reducing implicit bias by combining multiple interventions to "break the prejudice habit." These data "provide the first evidence that a controlled, randomized intervention can produce enduring reductions in implicit bias."²⁸ While earlier studies have found implicit bias to be less malleable, Devine et al. have replicated their study and are poised to publish a second article describing their findings in 2015.

However, because of the challenges of both reducing bias and ensuring that it does not re-emerge, employers must recognize that to ensure that bias does not translate into discriminatory practices within institutions, they will be required to establish procedures to "over-ride" bias.

Two forms of interventions will be required. The first are interventions devised to address the effects of implicit bias on decision-making. This will require an assessment of areas in which subjectivity may enable bias to affect outcomes: hiring criteria generally, resume review, assignment decisions, evaluation procedures, promotion decisions must all be assessed. For a description of actions to decrease the likelihood that implicit bias will affect decision-making, consider the following recommended by a group led by Jerry Kang including social scientists, law professors, and a federal judge.²⁹

Doubt Objectivity
When people presumes that they are objective, they increase the risk that they will inadvertently allow bias to influence decision-making because they will trust their automatic reactions which are likely to be influenced by unconscious stereotypes or attitudes. There is some evidence to suggest that teaching people about implicit bias will lead them to be more skeptical of their own objectivity and, as a result, be better able to guard against biased evaluations.

Improve Conditions of Decision-making
Implicit biases are a function of automaticity. "Thinking slow"³⁰ by engaging in mindful, deliberate processing prevents implicit bias from determining our behaviors. Ideally, decisions are made in a context in which one is accountable for the outcome, rather than in the throes of any emotion (either positive or negative) that may exacerbate bias. In addition, providing clear criteria for evaluation may prevent against phenomena in which racial and ethnic minorities and females of all groups tend to be disfavored such as confirmation bias, the malleability of merit, or attribution error.

Count
Implicitly biased behavior is best detected by using data to determine whether patterns of behavior are leading to racially or ethnically disparate or gendered outcomes. Perhaps not surprisingly in light of the assumptions many make about the decrease in discrimination in our society, research has shown that people are more likely to detect discrimination when it is presented in the aggregate rather than on a case-by-case basis. Once one is aware that decisions or behavior are having disparate outcomes, it is then possible to consider whether and how the outcomes are linked to bias.

Applying these interventions to each step in the employment process has enormous potential to address the cognitive dimensions of implicit bias. However, as noted above, implicit bias also affects relational dynamics which are equally salient in determining whether those who are in non-dominant groups within particular workplaces have an equal opportunity to succeed or whether they will be subject to a set of discriminatory barriers established by the response of employers to their race, ethnicity, gender, or aspects of identity. Along with the decision-making that determines which resumes will result in call-backs, which criteria will be used to assess proficiency in performance and the necessary elements for promotion, employers must also ensure that implicit bias does not affect the treatment of candidates when they interview, or the way in which critical feedback is communicated, or whether employees are included in team-building social activities and welcomed along with their white or male counterparts. These "soft" interactions are often as critical to success - who is hired, encouraged to stay, recommended for promotion, and finally given opportunities for leadership.

Because both evaluation and inter-personal dynamics matter, employers must also take into account that those who adhere to egalitarian norms are likely to be deeply concerned and upset when they learn that they have not successfully shed the effect of noxious stereotypes. This reaction can be helpful if it creates incentives to adopt the interventions described above to ensure that behavior is not dictated by implicit biases. Without careful management, it can also trigger racial anxiety and its attendant harms described above.

B. Reducing or Preventing Racial Anxiety

As I explain above, racial anxiety can infect the dynamics between whites and minorities when whites manifest anxiety in inter-group interactions and particularly if anxiety results in avoidance of people from other races and ethnicities. Racial anxiety can also lead to a reluctance to provide the necessary feedback employees need to develop their skills and to become successful team members. Social scientists have found that intergroup contact can lead to reduction in racial anxiety, yet this positive outcome is rarely immediate and does not necessarily occur. The conditions of the contact situation can either undermine or facilitate the potentially positive effects.³¹ How employers organize their workplaces and structure the environment will strongly determine whether racial and ethnic minorities are treated as valued members of the team or as outsiders.

It has long been recognized that certain factors are of particular importance, including the establishment of equal status between groups, cooperation, common goals, and institutional support for the contact.³² Providing clear instruction for inter-group feedback and opportunities to inter-group social contact can also be critical.

In order to reduce racial anxiety and its attendant negative consequences for minority employees, employers can institute practices relying upon effective mentoring studies and "behavioral scripts."³³ If supervisors are taught that they will be perceived as less biased if they provide critical feedback than if they provide empty praise, as long as the critical feedback is coupled with affirmation that they have high expectations of the person who is receiving the criticism and have confidence that the person can meet those expectations, they will be less likely to experience racial anxiety when providing the crucial feedback. This intervention can help prevent the adverse effects that whites' racial anxiety by addressing nonverbal as well as verbal cues. For instance, if a white person in a position of authority knows that she is doing right by her employees, she is likely to feel more confident and less anxious in the interaction and may therefore be less likely to engage in distancing or avoidant behavior.

This hypothesis was confirmed in a study by Avery and colleagues who tested the utility of providing "defined social scripts (i.e., norms dictating expected interpersonal behavior)" to white participants prior to black white interracial interactions.³⁴ Their goal was to reduce behavior that would stem from anxiety felt by white participants. Their research built upon earlier researching findings that whites reported feeling more comfortable in scripted interactions with blacks (for example, serving a black customer in a restaurant) than in unscripted interactions (sitting in a crowded table in a library where a black person is already sitting). Researchers were interested in white participants' behavior rather than self-reports, and in behaviors detectable to black people and which trigger avoidance on both sides of the racial dyad. Using video telephone conversations as a vehicle, researchers in this study found that scripted encounters were effective in reducing white anxiety as measured by third-party observers and suggested that providing scripting is particularly important for initial interactions.

Extrapolating the results, Avery et al. suggest that institutions should provide structured interactions for first encounters - such as asking people to "tell each other three interesting things about yourself" or to "describe your role in the organization."³⁵ Such scripts are particularly important for initial and critical interactions such as job interviews and initial meetings with supervisors when the cues received by the minority interviewee or employee are likely to affect their own response and thus likely to affect the impression they make upon the interview panel or supervisor.

* * * *

In conclusion, as the EEOC considers its mission to reduce discrimination in the workplace and to ensure that people have an opportunity to succeed regardless of their race, ethnicity, gender, or other identity characteristics, it will be crucial to recognize that phenomena other than explicit animus are likely to create more significant obstacles to these goals. The vast majority of employers and supervisors consciously affirm egalitarian values - but social science research and the continued disparities we see in workplace outcomes tell us that behavior does not automatically follow our conscious values. Instead, as a result of operations of our unconscious and the continued prevalence of stereotypes linked to race, ethnicity, gender, and other identity characteristics, implicit biases are more determinative of our behavior than conscious values. Simultaneously, because we fear that we may not live up to our country's norm of equality, particularly in our interactions with people of other racial and ethnic groups with whom we may be unfamiliar, racial anxiety is also an impediment in the workplace that differentiates the experience of those in the dominant group from those in minority groups. The EEOC is unique in its institutional authority to catalyze employers to take the necessary steps to address these phenomena and to provide an equal playing field for every employee to achieve to their capacity and to have those achievements recognized and rewarded without the interference of bias or anxiety.

---

Footnotes

¹ This testimony draws liberally from Rachel D. Godsil, Linda R. Tropp, Philip Attiba Goff & john a. powell, The Science of Equality (2014)(reviewing literature) http://perception.org/app/uploads/2014/11/Science-of-Equality-111214_web.pdf.

² Id. (citing R. Song Richardson & Philip Attiba Goff, 2012).

³ Jerry Kang et al, Implicit Bias in the Courtroom, 59 U.C.L.A. REV. 1124 (2012).

⁴ President Obama referred to the challenges of implicit bias in the workplace in his moving eulogy of Rev. Clementa C. Pinckney, Kevin Sack and Gardiner Harris, President Obama Eulogizes Charleston Pastor as One Who Understood Grace, New York Times, June 26, 2015.

⁵ Godsil et al, supra note 1, citing (Brewer, 1999; Tropp & Molina, 2012).

⁶ Id. (citing Reskin, 2002).

⁷ Godsil et al., supra note 1 (citing Dovidio & Gaertner, 2004).

⁸ Id. (citing Reskin, 2000).

⁹ Id. (citing Reskin, 2000; Brewer & Brown, 1998).

¹⁰ Id. (citing Nextions, 2014).

¹¹ Id.

¹² Id. (citing Hugenberg & Bodenhausen, 2003).

¹³ Id. (citing Jerry Kang et al, 2010).

¹⁴ Id. (citing Yogeeswaran & Dasgupta, 2010).

¹⁵ Id. (citing Pager et al., 2009).

¹⁶ Marc Bendick, Jr. & Ana P. Nunes, Developing the Research Basis for Controlling Bias in Hiring, 68 J. SOC.ISSUES (forthcoming 2012), available at http://www.bendickegan.com/pdf/Sent_to_JSI_Feb_27_2010.pdf.

¹⁷ Kang et al, supra note 3.

¹⁸ Laurie A. Rudman & Peter Glick, Prescriptive Gender Stereotypes and Backlash Toward Agentic Women, 57 J. SOC. ISSUES 743, 757 (2001).

¹⁹ Kang et al, supra note 3.

²⁰ Id.

²¹ Godsil et al., supra note 1 (citing Dovidio et al., 2008)

²² Id.

²³ Id. (citing Butz & Plant, 2011; Dovidio et al., 2006; Plant & Butz, 2006; Plant & Devine, 2003; Tropp, 2003).

²⁴ Id. (citing Harber, 1998); see also K.D. Harber et al., Students' Race and Teachers' Social Support Affect the Positive Feedback Bias in Public Schools, 104 J. OF EDU. PSYCHOL. (2012).

²⁵ Id.

²⁶ Godsil et al, supra note 1 (citing Yeager et al., 2013; Cohen et al., 1999; Cohen & Steele, 2002).

²⁷ Id. (citing Harber, 1998).

²⁸ Id. (citing Divine et al., 2012).

²⁹ Kang et al, supra note __.

³⁰ Daniel Kahanaman, 2011.

³¹ Godsil et al., supra note 1 (citing Tropp & Page-Gould, 2014).

³² Id. (citing (Allport, 1954; Pettigrew & Tropp, 2006).

³³ Id. (citing Cohen et al., 1999; Goff et al.,

³⁴ Id. (citing Avery et al, 2009).

³⁵ Id.