1 Introduction

Across the U.S., institutional diversity efforts are forcing change. The UC Davis ADVANCE Program seeks to remove identifiable intra-organizational barriers that white women and people of color face when pursuing STEM careers in academia. As a science-based initiative, ADVANCE targets the STEM disciplines and asks how we move beyond “fixing the numbers” (getting more white women and people of color into STEM) to “fixing the culture”—creating and sustaining an environment in which a diverse array of faculty with a diverse array of social identities can fully participate and thrive.

Psychology and management studies figure prominently in diversity, equity, and inclusion (DEI) literature, much of it focusing on the role of individual/cognitive processes in accomplishing cultural change; in what follows, I include this literature but also provide significant sociological grounding for understanding barriers to inclusion, because social and individual processes are inevitably entwined. Such grounding might seem self-evident to some readers, but my seven years of collaborating on ADVANCE initiatives on my own campus with STEM and non-STEM colleagues alike has convinced me that it’s essential to get on the same page with respect to what concepts mean and how to effectively deploy them in transformational efforts, particularly when that transformation requires interdisciplinary collaboration. Consider this chapter a primer of sorts for those seeking a common language with which to pursue diversity, equity, and inclusion in their own organization or institution.

2 Difference, Hierarchy, and Inequality

People are social creatures—we live, work, and play in groups and communities of varying sorts that we ourselves create and sustain. These groups may reflect family, kin, or neighborhood ties, professional networks, religious or political beliefs, shared abilities/disabilities, and/or national allegiances, among other possibilities. Communities may be governed by informal norms or formal laws; their boundaries may be relatively porous or strictly policed. Boundaries define who’s inside and outside of the group: consequently, groups and communities impart a shared sense of belonging as well as a shared sense of difference from others—which may (but does not automatically) lead to the creation of hierarchy.

2.1 Hierarchy

“Difference” is a necessary but not sufficient prerequisite to hierarchy unless the difference makes a difference in relation to systems of power. Wade and Ferree (2019) use the example of tongue aptitude versus sex category. Because of the presence or absence of a gene, some people are able to curl their tongues while others cannot, yet tongue aptitude is not a socially significant difference that we organize hierarchical relations around, unlike biological sex (female/male). Differences give rise to hierarchies which create/sustain inequalities when power—economic, political, cultural, and social—is unequally distributed among groups systematically and over time (Fig. 1). Other terms for hierarchy include “stratification” and “tiering”—metaphors that suggest the layering of groups on the basis of status and/or access to resources and opportunities. Those of the top are better off than those at the bottom.

Social hierarchies may be fixed or open to varying degrees. Fixed hierarchies are akin to caste systems: individuals are born into their tier and the status/resources conferred by tier membership is a function of birthright, not specific skills, attributes, or accomplishments. Mobility within a tier may be possible, but not between tiers. Slavery in the U.S. was a fixed hierarchy organized around race. By contrast, open hierarchies are said to allow social mobility (the more mobility, the more “open”)—social class is a typical example. Today, the U.S. class structure is generally understood to be hierarchical but in an open way; popular narratives suggest that if you work hard enough you can move up in the hierarchy, thanks to your own initiative, to educational opportunities, or some combination of both. Yet even open hierarchies are still hierarchies, and they can be and often are discriminatory if differential access to the resources needed to be socially mobile (either within or across tiers) creates systemic barriers to that mobility.Footnote 1

Access to higher education is one of those resources, yet access is not equally available to everyone and some who gain access are made to feel they don’t belong. Consider a working-class Latina student whose parents are migrant farmworkers. Because of gender, race, and class disadvantage, along with, potentially, English language barriers, she will not have the same access to the economic capital (resources), social capital (networks), and cultural capital (culturally-valued knowledge and culturally-valued ways of demonstrating that knowledge) as a U.S.-born young man from the white middle class. These structural disadvantages limit access to high-quality educational experiences at every stage of the academic pipeline, and, ultimately, career choice (see Mayer, 2010). Should both students eventually attend a prestigious four-year university, their experience of the institution may (and likely will) differ dramatically; for the young man, it may seem a natural stage along a predetermined path, whereas, for the young woman, it may seem a strange and alienating place.

Fig. 1
figure 1

Social hierarchy. Illustration by Mengmeng Luo

2.2 Social Identity

Sociologists have long used the concept of identity not only to study how people understand who they are (self-identity), but also, and more important, how that understanding is shaped by social/group membership (social identity). Social identity is thus one’s sense of self as a member of a social group.

Typical categories that give rise to social identities in the U.S. include but are not limited to gender, race, ethnicity, sexual orientation, social class, age, ability/disability, religious affiliation, political ideology, occupation or work status, geographic location, and national origin. Some social identities and the categories from which they are derived are ascribed (based on “innate” qualities) and some achieved (based on chosen or elective criteria), although ascription and achievement are not mutually exclusive; consider that one might be assigned male at birth but identify as a woman later in life, or that an ethnic identity may be rooted as much in cultural practices as in phenotype or national origin. Moreover, even so-called innate qualities must be interpreted and given meaning in a social context. Although seemingly self-determining, social identities are formed and embraced in relation to others and so the perspectives of others—whether inside or outside one’s own group—can influence identity-formation. In other words, although no one else can force you to identify in a particular way, the options available to you and the meanings attributed to those options are not exclusively within individual control.

Although identities are by no means static or unchangeable, and although some identities are more strongly held than others, they typically constitute a core sense of sense, at least while they are held. When aspects of the self are more loosely embraced or short-lived, they may be better described as roles. As the theatrical metaphor implies, roles are parts people play in their lives—in the family (daughter), workplace (teacher), and other organized groups such as sports teams (goalkeeper), civic organizations (treasurer), or professional associations (council member). There can be a blurry line between “identities,” “roles,” and “occupations.” For example, “mother” can be both an identity and a role because it is potentially a core sense of self but also a part played within the family or kinship network. Likewise, “scientist” is an occupation/profession but may also be an identity.Footnote 2

Within any given identity category exists a range of possible social identities. The category “race” may encompass Latinx, indigenous, African American or Asian American identities. Identities tend to be more dynamic than the categories from which they’re derived in response to changing personal and/or socio-political influences. Any one individual may hold multiple identities simultaneously: within or alongside racial identifications, people will have different gender, class, religious identifications, etc., although which identities are claimed or felt to be salient can vary considerably by context. This notion of multiple identities is related to but not synonymous with the terms “social location,” “positionality,” and “intersectionality.”

3 Social Location, Positionality, and Intersectionality

Social location refers to the combination of identity categories to which a person belongs (Fig. 2). Theoretically, everyone has a social location—straight, white, middle-class men no less than queer, working-class women of color. However, people may or may not recognize/acknowledge their social location as a social location or as influencing their worldview—how they know what they know and make sense of the world around them. Positionality, a term coined by feminist philosopher Linda Alcoff (1988), refers to the effect of social location on perspective. According to Alcoff, the key issue is not that some social locations are “objective” and others “biased,” but that all social locations are perspectival and some perspectives are more important to prioritize than others in understanding and challenging social inequalities.

Intersectionality is a framework that seeks to understand how systems of power converge in patterned ways to amplify oppression. Coined by the legal scholar Kimberlé Crenshaw in 1989 but theorized by women of color scholars and activists long before this, “intersectionality” denotes the ways in which race, class, gender, sexuality, and other axes of hierarchy operate not as mutually exclusive categories but as reciprocally constituted ones that shape social experience. In its early formulations, the concept of intersectionality underscores the multidimensionality of marginalized subjects’ lived experiences as the result of interlocking structures of oppression (see Truth, 1851; The Combahee River Collective Statement, 1977; Dill, 1983; Anzaldua, 1987; King, 1988; Crenshaw, 1989, 1991; Collins, 1990/2000). Other terms that predate or are in dialogue with the concept of intersectionality are “multiple jeopardy” (King, 1988) and “matrix of domination” (Collins, 1990/2000).

A legal scholar, Crenshaw developed the framework of intersectionality to address the fact that the experiences and struggles of women of color have been rendered invisible in U.S. discrimination doctrine because, in legal as in popular discourse, the iconic subject of racial discrimination is assumed to be male and the iconic subject of gender discrimination is assumed to be white (Crenshaw, 1989, 1991). Black women, and women of color more generally, are neither men nor white, and their unique experiences of oppression in society often reflect the simultaneous inflection of their specific social locations, race and gender among them. One needn’t possess a specific constellation of social identities to experience intersectional subordination, nor is intersectionality “additive” (the more layers, the more oppression). The point is to understand how different vectors of power/inequality come together to shape experience. A Vietnamese-American waitress isn’t necessarily more or less oppressed than, say, an immigrant Latina scientist, but the specificity of their social locations will position them to experience subordination differently. Intersectionality is an analytic framework or lens, not a collection of identities.Footnote 3

This is not to suggest that social identities are irrelevant in recognizing (and, conversely, failing to recognize) intersectional vulnerability. It is no accident that it was women of color feminists who developed the concept of intersectionality; people who are marginalized on the basis of social location do not have the luxury of ignoring its impact on their lives. An important aspect of social privilege is believing that one is somehow unmarked or “neutral” with respect to identity categories (being white and male, for example). In actuality, there is no neutral person who doesn’t have a social location, nor is there a person who occupies only a single social location—there is no “raced” person who isn’t also gendered, for instance, just as there is no gendered person who is raceless. But the degree to which individuals recognize their positionality—the intersections shaping their experience—often reflects the degree to which they are marginalized (and privileged) in relation to the category.

Fig. 2
figure 2

Social location and social identity. Illustration by Mengmeng Luo

4 Controlling Images

Social identities are learned, acquired throughout life in various ways: face-to-face interaction, shared experience, explicit instruction, cultural representations, formal assignment by rule of law, etc. This means social identities arise from both inside and outside groups and communities and may require the negotiation and resolution of conflicting messages. Socialization within the family or school may encourage/support one social identity, and the attributes associated with it, while peer networks or the media may support quite another; some groups have more influence than others in identity-formation, and this is true of both self-identity and social identity.

Social identities, and their associated attributes, are not equal in status in the broader society, nor, to reiterate a point made earlier, are they simply a matter of self-determination—of deciding, among competing options, which one you prefer. Generally speaking, the more marginalized are members of a community in terms of access to power and resources, the less able they may be to define for themselves who and what they are in the broader society and the more vulnerable they are to being negatively defined by others, since a key aspect of power is the ability to determine how people are perceived and treated. A clear example is the use of what Collins (1990/2000) calls “controlling images” to denigrate African-American women. Controlling images go beyond visual imagery to encompass societal representations more generally and they are related to but not synonymous with stereotypes, since stereotypes can be positive or neutral as well as negative and harmful. Controlling images facilitate “othering,” whereby dominant groups create/sustain the subordination of “others” by constructing these others as inferior, thereby reaffirming the apparent legitimacy of the arrangement (see also Hooks, 1989).

When internalized, controlling images may not only shape how marginalized groups make sense of their lives but also effectively hinder their ability to challenge their subordination (see Pyke, 2010; Pyke & Johnson, 2003). Collins was focused primarily on controlling images of black womanhood—the mammy, the matriarch, the welfare mother, and the jezebelFootnote 4—but her concept has been usefully applied to other groups and communities, including Asian American women (Pyke & Johnson, 2003) and Latinas (Mendible, 2007). As Pyke and Johnson (2003) note, pathologizing the femininity of women of color renders white constructions of femininity as “normal” and superior, thus according white women racial privilege, even as they may suffer gender oppression relative to white men.

The concept of controlling images reminds us that societal-level representations—historical narratives, cultural images, political ideologies, and/or laws and policies—do not provide a panoply of “neutral” options for identity-formation. Because societal-level representations reflect the priorities and experiences of those with the greatest power to author them (wealthy straight white men), they may reinforce, intentionally or not, demeaning (racist, sexist, homophobic, etc.). characterizations. Changing societal-level representations by diversifying their authorship is one reason why diversity in social institutions such as law, medicine, education, and politics matters, and why it’s imperative to challenge the exclusion of marginalized voices from these spaces.

5 Collective Action and Institutional Change

Of course, marginalized groups and communities do not automatically internalize others’ ideas of what their identities signify, although this happens. They can and do define themselves, sometimes in explicit opposition to mainstream narratives and ideologies. Identity-formation can involve a dynamic interplay of social forces, including collective action. The labor movement of the 1930s, the black and brown power movements of the 1960s, and the feminist and queer rights movements of the 1970s were key moments in U.S. history when racial, ethnic, class, and gender identities were actively renegotiated and reframed, in some respects quite profoundly.

Along with other U.S. institutions, institutions of higher education were partially transformed by these movements, albeit unevenly and incompletely, as newly-founded gender and ethnic studies programs demanded an expansion in what kinds of knowledge got produced and who got to produce it. More traditional disciplines “diversified” much less readily, particularly the STEM fields, prompting the emergence of federal grant initiatives like the one that inspired this book. Today, across the U.S., the majority of under-represented minority (URM) faculty—and to a lesser degree white female faculty—in academia remain clustered in ethnic and gender studies programs. Not coincidentally, these programs have been chronically under-funded, under-valued, and subject to conservative political attack. Consequently, although academia has widened the range of social identities represented among its faculty, it mirrors the culture at large in that faculty from historically marginalized groups are clustered at the “low” end of the occupational hierarchy (in gender and ethnic studies) and grossly under-represented at the “top” (in STEM).

Institutional transformation efforts such as the NSF ADVANCE Program exist in no small measure because of social justice efforts, but this history is indirect—the more immediate legacy is the notion of “diversity education” that began in corporate, military, and educational contexts in the post-civil rights era, solidifying in the 1980s and 90s when businesses began hiring diversity trainers as a way to avoid civil rights lawsuits (Kelly & Dobbin, 1998; Vaughn, 2007). Although ethnic and gender studies scholars, along with humanities-oriented social scientists, are critical of terms such as “diversity” and “inclusion” because of their ascension within corporate workplaces and their prioritization of individual-level behavioral/cognitive understandings of social inequality over deeply structural ones, the individual and the social inevitably come together and touch down in different ways in a variety of settings, including STEM.

Initiatives such as ADVANCE are strategic and necessarily limited interventions. The ADVANCE Program at UC Davis aims to change the organizational culture of STEM fields; it is not a broad social justice project designed to eliminate racial and gender inequality in higher education as a whole (if it were, time and resources would be also be devoted to improving the funding and working conditions for the humanities and social science disciplines that are, comparatively, already quite diverse), let alone society writ large. Rather, as a science-based initiative, it attempts to address identifiable, measurable, and (ideally) correctable barriers within academia that hinder the diversification of STEM faculty by discouraging white women and people of color from full participation and inclusion. Since STEM is the primary target for institutional transformation, the vision and the strategies for implementing it have to come from within the STEM community, not imposed from outside. UC Davis ADVANCE has a particular emphasis on Latinas in STEM, given the demographics of California and the mission of the university to serve the diverse peoples of the state.

6 Diversity and Inclusion

As discussed in the first chapter, ‘From Affirmative Action to Inclusion,’ “diversity” and “inclusion” are related but not synonymous: being diverse means the group or community in question encompasses a range of social identities representing a range of social locations/backgrounds, whereas being inclusive means that people of different backgrounds and identities feel—and are—valued members of the group who participate on equal footing. Sherbin and Rashid (2017) quote diversity advocate Verna Myers in explaining the difference: “diversity is being invited to the party, inclusion is being asked to dance.” Diversity is necessary but not sufficient to achieving inclusivity within social institutions; moreover, whereas diversity is relatively easy to measure (e.g., a headcount), inclusion is less so because it is an aspect of culture and must be interpreted and narrated. Why are diversity and inclusion important? Why should we strive for them? Why should we care about them?

In their masterful book, An Inclusive Academy: Achieving Diversity and Excellence, Stewart and Valian (2018), draw on a range of data across multiple disciplines to address these and other questions related to defining, understanding, and fostering diversity and inclusion in the academic workplace both in and outside STEM. Their review of the literature suggests that embracing diversity in an inclusive way yields multiple benefits: increased innovation and creativity; greater capacity to challenge received wisdom; the development of new fields of inquiry; and the ability to reach and inspire students of different backgrounds and experiences. In no instance will mere demographic diversity guarantee any of these benefits just as homogeneity won’t automatically preclude them (as Stewart and Valian point out, many important advances have occurred despite the historic homogeneity of higher education); but diversity will be make such benefits more likely because it is a proxy for diverse thinking—and thus more varied ways of seeing, understanding, and inhabiting the world, which has an impact not only on knowledge itself but also those aspiring to become knowledge-producers (see Chap 2 in Stewart & Valian, 2018).

7 Stem in Academia: Barriers to Diversity and Inclusion

Having discussed a number of general sociological processes and concepts of broad relevance to understanding diversity and inclusion as well as the benefits of diverse, inclusive environments, I now turn to some of the specific challenges STEM fields confront in pursuing those two interrelated goals with regard to their faculty. Below we focus on six different but interrelated phenomena: exclusion, the ideal worker norm, positionality, stereotypes, microaggressions, and implicit bias.

7.1 Exclusion, Isolation, and Tokenism: The Problem of (Not) Belonging

Despite increasing numbers of white female graduate students and early-career scholars in certain STEM fields such as biology and veterinary medicine, the STEM disciplines in U.S. research universities have generally remained largely dominated by white men, especially at the senior ranks of the faculty and in executive administration. As STEM slowly diversifies to include white women and domestic URM scholars, members of these groups, particularly women of color, may experience a sense of social exclusion and isolation, as well as tokenism if they are “the only____” (fill in the blank) in their department or campus unit.

The problems associated with being a token are well known (Grey, 2006; Kanter, 1977; Laws, 1975; Niemann, 1999; Yoder, 1991). Tokenism makes individuals hyper-visible and puts them in the position of representing all members of their group to others—what Shohat and Stam (1994) call the burden of representation—as well as educating others about any misperceptions or stereotypical assumptions they might hold. Performance anxiety is heightened because failure will reflect on the group, not just the individual—a stressor that members of dominant identity categories don’t face. Self-doubt, emotional distress, and increased “stereotype threat” (discussed below) may result, potentially compromising performance. In a self-fulfilling prophecy, heightened visibility and anxiety about failure may make failure more likely (see Steele, 1997, 2011).

The isolation of being the only or one of the only representatives of a group may be experienced as a lack of connectedness or belonging to others in the workplace, regardless of performance (Fig. 3). The sense that one doesn’t belong because of the absence of “people like me” can discourage individuals from participating in certain activities or taking risks that could lead to rewards, from remaining in an occupation over time, or even from seeking to enter an occupation in the first place (Niemann, 1999). In this sense, diversity begets diversity, and lack of diversity sustains lack of diversity, because people’s aspirations are affected by who they see (and the positions those people occupy) in their environment.

In the chapter titled ‘Seeing Self’  in this volume, the authors use the term “seeing self” to emphasize the importance of having people of one’s own identity categories not just present in but truly integrated into a workplace. In academia, this not only requires hiring faculty of diverse experiences and backgrounds, but also building networks, partnerships, and shared spaces of knowledge-production that enable members of historically underrepresented communities to thrive.

Fig. 3
figure 3

Sense of exclusion versus belonging. Illustration by Chastine Leora Madla

7.2 The Ideal Worker Norm

Token or superficial efforts at diversity enable workplace cultures to resist meaningful change. The persistence of what sociologists have termed “the ideal worker norm” (see Acker, 1990; Williams, 2000) is a good example. If a biochemistry department consists primarily of white male faculty and the culture of the workplace is to prioritize research above everything else in life, as if one had no familial obligations outside the lab, scientists who have such obligations will be disadvantaged; instead of the workplace changing to accommodate realities of having a family or other care-work responsibilities, the individual worker will either have to struggle to conform to the norm or risk being seen as less dedicated to the job. The individual has to make a “choice” that his or her colleagues don’t have to make.

Like other women workers, women scientists, especially those with children, struggle far more than men to embody the ideal worker norm because of the deeply-entrenched gendered division of labor in the broader society that assigns housework and childcare to women; indeed, the norm itself presupposes this gendered division of labor. The separation of work (paid) and home (unpaid), combined with the lack of high-quality affordable daycare, creates a “second shift” for many working women (Hochschild, 1989/2012). Meanwhile, unless they embrace an equal share of housework and parenting, male employees with children get the benefits of fatherhood without violating the expectation of dedication to work. This social arrangement is also racialized. Consider that, because racial inequality has a strong economic dimension, white and Asian-American men are more likely than black or Latino men to earn salaries that support a stay-at-home partner (whose unpaid labor, in turn, upholds the masculinization of the “ideal” worker), and that white women are more likely able to afford outsourcing domestic labor to other women, typically black and brown women. Consider too, in the case of Latinas, especially those of working-class origins, that the expectations, commitments, complexities, pleasures, and rewards of la familia may play a stronger role in their identity-formation compared to their white or otherwise non-Latina female counterparts (see Hurtado, 2003; Gutierrez et al., 2012).

Parental leave and other work-life balance policies are designed to challenge the ideal worker norm in academia and elsewhere, and they are critical interventions, as discussed in the Chapter, ‘Work Life Integration in Academia: From Myth to Reality.’ But until fathers contribute equally to raising children (and everything else associated with family life), men won’t need such policies as much as women and they can continue to embody the ideal worker. The strength and persistence of the ideal is underscored by studies that show women with children suffer a “motherhood penalty” in the labor force while men with children enjoy a “fatherhood bonus” (Budig, 2017; Budig & England, 2001; Gough & Noonan, 2013). Data further show that married fathers earn 4–7% more than married men without children, and that premium shrinks or disappears when you control for race, meaning the father premium is larger for white men than for men of color (Glauber, 2008; Hodges & Budig, 2010; Killewald, 2013).

7.3 Denying Positionality

Clearly, “diversifying” academia is not a matter of sprinkling a few white women or URM scholars into otherwise unchanged environments; the norms, values, and practices that created and continue to sustain white male homogeneity in the first place have to shift. This poses a particular challenge in many STEM fields because of the strong belief that science, when conducted properly, is neutral, objective, and value-free. In combination with the ideal worker norm, this construction of science can hinder diversity in two ways. First, it suggests that social identity is irrelevant to conducting good science and so it doesn’t matter what race or gender the scientist is. Second, it obscures the fact that, in Western cultures, objectivity and neutrality have been (and often still are) associated with masculinity, especially white masculinity; indeed, they help to construct the very definition and meaning of masculinity itself (see Keller, 1985; Harding, 1986; Hubbard, 1990; Martin, 2001; Fausto-Sterling, 2000/2020; Alcoff &Potter, 2013). Culturally speaking, apparently oppositional qualities such as objectivity/subjectivity, reason/emotion, and logic/intuition are deeply gendered, mapping onto and reinforcing the binary construction of men/masculinity and women/femininity in predictable ways.

The point here is not that scientists shouldn’t aim to conduct careful, rigorous research that is as objective as possible, or that women and men of color aren’t capable of conducting such research, but that the very ideals of the scientific enterprise themselves aren’t neutral with respect to social identity—they more “naturally” align with some people than others. This is important to acknowledge and confront, especially because the group generally advantaged by this alignment—middle-class white men—may not see themselves as occupying a social location at all. As mentioned earlier, an important aspect of privilege is appearing to be unmarked or neutral with respect to identity; this is because dominant- or majority-group members represent the implicit norm against which others are “marked” and measured. Their “normalcy” is part of their privilege. Consequently, privilege is difficult to see for those who possess it, and what appears to them as a neutral environment may actually be supportive and inclusive for them—and, conversely, unsupportive or alienating for everyone else (Fig. 4). Positionality is the willingness and ability to consider how your own social location—the constellation of identity categories that you occupy—may influence how you view and move through the world, whether with ease or great struggle.

Fig. 4
figure 4

The invisibility of privilege. Illustration by Mengmeng Luo

7.4 Stereotypes and Stereotyping

Stereotypes are over-simplified, and often distorted, typifications or ideas of what people are like (Fig. 5). When we stereotype, we think in terms of fixed and inflexible categories. Individual behavior is perceived and understood in terms of these categories; relatedly, stereotypes associated with a group are presumed applicable to all members of that group. Stereotypes abound, but some are more culturally familiar and consequential: men as “naturally” strong and competent leaders, women as “naturally” nurturing and caring, Asian-Americans as “naturally” inclined toward science, African Americans as “naturally” gifted at sports, etc. Of course, some men are competent leaders, some women are nurturing, some Asian-Americans excel at science, and some African Americans are impressive athletes. The problem is that stereotypes limit our perspective, preventing us from seeing the full range of who and what people are. When stereotypes are pejorative and amplified across multiple social fields to justify systematic subordination, they become “controlling images” (Collins, 1990/2000) as discussed earlier: the black woman as jezebel, the Asian woman as exotic lotus blossom, the “fiery” Latina, the dangerous, threatening black man, etc.

Stereotypes are thus cognitive shortcuts that may or may not be rooted in reality; regardless, they have consequences for those at whom they’re directed. “Stereotype promise” (also called “stereotype lift”) is a phenomenon where being viewed through the lens of a positive stereotype can lead to a strong performance that confirms the stereotype; conversely, “stereotype threat” is a phenomenon where being viewed through the lens of a negative stereotype can lead to a poor performance that confirms the stereotype. Claude Steele (1997) developed the concept of stereotype threat in an effort to understand and explain the effect of negative stereotyping on the academic performance of black students, especially in the context of standardized test-taking. Anxiety about the stereotype creates cognitive stress; as a task becomes more difficult, people waste time and energy by focusing on the implications of the stereotype, potentially amplifying it. The concept has since become familiar across different levels of academia. The prevalent stereotype that Latinx and black faculty are less intellectually capable than their white counterparts makes faculty from these groups particularly vulnerable to the self-undermining effects of stereotype threat. As research has shown, low expectations don’t need to be directly communicated to individuals to impair their performance (see Spencer et al., 2016); moreover, making different identities salient has different effects. Consider the following example provided by Stewart and Valian (2018): if Asian women are subtly primed with cues about their ethnicity, they experience stereotype lift, but if subtly primed with cues about their gender, they experience stereotype threat.

Stereotypes are but one of many cognitive shortcuts that people rely on to help manage the swirl of external stimuli; they are a natural way for human brains to work. When internalized, stereotypes may become building blocks for implicit bias, or what Stewart and Valian (2018) call schemas; when externalized and acted upon, they may become building blocks for microaggressions.

Fig. 5
figure 5

Stereotyping and loss of self. Illustration by Meghan Crebbin-Coates

7.5 Implicit Bias

Closely linked to the concept of stereotyping is the concept of implicit bias, a subcategory of implicit social cognition. Implicit social cognition refers to thoughts and feelings outside of conscious intention, awareness, or control (see Greenwald & Banaji, 1995; Nosek & Greenwald, 2009; Nosek et al., 2011). When unconscious or subconscious thoughts reproduce stereotypes based on social-identity categories such as race and gender, we talk of implicit bias (Fig. 6). One common measure of implicit bias is the IAT—the Implicit Association Test, which measures subconscious beliefs by comparing how quickly we can make connections between items; for example, people very quickly link male names to the word “mechanic” and female names to the word “secretary,” but not vice versa. Similarly, when presented with a random list containing masculine, feminine, and gender-neutral words and later asked to recall them, people would cluster the words by gender, sometimes even adding gendered words that weren’t on the original list (Bem, 1981). The IAT reveals that when people have to make decisions quickly, virtually all of us revert to stereotypical thinking. Going against the grain of stereotypical thinking requires time for “shifting gears.”

Of course, having thoughts is not the same as acting on them—possessing an implicit bias does not guarantee it will influence behavior. Yet we know that often it does.

One study showed that doctors with greater implicit racial bias were less likely to recommend an appropriate treatment for coronary heart disease to black patients than doctors with less implicit racial bias (Green et al., 2007); another showed that black convicted felons with stereotypically African features were more likely to receive death sentences than those without such features (Eberhardt et al., 2006). Studies have also revealed race and gender bias in the job application process. Bertrand and Mullainathan (2004) found that resumes with white-sounding names were 50% more likely to receive a callback for an interview than the same resumes with black-sounding names. Similarly, Moss-Racusin et al. (2012) asked 127 STEM faculty to evaluate resumes for a lab manager position that differed only in name—John versus Jennifer. Despite their identical qualifications, Jennifer was perceived as significantly less competent, less mentorable, and deserving of a lower salary than John. Consequently, had Jennifer been a real applicant, she would have faced a number of disadvantages that would have hindered her career advancement.

These are experiments, but real-world examples abound, too. Consider the “blind” orchestral audition. Shifting to blind auditions—where musicians play behind a large screen and their footsteps crossing stage are muffled by carpets or the removal of shoes—was meant to reduce bias, and it did, substantially increasing the likelihood that women would advance to the final rounds (Goldin & Rouse, 2000). Consider too, Kristen Schilt’s research on transgender men (male-identified men assigned female at birth), who transitioned while on the job. The men reported getting more respect, better assignments, more opportunities for promotion, and higher evaluations after transitioning than before—but only if they were white. Transmen of color did not report such advantages (Schilt, 2006). There are countless additional examples of gender and racial bias in the workplace, including in academic STEM fields vis a vis recruitment, hiring, evaluation, and promotion of faculty. Job advertisements, letters of recommendation, criteria for merit, interview protocols, and the like have all been rich data for challenging claims to objectivity and neutrality (for an excellent overview, see Stewart & Valian, 2018; see also Bilimoria & Lang, 2011).

Sociologically speaking, for women, the issue is not just implicit sexism (valuing men over women) but also implicit androcentrism (valuing masculinity, as a constellation of traits, over femininity, as a constellation of traits). Androcentrism helps explain why daughters who are tomboys are less worrisome to parents than sons who are “sissies” (Kane, 2012), or why women stereotypically associated with masculinity (such as “butch” lesbians) fare better in male-dominated construction jobs than do stereotypically feminine women (see Paap, 2006). “Too much” masculinity in women, however, is socially undesirable. Implicit androcentrism means that, for women as a group, “doing gender” is a balancing act: one has to embrace some aspects of masculinity to be perceived as competent and taken seriously, especially in the workplace, but too much masculinity—too much strength, assertiveness, or competitiveness—must be tempered with some femininity—deference, nurturance, empathy—in order to avoid censure and reassure everyone (oneself included, perhaps) that the gender order is still intact. Men, on the other hand, can get away with doing little or no femininity at all—although many men today reject this option, especially in relation to parenting. The balancing act creates a double bind for women workers, in that competence and femininity are seen as mutually exclusive.

The question then arises, if implicit bias is implicit—that is unconscious, subconscious, or semi-conscious, how can we overcome it? Can we overcome it? As Nosek and Riskind (2012) observe, simply telling people not to be biased won’t prevent biased behavior if people don’t recognize (a) they have biases or (b) their biases influence action. To complicate matters, believing that one is objective may actually increase the likelihood of biased behavior because, whatever the behavior, it is seen as stemming from a place of objectivity (Lindner et al., 2011; Uhlmann & Cohen, 2007). Although there is no sure way to eliminate implicit bias, it appears possible to reduce its incidence. The first step is to understand what implicit bias is and how it works—nothing can be accomplished unless we acknowledge it is real. Beyond that, strategies include exposure to counter stereotypes (Dasgupta & Greenwald, 2001); increasing motivation to respond without prejudice (Devine et al., 2002); following prescriptive standards that are meant to reduce bias (Johnson, 2003); comparing identity-conscious versus identity-blind evaluation strategies to ascertain whether implicit bias is a contributing factor; and ensuring a diverse group of evaluators, conceptualizing “diversity” in ways relevant to the evaluation (Nosek & Riskind, 2012).

Fig. 6
figure 6

Implicit bias: thoughts and feelings outside of conscious awareness. Illustration by Mengmeng Luo

7.6 Microaggressions

The study of microaggressions, like that of implicit bias, comes primarily out of psychology. Microaggressions are subtle forms of discrimination, often prompted by stereotypes consciously or unconsciously held, that communicate hostile, derogatory, or insulting messages to the target individual or group (Nadal, 2011; Hernandez et al., 2010; Sue, 2010; Sue et al., 2007); microaggressions, whether intention or unintentional, are “brief and commonplace daily verbal, behavioral, and environmental indignities” (Sue et al., 2007: 273). In effect, microaggressions are implicit biases made visible. They are so pervasive and automatic that they can be dismissed as innocuous or inconsequential, but they have real detrimental effects “because they impair performance across a multitude of settings by sapping the psychic and spiritual energy of recipients and by creating inequities” (Sue et al., 2007: 273; see also Franklin, 2014) (Fig. 7).

In their research on racial microaggressions, Sue et al. (2007) distinguish between microassaults, microinsults, and microinvalidations, each representing diminishing levels of “overtness.” As the most overt, microassaults can manifest as verbal and nonverbal attacks, avoidance, or purposefully discriminatory action. Examples include using racial epithets, deliberately serving a white patron before someone of color, and displaying a swastika. Microinsults are rude or insensitive remarks that demean a person’s racial heritage or identity—as when an employer asks a person of color “how did you get your job?” (implying he or she isn’t qualified and/or the job was not obtained meritoriously) or when a white teacher fails to acknowledge students of color in the classroom. Microinvalidations occur when a person negates or denies the thoughts, feelings, or experiences of a person of color—complimenting US-born Latinx or Asian Americans on their use of English (implying they are perpetual foreigners), or saying to a person or color “I don’t see race,” or “we’re all just human,” which invalidates the person’s experience of racism in their everyday lives.

Gendered microaggressions tend to be either sexual in nature (comments about appearance or clothing) or aimed at challenging women’s competence, capability, and motivation (comments expressing doubt about whether women can handle a task or assignment—also characterized in the sociological literature of “benevolent sexism.”) Both forms tend to increase as the representation of women in a field or occupation decreases (Allan & Madden, 2006). Gendered microaggressions constitute subtle expressions of sexism that stand in contrast to what gender scholars call “hostile sexism,” which is blatant—the use of harassments, threats, and violence to enforce women’s subordination. Like overt racism, overt sexism is less socially acceptable today than in the past; it hasn’t disappeared, however, but rather is driven underground and manifests in more subtle ways. As Basford et al. (2014) note, subtle discrimination creates the kind of uncertainty about perceptions of prejudice associated with anxiety and depression (see also Sue et al., 2008; Banks et al., 2006; Noh et al., 2007) as well as affecting job satisfaction and workplace commitment (see Foley et al., 2005).

Research shows that both racial and gender microaggressions tend to express certain themes of exclusion. Sue et al. (2007) identify eight different themes related to race, four of which are especially pertinent to faculty in academic settings: (1) ascribing intelligence on the basis of race, which disadvantages black and Latinx scholars relative to whites and Asians; (2) assuming color blindness/denying individual racism, which invalidates the daily, lived experience of racism; (3) believing the institution is a race-neutral meritocracy, which fails to see how the very criteria of meritocracy advantage dominant groups; (4) environmental messages of exclusion which preclude a sense of belonging, as when a department faculty consists mostly of white men, or all the buildings in an institution are named after white men.

Williams and Hall (2016), in a study gender bias experienced by female STEM faculty, identify four themes representing types of bias that hinder the advancement of women in STEM (they do not specifically use the term “microaggressions,” but their findings are germane). The first theme is Prove it Again, in which women often have to provide more evidence of competence than men in order to be seen as equally competent. Here, women of color face a double jeopardy because their competence may be questioned on the basis of race and gender simultaneously. The second theme, the Tightrope, reflects the imperative that women balance the “masculine qualities” demanded of STEM scientists with enough femininity to avoid being perceived as too angry, assertive, or unfeminine. The tightrope is especially difficult for black women, who, historically, have been perceived as “naturally” less feminine than their white, Latinx, or Asian counterparts. The third and most common theme, the Maternal Wall, reflects the prescriptive bias that motherhood is incompatible with a science career; having children is incorrectly assumed to derail women’s commitment to rigorous scientific investigation, thereby impacting performance. Fourth, the Tug of War refers to the ways in which women may not support—and may even distance themselves from—other women, particularly if experiences of gender discrimination began early in their education or career. Although previous studies attributed this phenomenon to the personality characteristics of senior women, the authors attribute lack of collegial support to the overtly hostile workplace environments senior women were subjected to in the past.

Fig. 7
figure 7

The cumulative impact of microaggressions. Illustration by Meghan Crebbin-Coates

8 Conclusion

Discussions of exclusion, positionality, stereotyping, implicit bias, and microaggressions are by now common features of many diversity training programs in academia. As mentioned earlier, social science and humanities scholars who have spent their careers studying the structural dimensions of social inequality are skeptical of the hefty administrative resources devoted to such training when their own research remains undervalued and underfunded. But unless STEM faculty have a liberal arts background or have deliberately taken it upon themselves to learn about systemic racism, sexism, homophobia, etc., a training or series of trainings on implicit bias or microaggressions might be the best or only exposure they get. Context also matters: academic faculty are generally committed to the ideals of diversity, equity, and inclusion in the workplace and they are generally motivated to realize these ideals. This commitment makes academia quite different from, say, law enforcement in the U.S., whose departments and agencies routinely incorporate implicit racial bias training into their protocols with little effect on stemming the tide of state violence against black and brown communities.

Race, gender, sexual orientation, and other social identity categories are not fixed or unitary. They are multiple and varied, and the identities they give rise to are linked to complex histories as well as contemporary experiences. Experiencing one form of discrimination doesn’t prevent people from reinforcing other forms, any more than understanding one form of discrimination guarantees knowledge of another. Men of color, who experience racism, can be sexist; white women, who experience sexism, can be racist; heterosexual women of color, who experience both racism and sexism, can be homophobic; Asian or Latinx individuals can be anti-black, etc. Moreover, knowing about, and understanding, inequality doesn’t automatically prevent discriminatory behavior, if certain biases are unconscious. Likewise, one can be consciously anti-racist in attitude and interaction while contributing to systemic racism.Footnote 5

Interventions such as the NSF ADVANCE program are not panaceas to systemic social problems, nor are they meant to be. But incentivizing institutional transformation in the academy, in places where it is needed and in ways that faculty will accept, can make a difference. At the same time, as a feminist cultural sociologist, I hope university administrations will see the big picture, and consider programmatic diversity and inclusion efforts in STEM as complementary to and not a substitute for a comprehensive approach to transformation that supports those pockets of the university where the faculty are already both diverse and conducting important diversity-relevant work.