How Does Racism Work?

Introduction

Imagine an employer who sincerely believes that hiring should be fair. The employer never uses a racial slur, never announces a preference for one racial group, and applies the same written criteria to every applicant. Could the hiring process still help reproduce racial inequality?

The answer is: possibly—but we would need evidence. We would need to know how the criteria were chosen, what they measure, how they are applied, what opportunities applicants had before applying, and what outcomes follow. An unequal result alone does not prove racism, just as a stated commitment to fairness does not prove that a process is fair.

This hypothetical hiring process will return throughout this article. It gives us a concrete case to test as we explore a larger question: How can racism operate through individual actions, institutional routines, and social arrangements—even when explicit racial hostility is absent?

Racism is often described as an individual attitude: a person holds a stereotype, dislikes someone because of race, or deliberately treats another person unfairly. Those actions matter. But a sociological explanation also examines how racial meanings become connected to power, how institutions turn assumptions into routine decisions, and how practices interact to produce patterned outcomes. The goal is not to assume that every disparity is racism, but to investigate which mechanisms, if any, explain it. Learn more about Dr. Stokes's background in the biography and explore related scholarship on the research page.

Can racism persist even when people reject racial hatred?

Prejudice refers broadly to attitudes, evaluations, or assumptions about people based on group membership. Discrimination refers to differential treatment. Racism can involve both, but it can also be understood as a wider social process in which racial categories and meanings are connected to unequal power and institutional arrangements.

The distinction matters because a society can reproduce racial inequality even when many individuals reject explicit racial hatred. Eduardo Bonilla-Silva's analysis of structural and color-blind racism describes how contemporary racial orders can operate through subtle, institutional, and seemingly nonracial practices, while racial ideology helps justify or normalize those arrangements (Bonilla-Silva 2006, 2015). Joe Feagin's account of systemic racism likewise emphasizes the connections among historical relations, institutions, racial ideology, and material inequality (Feagin 2006).

Return to the hiring example. One committee member might rely on a stereotype; the committee might use a routine that rewards a narrow kind of experience; and several organizations might reproduce similar patterns over time. These are related possibilities, not interchangeable explanations. To understand what is happening, we need to identify the level at which a process operates and how the levels connect.

Consider three related levels of analysis:

• Interpersonal action: A person makes a decision, uses a stereotype, or treats someone differently.
• Institutional process: A school, employer, court, agency, or other organization uses routines, rules, or standards that shape who receives opportunities, resources, penalties, or protection.
• Systemic pattern: Multiple institutions, policies, social expectations, and historical conditions interact over time, producing or reproducing patterned racial inequality.

Key idea

Racism can involve individual actions, institutional processes, and systemic patterns; these levels are connected but not interchangeable.

These levels overlap, but they are not interchangeable. An individual's action may reflect an institutional norm; an institutional practice may persist even when particular employees change; and several institutions may reinforce one another. A strong explanation investigates the links rather than assuming that any one level explains everything.

If race is socially constructed, how does it become powerful?

Race is not simply a biological fact that automatically dictates social position. Racial categories are socially produced, historically variable ways of classifying people. But classification is not merely a matter of words. When categories become attached to assumptions about intelligence, morality, danger, trustworthiness, competence, or belonging, they can influence how people are treated and which opportunities are available to them.

Michael Omi and Howard Winant developed racial formation theory to explain how race is created, transformed, and made consequential through social and political life. First presented in their 1986 book Racial Formation in the United States, the framework challenges explanations that treat race as a fixed biological fact or reduce it to another category, such as nationality or economic class. Race is neither an unchanging set of natural divisions nor simply an illusion. It is a changing social structure and a system of meanings: racial categories are made and contested historically, and they can shape people's identities, experiences, opportunities, and treatment (Omi and Winant 2014).

For Omi and Winant, racial formation happens through struggles over what racial categories mean and what consequences follow from them. Laws, public policies, political arguments, institutions, and everyday interactions can all help create or change racial meanings. Those meanings are not just ideas floating apart from material life: they can become connected to rules, institutional practices, and the distribution of resources and opportunities. At the same time, the theory does not imply that race works identically in every society or historical period. The categories people use, the meanings attached to them, and their consequences must be understood in context.

The central concept is the racial project. Omi and Winant define a racial project as both an interpretation, representation, or explanation of racial dynamics and an effort to reorganize or redistribute resources along racial lines. In other words, a racial project connects what race is understood to mean with how social life is organized and how people experience it. A project might portray a group as dangerous, deserving, foreign, or especially qualified; that interpretation can then be connected to decisions about surveillance, public support, belonging, hiring, housing, or access to other resources. Not every racial project has the same aims or effects, and identifying one requires examining both its racial meanings and its practical consequences.

Consider a government census. Its categories do not simply record a set of self-evident groups: they reflect choices about which distinctions matter and how people can describe themselves. Those classifications can affect which populations become visible in public debate, how institutions measure inequality, and how policies identify the people they are intended to serve. The example illustrates why racial formation theory matters: it helps us ask who defines racial categories, what meanings those categories carry, how they connect to institutional arrangements and the distribution of resources, and how they are challenged or changed. Anthony Marx's work on race-making and the nation-state likewise examines how racial classification becomes part of the organization of political communities (Marx 1996).

Key idea

Racial meanings become consequential when they connect to power, institutional decisions, and the distribution of resources and opportunities.

Power is the capacity to shape what happens: to influence decisions, define the available choices, establish rules, control resources, set standards, or make some descriptions of the world seem authoritative while others are ignored. Power is not limited to forcing someone to do something. It can also operate by shaping what people regard as normal, reasonable, legitimate, or possible. For example, a hiring manager has direct power to decide whom to interview. An institution also has power when it defines which credentials count as evidence of competence, controls access to professional networks, or sets procedures that determine whose work receives attention.

There is no single definition of power that fits every question and every context. What counts as power, how it is exercised, and what makes it legitimate vary across historical periods and cultural settings. Max Weber's influential definition emphasizes the ability to carry out one's will even in the face of resistance (Weber 1978). That approach helps identify domination and conflict, but if treated as a complete or universal definition, it can leave less visible forms of power in the background: the power to define categories, establish standards, shape desires and expectations, and determine whose knowledge counts. It can also make historically specific arrangements of conquest, colonial rule, and imperial expansion appear as ordinary expressions of power rather than relationships with histories that should be examined. This is not a reason to discard Weber; it is a reason to ask what a particular definition reveals and what it leaves out.

Power matters for racism because people and groups do not have the same capacity to make racial meanings consequential. An individual can express racial prejudice regardless of their own racial identity. But the reach of an action depends on social position and context: who has decision-making authority, what resources they control, what histories and policies shape the situation, and whether their actions are backed by organizational routines or state power. A prejudiced remark and an institutional decision that restricts access to housing, employment, education, or legal protection are not equivalent simply because both involve racial attitudes.

This distinction avoids two opposite mistakes. One is to treat racism as nothing more than individual hostility. The other is to assume that every action by every person has the same structural power. People of any racial group can hold racial stereotypes, express prejudice, or act discriminatorily. In the United States, however, the same statement or action can have different consequences depending on the speaker's social position and the historical and institutional relationships it activates. When a white person expresses a racist idea, that act can draw force from a wider history of white advantage and from institutions in which white people have held disproportionate authority and control over resources. A person of color can also express or act on racist ideas—including against their own group or another group—but that act does not automatically activate the same history, institutional backing, or structural reach. The distinction is not that prejudice or harm from a person of color is impossible or irrelevant; it is that the actions do not necessarily have equivalent social power. The task is to examine how attitudes, authority, institutional arrangements, historical context, and consequences relate in each circumstance, rather than denying individual prejudice outside dominant groups or treating all acts as structurally interchangeable.

Key idea

Racial meanings become consequential when they connect to power, institutional decisions, and the distribution of resources and opportunities.

How can ordinary institutions reproduce inequality?

In our hiring example, the institution is not just the company or the committee room. It also includes the criteria for being shortlisted, the meaning attached to “professionalism,” the sources of referrals, the way interviews are scored, and the discretion evaluators have when interpreting answers.

Institutions are not only buildings or formal organizations. They also include established rules, routines, norms, expectations, categories, and ways of evaluating people. Laws and policies matter, but so do the ordinary practices through which people interpret and apply them.

The state refers to the relatively enduring set of institutions that exercise public authority over a territory and population, including the legal system, courts, police, military, and government bureaucracy. A government is the group of leaders and officials directing those institutions at a particular time. A new president, party, or administration can change government without replacing the state itself. This distinction matters because institutional practices and historical arrangements can continue across changes in political leadership. The state is not automatically neutral: its laws, classifications, enforcement practices, and allocation of resources can help produce or challenge racial inequality (Bracey 2015).

An institution may reproduce inequality through explicit rules. It may also do so through routines that seem unrelated to race, through criteria that reward some experiences more than others, or through decisions that repeatedly favor people who resemble those already recognized as qualified or belonging.

Institutional processes can continue beyond the intentions of any one participant. A hiring committee member may oppose racism but still rely on familiar indicators of competence. A teacher may value fairness but use an evaluation practice without examining whether it measures the skill it claims to measure. A public agency may apply a general rule while overlooking how existing conditions shape who is most exposed to enforcement. These examples do not establish that every such decision is discriminatory. They identify questions that deserve investigation: How was the practice designed? What does it measure? How is it applied? What patterns follow? What alternative explanations are plausible?

Institutional analysis asks how decisions are made repeatedly, what incentives shape them, what assumptions are built into the process, and whether similar outcomes recur across settings. It also examines how one institution's decisions interact with others. Unequal outcomes in education, for example, may reflect not one isolated decision but the cumulative effects of resources, networks, prior opportunities, evaluation practices, and access to later opportunities.

Can a policy have unequal effects without anyone intending discrimination?

Suppose the employer believes that a particular credential is the best evidence of competence. That belief may be sincere. But sincerity does not answer whether the credential predicts job performance, whether it is essential, or whether access to it has been unequal. Intent is relevant: it can help explain why someone acted, whether a decision was deliberate, and what kinds of accountability may be appropriate. But intent alone cannot establish the full social meaning or consequences of an action.

Key idea

Intent matters, but understanding a policy also requires examining its effects, mechanisms, and plausible alternative explanations.

A policy can use the same language for everyone and still operate in a context where people have unequal resources or face different levels of scrutiny. A person may unintentionally repeat a stereotype or apply an assumption unevenly. An organization may sincerely endorse equal treatment while retaining procedures that produce persistent disparities.

The correct response is neither to ignore intent nor to assume that an unequal outcome proves discriminatory intent. Laws, policies, and actions can have consequences their authors did not anticipate or intend, sometimes because a rule interacts with existing inequalities or because the way it is implemented differs from the way it was described. A policy may be written in the same terms for everyone yet place heavier burdens on some groups because people begin from different circumstances or encounter different levels of enforcement. An unintended effect is not, by itself, proof of racism; its causes still need to be established. But intent cannot be the only test, because a policy can contribute to racial inequality regardless of whether its authors meant to produce that result.

It is therefore important to examine both intent and consequences. What did the decision-maker intend? What did they do? What effects followed? Are the effects patterned by race? What mechanism could explain that pattern? Does evidence support that mechanism, and what other explanations should be considered? Intent can help explain responsibility and the kind of response that is appropriate, while evidence about effects and mechanisms helps establish what a practice actually does.

Research can help answer these questions. In a widely cited field experiment, Bertrand and Mullainathan (2004) sent fictitious resumes to help-wanted advertisements in Boston and Chicago, randomly assigning names associated with different racial groups to otherwise comparable resumes. Resumes with White-associated names received more callbacks than those with African American-associated names. Because the name assignment was randomized, the study provides evidence that perceived racial signals affected employer responses in the settings studied. It does not prove that every employer discriminates, nor does one experiment explain every labor-market disparity. Its value is more precise: it demonstrates how a racialized signal can affect an early opportunity even when the resumes are otherwise comparable.

If the same rules apply to everyone, why examine them?

The hiring example makes the issue concrete: perhaps the employer values a prestigious degree, years of experience, and a particular interview style. Any of these might provide useful evidence, but none should be treated as self-justifying. What ability is each criterion supposed to measure, and does it actually measure that ability?

Some standards are necessary. Employers need to assess whether applicants can perform a job; schools need to assess learning; professional bodies need to evaluate competence. The question is not whether standards should exist. It is how particular standards are defined, justified, measured, and applied—and what they actually do.

Key idea

A rule's stated neutrality does not establish that it is valid, consistently applied, or fair in practice.

A standard may be formally identical for everyone while interacting with unequal opportunities, resources, or institutional expectations. It may reward skills genuinely relevant to a task, but it may also reward familiarity with the culture of the organization, access to prestigious institutions, or opportunities that were unequally distributed. These possibilities need to be investigated, not presumed.

A useful evaluation asks:

• Purpose: What ability, knowledge, or behavior is the standard supposed to measure?
• Validity: Is there evidence that the criterion measures what it claims to measure?
• Opportunity: Have people had reasonably comparable access to the resources or experiences required to meet it?
• Application: Is the criterion applied consistently, or do decision-makers interpret it differently across candidates or groups?
• Consequences: What patterns emerge, and what mechanisms plausibly produce them?
• Alternatives: Could the same legitimate goal be achieved with a more valid or less exclusionary measure?

A disparity is a reason to investigate, not a shortcut around investigation. Conversely, a claim of neutrality is not proof that a practice is fair. Both the asserted purpose and the real-world operation of the standard matter.

Is merit always what we think it is?

Imagine that the employer says, “We hired the most meritorious applicant.” That may be true—but the statement does not, by itself, tell us how merit was defined, what evidence counted, or whether the measures were appropriate.

Merit is often treated as though it were self-defining: the most qualified person is simply the person with the most merit. In practice, institutions must decide what counts as merit and how it will be measured. In that sense, merit is socially constructed: people and institutions select the qualities they value, decide how to recognize them, and establish which evidence will count. Those choices can be reasonable and relevant to a task, but they are not culturally neutral by default. Ideas about excellent writing, professional demeanor, leadership, communication, or the right kind of experience can reflect the values and expectations of some cultural settings more than others.

This does not mean that ability, effort, or competence are imaginary, or that every standard is unfair. It means that claims to evaluate people purely on merit can conceal the choices built into the definition and measurement of merit. They can also reproduce inequality when the valued credentials or experiences are easier to obtain for people who already have access to money, time, mentorship, networks, institutional prestige, or opportunities. A criterion can measure something real and still be incomplete, poorly matched to the task, or applied in ways that preserve existing advantages. Sometimes what is called “merit” is not a reliable measure of the ability that matters for the decision. Merit should therefore be examined rather than simply accepted or dismissed.

A publication record, degree, award, test score, grant, or prestigious institutional affiliation may reflect real ability and effort. It may also reflect access to time, funding, mentorship, networks, institutional resources, and earlier opportunities. Recognizing those structural influences does not mean denying individual achievement. It means evaluating achievement in context and asking whether the selected indicators are appropriate for the decision being made.

Key idea

Merit is defined and measured through choices; those choices should be evaluated rather than automatically accepted or dismissed.

Wang's study of academic gatekeeping, based on interviews with 50 sociology and business professors across seven U.S. research universities, examines how merit, fit, and diversity shape the evaluation of tenure-track candidates. It finds that apparently objective signals of merit can obscure structural constraints, while different evaluation logics can jointly affect selection (Wang 2025). Because the study focuses on a specific set of research universities and disciplines, its findings should not be assumed to describe every hiring process. It nonetheless provides a concrete example of how evaluative standards can be socially organized.

The key questions are practical: What does this credential predict? Is a particular measure essential for the role? Are candidates being compared in light of different opportunities and responsibilities? Are evaluators relying on a single quick signal when multiple forms of evidence could better establish competence?

The point is not that merit is meaningless. The point is that merit is defined and measured through social choices, and those choices should be open to scrutiny. We should ask what a criterion actually measures, whether it is relevant to the task, whose cultural expectations it reflects, what opportunities made the achievement possible, and whether the claim of merit is being used to explain a decision or to shield that decision from examination. Merit matters—but the label alone does not prove that an evaluation is valid or fair.

When does “fit” become a way to keep people out?

“Fit” can refer to a legitimate question: can someone work effectively in a role, collaborate with colleagues, or contribute to the institution's mission? But fit can also become a vague judgment about whether someone feels familiar, comfortable, or similar to the people who already belong.

This distinction is especially important when decision-makers assess communication style, demeanor, self-presentation, collegiality, cultural interests, or a candidate's sense of belonging. Some of these qualities may be relevant in particular roles. Others may act as proxies for familiarity with dominant organizational norms. Without clear criteria, “fit” can be difficult to evaluate consistently and can reward insiders while excluding people who would contribute effectively but do not resemble existing members.

Gatekeeping occurs at points where people control access to opportunities: who receives an interview, whose work is treated as promising, who is admitted, who is promoted, and who is viewed as belonging. Social closure describes processes through which groups restrict access to opportunities or advantages. These processes need not involve an explicit declaration that a group should be excluded. They can operate through criteria, networks, informal judgments, and repeated decisions that define who counts as qualified or suitable.

Wang (2025) connects merit, fit, and diversity to different stages of academic hiring and argues that they can interact to shape inclusion and exclusion. Sensoy and DiAngelo (2017) likewise examine how faculty hiring committees can reproduce established patterns even while affirming support for diversity. Together, these sources support a focused question: can an institution's stated values and its everyday selection practices pull in different directions?

Key idea

Fit is most defensible when it is clearly defined, relevant to the role, and assessed consistently.

The answer should be established through evidence in each case. The presence of a fit criterion does not automatically prove bias. Strong analysis specifies what fit means, whether it is connected to the role, how it is assessed, and what patterns its application produces.

How do we learn to see some people as qualified, dangerous, or out of place?

People make the social world intelligible by categorizing, naming, and interpreting what they encounter. Recognition involves taking someone or something as a particular kind of person, object, group, or situation. These processes are necessary to ordinary social life. They become consequential when categories carry unequal meanings, when others' self-understandings are overridden, or when socially produced classifications are treated as natural and inevitable.

Misrecognition occurs when a person, group, or situation is wrongly understood or classified. In racial contexts, it can include treating socially assigned group characteristics as inherent qualities: assuming that a racial category determines intelligence, morality, danger, ability, or leadership. It can also involve failing to recognize the operation or effects of a racialized practice because a familiar institutional routine is taken to be neutral.

Recognition and misrecognition operate at different levels. People learn meanings through families, peers, education, media, public institutions, and everyday interaction. Organizations formalize categories in records and decisions. States can define or enforce categories through laws and policy. These levels interact, although no one level fully determines every person's beliefs or actions.

This framework helps explain how people can reproduce racial meanings without consciously choosing to do so. It also points toward a method of inquiry: ask who is being classified, who has the authority to classify, what meanings accompany the category, what consequences follow, and what evidence might challenge the classification.

Does race operate separately from other forms of inequality?

Racism does not operate in isolation from gender, class, sexuality, nationality, Indigeneity, religion, or other relations of power. These relationships vary by context. An intersectional approach asks how systems of inequality interact to shape specific experiences and outcomes rather than treating each category as a separate variable that always works the same way.

Collins (1998), for example, examines the mutual construction of gender, race, and nation. In practice, a racialized expectation may also be gendered; ideas about national belonging may affect how people are classified; and economic position may shape access to the resources needed to challenge unequal treatment. These are questions to investigate in context, not a claim that all forms of inequality are identical.

The practical implication is to avoid explanations that are too narrow. A particular racial disparity may be shaped by multiple institutions and social relationships. Understanding those interactions requires specifying the case and evidence rather than assuming that one general explanation applies everywhere.

How can we tell whether a practice is reproducing racial inequality?

At the end of our hypothetical hiring process, we should not jump from a disparity to a verdict, or from a claim of neutrality to an acquittal. We should examine the actual process and test explanations against evidence. The following questions offer a practical way to do that:

Key idea

A disparity is a reason to investigate, not a complete causal explanation; test proposed mechanisms against evidence.

1. Define the practice. What exactly is the rule, decision, standard, or behavior?
2. Identify its stated purpose. What is it intended to accomplish?
3. Trace its history. How was it created, and what conditions shaped it?
4. Examine implementation. Who applies it, with what discretion, resources, and accountability?
5. Look at outcomes. Who receives opportunities, penalties, resources, or protection? Are there recurring patterns?
6. Test explanations. What evidence supports the proposed mechanism? What alternative explanations remain plausible?
7. Consider revisions. Can the legitimate purpose be achieved more accurately or fairly? How would changes be evaluated?

This framework avoids two shortcuts: assuming that a formally neutral rule is necessarily fair, and assuming that every unequal outcome proves intentional discrimination. It directs attention toward evidence, mechanisms, and consequences.

So how does racism work?

Could the employer's process reproduce racial inequality even without explicit racial hostility? It could—but the answer depends on the evidence about the criteria, how they are applied, the opportunities available to applicants, and the outcomes they produce. The example illustrates the method, not a verdict about a real employer.

More broadly, racism can be reproduced through individual conduct, institutional routines, social expectations, and the interaction of multiple systems. It does not always require conscious hostility, and it cannot be fully understood by examining intentions alone. But neither should the concept be used as a substitute for evidence. Strong analysis identifies the process at issue, explains how it works, and evaluates its consequences in context.

The central task is to examine how racial meanings become connected to power, how classifications become consequential, how standards and institutions operate in practice, and how patterns of unequal treatment are sustained or challenged. This approach makes room for accountability without reducing the explanation to individual blame, and for institutional critique without assuming that every disparity has the same cause.

Next: Why Does Racism Still Exist? · Race and Racism: A Sociological Guide

Selected references

• Bonilla-Silva, Eduardo. 2006. Racism without Racists: Color-Blind Racism and the Persistence of Racial Inequality in the United States. 2nd ed. Lanham, MD: Rowman & Littlefield.
• Bonilla-Silva, Eduardo. 2015. “The Structure of Racism in Color-Blind, ‘Post-Racial’ America.” American Behavioral Scientist 59(11):1358–1376. https://doi.org/10.1177/0002764215586826
• Bertrand, Marianne, and Sendhil Mullainathan. 2004. “Are Emily and Greg More Employable Than Lakisha and Jamal? A Field Experiment on Labor Market Discrimination.” American Economic Review 94(4):991–1013. https://doi.org/10.1257/0002828042002561
• Brubaker, Rogers, Mara Loveman, and Peter Stamatov. 2004. “Ethnicity as Cognition.” Theory and Society 33(1):31–64.
• Collins, Patricia Hill. 1998. “It’s All in the Family: Intersections of Gender, Race, and Nation.” Hypatia 13(3):62–82.
• Feagin, Joe R. 2006. Systemic Racism: A Theory of Oppression. New York: Routledge.
• Marx, Anthony W. 1996. “Race-Making and the Nation-State.” World Politics 48(2):180–208.
• Omi, Michael, and Howard Winant. 2014. Racial Formation in the United States. 3rd ed. New York: Routledge.
• Sensoy, Özlem, and Robin DiAngelo. 2017. “‘We Are All for Diversity, but . . .’: How Faculty Hiring Committees Reproduce Whiteness and Practical Suggestions for How They Can Change.” Harvard Educational Review 87(4).
• Weber, Max. 1978. Economy and Society: An Outline of Interpretive Sociology. Edited by Guenther Roth and Claus Wittich. Berkeley: University of California Press.
• Wang, Leping. 2025. “Unpacking Merit, Fit, and Diversity: A Multifaceted Framework to Academic Gatekeeping in Social Sciences at U.S. R1 Research Universities.” Sociological Inquiry. https://doi.org/10.1111/soin.70026