Discriminatory behaviour at work is not a single phenomenon. It operates across a spectrum that runs from explicit statutory violations captured in federal enforcement data to covert behavioural patterns that accumulate over years and reshape career trajectories without ever producing a formal complaint. Understanding that full spectrum, and building organisational systems that address it at every point, is the core technical challenge for DEI practitioners and HR professionals who are serious about producing equitable workplaces rather than compliant ones.
This article covers the legal architecture that defines workplace discrimination in the United States, the distinct behavioural categories through which it manifests, the structural conditions that allow it to persist, the data on what it costs organisations and individuals, and the practitioner-level framework for identifying and eliminating it.
Before any practitioner can classify or respond to discriminatory behaviour, they need to understand the statutory categories that define it under federal law. The U.S. Equal Employment Opportunity Commission enforces seven principal federal statutes covering workplace discrimination.
Title VII of the Civil Rights Act of 1964 is the foundational statute. It prohibits discrimination in hiring, firing, pay, promotions, job assignments, training, and any other term or condition of employment on the basis of race, color, religion, sex, or national origin. Title VII covers private employers with 15 or more employees, labour organisations, employment agencies, and most federal, state, and local government employers.
The Age Discrimination in Employment Act of 1967 prohibits discrimination against workers who are 40 years of age or older. The Americans with Disabilities Act of 1990 prohibits discrimination against qualified individuals with disabilities and requires employers to provide reasonable accommodation unless doing so would impose undue hardship. The Equal Pay Act of 1963 requires that men and women receive equal pay for substantially equal work performed under similar working conditions. The Genetic Information Nondiscrimination Act of 2008 prohibits discrimination based on genetic information, including an employee’s family medical history. The Pregnant Workers Fairness Act, which became effective in June 2023, requires employers to provide reasonable accommodations to employees with limitations related to pregnancy, childbirth, or related medical conditions.
These statutes establish what the law prohibits. They do not exhaustively describe the forms that prohibited behaviour takes, which is a critical distinction. The absence of an explicit statutory category does not mean an organisation has no obligation to address a behaviour. It means the behaviour has not yet been litigated to the point where its legal category is fully defined.
Federal law recognises two distinct theories under which an employment practice can constitute unlawful discrimination. Each describes a different mechanism of harm and requires a different evidentiary approach to establish.
Disparate treatment is intentional discrimination. It occurs when an employer treats an individual differently because of their membership in a protected class. A hiring manager who declines to interview qualified candidates with names that signal a particular racial or ethnic background is engaging in disparate treatment. A supervisor who passes over a female employee for a leadership role because of assumptions about her family commitments is engaging in disparate treatment. Intent does not need to be consciously acknowledged by the actor to be legally actionable. The legal standard looks at whether the protected characteristic was a motivating factor in the decision, not whether the decision-maker was aware of or admitted to that motivation.
Disparate impact is discrimination through facially neutral practices. A policy, standard, or selection criterion that does not appear on its face to discriminate on a protected basis but that produces significantly different rates of exclusion across demographic groups may constitute unlawful discrimination if the employer cannot demonstrate that the practice is job-related and consistent with business necessity. The classic example is a height or physical strength requirement that disproportionately excludes women or individuals from particular ethnic backgrounds from employment in roles where those requirements are not genuinely necessary for job performance.
Understanding how bias operates at both the intentional and structural levels is the practitioner prerequisite for building interventions that address the actual mechanisms generating discriminatory outcomes rather than their most visible surface expressions.
The EEOC received 88,531 new charges of employment discrimination in fiscal year 2024, a 9.2 percent increase over fiscal year 2023 and part of a sustained upward trend over multiple years. The agency recovered nearly 700 million dollars for approximately 21,000 victims of discrimination across private, state and local government, and federal workplaces that year. That 700 million dollars represents the highest monetary recovery in recent EEOC history and reflects the compounding costs that organisations incur when discriminatory behaviour is allowed to operate without correction.
Retaliation consistently accounts for the largest share of EEOC charge bases, appearing in close to 50 percent of all charges filed. This is not primarily because retaliation is the most common form of initial discriminatory behaviour. It is because retaliation charges frequently accompany or follow charges on other bases. When an employee files a complaint about race, sex, age, or disability discrimination and subsequently experiences adverse employment action, they file a retaliation charge. The high rate of retaliation charges is a structural indicator of how organisations respond to complaint activity, and it tells practitioners that the discriminatory behaviour problem at most organisations does not end when someone reports it. For many employees, it intensifies.
The sustained increase in discrimination-related lawsuits against organisations in the U.S. is not a legal anomaly. It is the downstream consequence of behavioural and structural patterns that have been left unaddressed inside organisations for long enough to become legally visible.
The range of discriminatory behaviour that DEI practitioners encounter in real organisations spans several distinct categories. Each operates through different mechanisms and requires different detection and response strategies.
Overt discriminatory behaviour involves explicit, direct conduct based on a protected characteristic. Slurs, derogatory comments, exclusion from meetings or communications based on demographic membership, denial of opportunities that are explicitly tied to an employee’s race, sex, age, religion, national origin, or disability, and hostile physical conduct all fall into this category. Overt discrimination is the most legally straightforward form to establish, because the conduct itself serves as evidence of the discriminatory intent.
Overt racism in the workplace does not look like a relic of a prior era. It continues to appear across industries and organisational levels, including senior leadership environments where its presence is least expected and most damaging to the broader culture signal that leadership behaviour sends.
Overt discrimination is also the category that generates the fastest escalation to formal legal action when it is not addressed promptly. Organisations that allow overt discriminatory conduct to persist or that respond to complaints about it in ways that are perceived as protective of the actor rather than the target create the conditions for hostile work environment claims under Title VII and for the retaliation charges that follow when reporting employees face adverse consequences.
Harassment is a specific legal category of discriminatory behaviour defined by the EEOC as unwelcome conduct based on a protected characteristic that is either severe or pervasive enough to create a work environment that a reasonable person would consider intimidating, hostile, or abusive, or that results in an adverse employment action. The legal threshold of severe or pervasive is a composite standard. A single extremely severe incident, such as a physical assault, can meet it. A pattern of less severe conduct that occurs frequently enough to alter the conditions of employment can also meet it.
The EEOC’s Strategic Enforcement Plan for fiscal years 2024 through 2028 documents that total harassment charges increased from 21,270 in fiscal year 2021 to 24,430 in fiscal year 2022. Race-based harassment charges rose from 7,755 to 8,524 over the same period. Sexual harassment charges increased from 5,581 to 6,201. These are charges filed with the federal agency. They represent a subset of harassment incidents that occur in workplaces, because research consistently shows that the majority of employees who experience harassment do not file formal complaints. The charge data captures the cases where the behaviour was severe or sustained enough, and the employee motivated and able enough, to initiate federal agency contact.
Both quid pro quo harassment, in which submission to harassment is made a condition of employment or employment decisions, and hostile work environment harassment are prohibited under Title VII. Harassment does not have to come from a direct supervisor. It can come from colleagues, subordinates, clients, or customers. The employer’s liability depends on who committed the harassment and whether the employer took prompt corrective action upon learning of it.
Covert discriminatory behaviour operates through mechanisms that are individually ambiguous but collectively systematic. This is the category that generates the widest gap between what employees experience and what organisations detect and address.
The spectrum from overt to covert discrimination in the workplace is not a gradient of severity. Covert forms often cause more sustained damage precisely because they are harder to name, harder to report, harder to investigate, and harder to correct through individual-level interventions. They include differential assignment of high-visibility versus low-visibility projects across demographic lines, inconsistent application of performance standards, exclusion from informal information networks, differential access to mentorship and sponsorship, and evaluation language that describes the same professional behaviours differently depending on who exhibits them.
None of these behaviours, in isolation, necessarily rises to the legal threshold of actionable discrimination. In combination, applied consistently across a cohort of employees over multiple years, they produce promotion gaps, compensation disparities, and retention differentials that are measurable in aggregated workforce data even when no individual incident is individually actionable.
Unconscious bias is the cognitive mechanism that produces much of this covert discriminatory behaviour. Ingroup favouritism, attribution error, affinity bias, and confirmation bias each influence how managers evaluate performance, allocate resources, and make advancement decisions. They do so without requiring conscious prejudice and without producing the explicit conduct that overt discrimination generates. Addressing them requires structural interventions in decision-making processes, not appeals to individual goodwill.
Systemic or institutional discrimination describes the condition in which discriminatory outcomes are produced by the cumulative effect of policies, practices, and cultural norms that have become embedded in how an organisation operates. No single actor need be engaging in intentional discrimination for systemic discrimination to exist and to cause measurable harm. The organisation’s systems, processes, and informal structures are themselves the vehicle of the discriminatory effect.
Institutional racism at work is the most analytically demanding form to address because it does not produce the discrete incidents that individual complaint processes are designed to capture. It produces statistical patterns across large datasets: representation ratios by level and function, promotion velocity differentials by demographic cohort, attrition rates segmented by identity group and tenure, and compensation distributions that correlate with race or sex after controlling for job-relevant variables.
The EEOC’s systemic enforcement programme is specifically designed to address this category. The agency’s 2024 Annual Performance Report documented 16 systemic case resolutions in fiscal year 2024, obtaining just over 23.9 million dollars for 4,074 victims of systemic discrimination. Systemic cases involve a pattern, practice, or policy of discrimination that creates broad harm across multiple employees, and their resolution typically requires both monetary relief and equitable remedies including changes to employment practices.
Most organisations operating in 2026 have explicit anti-discrimination policies, conduct some form of harassment prevention training, and have HR functions staffed to respond to complaints. Discriminatory behaviour continues at scale despite all of this. Understanding why requires looking at the structural conditions that allow it to persist rather than focusing on individual bad actors.
The first structural condition is the under-reporting problem. Employees who experience discriminatory behaviour face a rational calculation in which the expected benefit of reporting is weighed against the expected cost. The expected cost includes retaliation risk, the credibility asymmetry between complainants and respondents who hold more organisational power, the social cost of being labelled difficult or oversensitive, and the perception that the formal process will not produce a meaningful outcome. The consistent appearance of retaliation as the most frequently cited basis in EEOC charges is evidence that the rational calculation discouraging reporting is often accurate.
The organisational cost of employees being unwilling to speak up about discriminatory behaviour compounds over time. It means that managers who engage in discriminatory behaviour receive no corrective feedback from below, HR receives no data signal about the behaviour’s prevalence, and the organisation accumulates liability exposure without any mechanism for internal correction.
The second structural condition is the credibility asymmetry built into most investigation processes. Formal complaint investigations typically produce a binary outcome: the conduct either crossed a threshold or it did not. This binary framing systematically misses the cumulative pattern problem. A single instance of differential treatment by a manager may not constitute actionable discrimination. The same manager’s repeated pattern of differential treatment across multiple employees over multiple years is systemic discrimination. Individual complaint processes, designed to adjudicate discrete incidents, are not architecturally suited to detect or respond to that pattern.
The third structural condition is the performance evaluation system. Performance reviews are a documented site of discriminatory behaviour because they involve subjective judgment exercised by individuals who carry all of the cognitive biases that produce covert discrimination, applied to decisions that have direct consequences for pay and advancement. Research on performance evaluation language consistently shows that the same professional behaviours are described in positive terms for some demographic groups and negative terms for others. Organisations that do not audit their performance data for demographic correlation and do not calibrate evaluation language and criteria across rating managers are running a bias amplification system inside what appears to be a merit-based process.
Discriminatory behaviour does not operate cleanly along single demographic axes. Employees who hold multiple identity memberships that are each individually associated with structural disadvantage experience discriminatory behaviour through the intersection of those dimensions, producing a qualitatively distinct experience that a single-axis analysis cannot detect or describe.
A woman of colour in a professional role does not experience race discrimination and sex discrimination as parallel, independent streams. The specific form of discriminatory behaviour directed at her is shaped by the interaction of both identities, producing microaggressions, evaluation patterns, and access barriers that are different from those experienced by a white woman or a man of colour in the same role. Survey data segmented by single demographic dimensions will systematically miss this reality because it averages across experiences that are not comparable.
The hidden cost of ignoring intersectionality in DEI data collection and intervention design is that the employees who face the most severe and compounded discriminatory behaviour are precisely the ones who are least visible in single-axis analyses. They fall through the gap between category-level analyses that show their primary identity group is not dramatically underrepresented and the actual lived experience of differential treatment that shapes their daily professional environment.
Effective organisational response to discriminatory behaviour requires systems that operate at three levels simultaneously: incident detection and response, pattern identification and correction, and structural prevention.
At the incident level, organisations need reporting mechanisms that are genuinely accessible, genuinely confidential to the degree legally possible, and genuinely protective against retaliation. This means multiple reporting channels that do not all route through a single manager or HR generalist, investigations conducted by personnel who have been trained in trauma-informed interviewing and who do not have a prior relationship with the respondent, and documented timelines and outcomes communicated to complainants.
At the pattern level, organisations need to run regular analyses of workforce data disaggregated by demographic dimension and intersectional combination. This means tracking promotion rates, compensation distributions, performance evaluation scores, assignment patterns, and attrition by demographic cohort and analysing the data for differential patterns that are not explained by job-relevant variables. When those patterns appear, the analytical question is not whether individual managers intended to discriminate. It is what process or policy is generating the outcome and what intervention will correct it.
Navigating the most common DEI issues in the workplace at the practitioner level requires this combination of incident-level responsiveness and pattern-level analysis. Organisations that are only reactive to formal complaints are operating without visibility into the covert and systemic discriminatory behaviour that accounts for most of the inequitable outcomes in their workforce data.
At the structural prevention level, organisations need to build discriminatory-behaviour prevention into the architecture of their decision-making processes. This includes structured interview processes with predetermined evaluation criteria applied consistently across all candidates, calibration sessions for performance evaluations that surface and challenge differential language across rating managers, pay equity analyses conducted at regular intervals and corrected when disparities appear, and promotion process audits that examine both the formal criteria and the informal sponsorship networks that influence who enters the promotion pipeline.
Organisations that have actual or constructive knowledge of discriminatory behaviour and fail to take prompt corrective action face significantly expanded liability under federal law. The employer liability standard under Title VII for supervisor harassment that results in a tangible employment action is strict liability. For supervisor harassment that does not result in a tangible employment action, the employer can assert an affirmative defence by showing that it exercised reasonable care to prevent and promptly correct the behaviour and that the employee unreasonably failed to use the preventive or corrective opportunities the employer provided. For harassment by non-supervisory employees, the employer is liable if it knew or should have known about the conduct and failed to take appropriate corrective action.
This means that organisations cannot simply establish an anti-harassment policy and a reporting mechanism and consider their obligation discharged. The prompt corrective action standard requires actual investigation and response when discriminatory behaviour is reported, and the reasonable care standard requires that the preventive infrastructure the organisation has built is genuinely functional, not merely documented.
The fiscal year 2024 EEOC data makes the financial stakes concrete. Nearly 700 million dollars recovered for victims of discrimination across one fiscal year. That figure does not include the cost of internal investigations, legal defence, reputational damage, or the productivity and retention losses that discriminatory workplace environments generate long before any formal complaint is filed. The total organisational cost of discriminatory behaviour is substantially larger than the enforcement data captures, because enforcement data only measures the cases that reach the EEOC. Most discriminatory behaviour never produces a formal charge.
Discriminatory behaviour at work exists across a spectrum from explicit statutory violations to embedded structural patterns that produce inequitable outcomes without producing incidents that surface through formal complaint processes. Addressing it requires practitioners who understand both ends of that spectrum and who build organisational systems with the architecture to detect, respond to, and prevent discriminatory behaviour at every point along it.
The legal framework establishes the floor of organisational obligation. The EEOC enforcement data documents what happens when organisations fail to meet even that floor. The practitioner responsibility extends above the floor, into the structural and cultural dimensions of discriminatory behaviour that no federal statute fully describes but that produce measurable inequity in every workforce where they operate unchecked.
The organisations that manage this well are not those that avoid legal liability. They are those that build workplaces where discriminatory behaviour, in all its forms, cannot survive because the systems, processes, and cultural norms of the organisation are actively designed to identify and eliminate it.
The Diverseek podcast aims to create a platform for meaningful conversations, education, and advocacy surrounding issues of diversity, equity, inclusion, and belonging in various aspects of society.