Intersectionality

Working Definition

Intersectionality is the proposition that categories of social location—gender, race, class, disability, sexual orientation, age—do not operate as separate, additive burdens, but interact to produce distinct structural positions whose logic cannot be reconstructed by analyzing each axis in isolation. In its technical sense, originating in US antidiscrimination law, it is a critique of “single-axis” legal procedures: if a plaintiff must file a claim either on the basis of sex or on the basis of race, a person discriminated against specifically as a Black woman loses both tests. In popular usage, the term refers more loosely to the norm of taking multiple identities into account at once or serves as a label for a style of progressive politics. The difference is crucial: the technical version formulates a testable hypothesis about the structure of discrimination, whereas the popular version is a declaration of moral sensibility.

Genesis and Global Lineage

The term was coined by the legal scholar Kimberlé Crenshaw—educated at Cornell, Harvard Law School, and the University of Wisconsin, and a professor at UCLA and Columbia—in her 1989 article Demarginalizing the Intersection of Race and Sex. Her point of departure was a series of federal employment-discrimination cases, most notably DeGraffenreid v. General Motors (1976), in which courts dismissed claims brought by Black women because the company hired both women (whowere white) and Black workers (who were men). Crenshaw’s metaphor of a traffic intersection, where an accident can be caused by cars traveling from several directions at once, was designed to expose an evidentiary blind spot in legal doctrine rather than to build a general theory of identity. In Mapping the Margins (1991), she extended the analysis across structural, political, and representational dimensions using fieldwork on domestic-violence shelters. The concept’s deeper lineage reaches back to nineteenth-century Black feminism (Sojourner Truth, Anna Julia Cooper) and the 1970s–1980s writings of the Combahee River Collective (whose 1977 manifesto spoke of “interlocking” systems of oppression), Angela Davis, bell hooks, and Patricia Hill Collins’s sociological “matrix of domination” (1990). In the 2000s the concept migrated into quantitative and qualitative sociology (Leslie McCall called it in 2005 perhaps the most important theoretical contribution of women’s studies), then into European Union and UN equality frameworks, and after 2014 into mass media, where both sides of the culture wars turned it into a political banner.

Central European Lineage

We found no documented link between this concept and the intellectual history or historical experience of Central Europe.

Why Now

Three developments between 2018 and 2026 moved intersectionality from academic seminars into operational practice. First, audits of machine-learning systems gave the concept a crisp quantitative proof: Joy Buolamwini and Timnit Gebru’s Gender Shades study (2018) demonstrated that commercial gender-classification systems had error rates up to 34.7% for dark-skinned women compared with under 0.8% for light-skinned men, even though overall single-axis accuracy scores (87.9–93.7%) looked respectable; disaggregating error rates across intersectional subgroups has since become a standard requirement in AI auditing. Second, institutionalization advanced in European law: the European Union embedded intersectional analysis into its Gender Equality Strategy 2020–2025 and Anti-Racism Action Plan 2020–2025. Third, political backlash intensified: after 2020 in the United States—and at the federal executive level from 2025—intersectionality was placed on lists of terms stripped from government training programs and research grants, making it a primary trigger word in institutional politics. In Poland and Hungary, campaigns against “gender ideology” and the 2020 Polish debate over withdrawing from the Council of Europe’s Istanbul Convention meant that the term entered public circulation more often as a political accusation than as an analytical instrument.

Explanatory Power

The concept makes visible situations in which each variable looks benign on its own, yet their interaction produces severe exclusion. Consider the labor market: a manufacturing firm that employs many women in clerical roles and many minority men on the factory floor may employ zero minority women—and single-axis compliance statistics will report full equality on both counts. Consider public policy in Central Europe: as Angéla Kóczé’s 2009 study for the Central European University in Budapest showed, state programs for Roma targeted at “families” (implicitly represented by men) and gender-equality programs targeted at “women” (modeled on majority middle-class women) systematically bypass Romani women, whose specific barriers—school segregation, lack of healthcare access, compounded labor-market exclusion—fall between the two bureaucratic mandates. Consider clinical medicine and technology: a diagnostic algorithm that achieves 95% accuracy for women overall and 95% for older patients overall may drop to 70% accuracy for older women from an underrepresented ethnic group if they were virtually absent from the training data. Across all three domains, the underlying mechanism is identical: what econometrics calls an interaction term and epidemiology calls effect modification.

Axes of Debate

Underappreciated Angle

The least discussed consequence of intersectionality is a practical problem of statistical power and data infrastructure. Every additional axis slices the sample more thinly: in a survey of a few thousand respondents, a three-way intersection of gender, age, and minority status may leave only a handful of observations, making standard confidence intervals useless. In public registries across much of Central Europe, where states historically avoid collecting racial or ethnic data out of post-1945 legal caution, intersectional disparities are formally invisible to official statistics. The exact same bottleneck governs machine learning: subgroups at the intersection of multiple minority traits are the sparsest in training corpora, causing models to fail on them most often, while privacy rules that prohibit collecting sensitive attributes prevent auditors from even measuring the error. Intersectionality is therefore not just a thesis about social power, but a sharp technical challenge to how states and digital platforms design their data infrastructure: those whom the statistical system cannot count in combination, the legal system cannot protect in practice.