Intersectionality - What It Actually Claims

Intersectionality is among the most used and most misused terms in the social sciences. It is invoked constantly, frequently as a synonym for "multiple identities matter," and it has been attacked and defended by people who agree about almost nothing including what it says.

This lesson does what the rest of Part 5 has done: state the argument precisely, trace where it came from, show what it explains, and state honestly what it does not.

It began with a legal case that five women lost, for a reason that made no sense.

DeGraffenreid v. General Motors, 1976.

Five Black women sued General Motors for discrimination. The company's St Louis plant had, in effect, two workforces: a factory floor that had historically hired only men, and an office workforce that had historically hired only white women.

Black women fitted neither. And when the company made redundancies on a last-in, first-out basis, the seniority system — built on a hiring history that had excluded them — removed them first.

The court dismissed the case, and the reasoning is the point.

On the sex discrimination claim: the company employed women. There were white women in the office. Therefore no sex discrimination.

On the race discrimination claim: the company employed Black people. There were Black men on the factory floor. Therefore no race discrimination.

And on the claim that they had been discriminated against as Black women specifically, the court declined to proceed. Its stated concern was that allowing a compound claim would create "a new classification of 'black women' who would have greater standing than, for example, a black male" — and it warned that "the prospect of the creation of new classes of protected minorities, governed only by the mathematical principles of permutation and combination, clearly raises the prospect of opening the hackneyed Pandora's box."

So: the plaintiffs could not show race discrimination, because Black men were hired. They could not show sex discrimination, because white women were hired. And they could not combine the two.

The discrimination was real, documented and demonstrable — and there was no legal category in which it could be stated.

Kimberlé Crenshaw, a legal scholar, wrote about this case and two others in 1989, and named what she had found.

Crenshaw's argument in "Demarginalizing the Intersection of Race and Sex" (1989) is not that Black women face more discrimination. It is a structural claim about categories.

The dominant framework in both antidiscrimination law and social theory was single-axis. It analysed race discrimination by taking the experience of the most privileged members of the racial group — Black men — as the paradigm. It analysed sex discrimination by taking the experience of the most privileged members of the sex group — white women — as the paradigm.

Which means that in both frameworks, the paradigm case is someone who is subordinated on only one axis.

And the consequence follows rigorously: those subordinated on more than one axis fall outside both frameworks, not because their situation is more extreme but because it is structurally different in kind.

Crenshaw's metaphor, which is the one everyone remembers:

"Consider an analogy to traffic in an intersection, coming and going in all four directions. Discrimination, like traffic through an intersection, may flow in one direction, and it may flow in another. If an accident happens in an intersection, it can be caused by cars travelling from any number of directions and, sometimes, from all of them. Similarly, if a Black woman is harmed because she is in the intersection, her injury could result from sex discrimination or race discrimination... But it is not always easy to reconstruct an accident: sometimes the skid marks and the injuries simply indicate that they occurred simultaneously, frustrating efforts to determine which driver caused the harm."

The claim is about categories and their construction, not about the quantity of disadvantage.

What it is not: the additive model

Why "greater than the sum" is not a claim about severity, and this is where most misreadings begin.

The additive model treats "woman" as a stable category to which "Black" is added. But which woman? The category "woman," as constructed in the discrimination framework and in much feminist theory, was built from the experience of women who are not racially subordinated. A Black woman is not a white woman plus race.

Concretely: the stereotype of women as delicate, dependent and requiring protection from work — which underwrote protective labour legislation and the family wage — was never applied to Black women in the United States, who were expected to perform heavy labour throughout slavery and after. So the specific form of sexism operating on white women and the specific form operating on Black women are different in content, not different in quantity.

Sojourner Truth's 1851 speech makes exactly this point, and makes it 138 years before Crenshaw:

"That man over there says that women need to be helped into carriages, and lifted over ditches, and to have the best place everywhere. Nobody ever helps me into carriages, or over mud-puddles, or gives me any best place! And ain't I a woman? Look at me! Look at my arm! I have ploughed and planted, and gathered into barns, and no man could head me! And ain't I a woman? I could work as much and eat as much as a man — when I could get it — and bear the lash as well! And ain't I a woman?"

(A note on the source: the version quoted above is Frances Gage's 1863 reconstruction, published twelve years after the speech and rendered in a Southern dialect Truth — who was from New York and whose first language was Dutch — is unlikely to have used. Marius Robinson's contemporaneous 1851 report differs substantially and contains no "ain't I a woman" refrain. The argument is Truth's; the famous wording is probably Gage's. This course states such things.)

The point stands either way: the category "woman," as it operated, did not include her. That is not addition. It is a different construction of the same category.

The three forms

Crenshaw's 1991 essay "Mapping the Margins" distinguishes three, and the distinction is more useful than the metaphor.

Structural intersectionality — the ways in which the location of women of colour at the intersection of race and gender makes their actual experience of domestic violence, rape and remedial reform qualitatively different from that of white women.

Her examples are concrete: immigrant women whose residency status depended on remaining married, so that leaving an abusive husband meant deportation; shelters that could not serve non-English-speaking women; and job training programmes that presupposed skills and childcare arrangements the women did not have.

Political intersectionality — the ways in which women of colour are situated within at least two subordinated groups that pursue conflicting political agendas.

Antiracist politics has frequently suppressed discussion of domestic violence within communities of colour, for fear of confirming racist stereotypes. Feminist politics has frequently ignored race. The result is that women of colour are asked to choose, and that the price of solidarity in each movement is silence about the other.

Representational intersectionality — the ways cultural representations of women of colour, and the contestation of those representations, elide their specific position.

The framework generates predictions that can be checked, and this is what distinguishes it from a slogan.

The additive prediction: if disadvantages simply accumulate, then the outcome for any group should be predictable from the sum of its component effects — the race effect plus the gender effect.

The intersectional prediction: the outcomes will not be predictable by addition, because the mechanisms differ by group.

Where the intersectional prediction has been borne out:

Maternal mortality. In the United States, Black women die of pregnancy-related causes at roughly three times the rate of white women, and — the finding that defeats the additive model — the disparity persists across education and income levels, with college-educated Black women experiencing worse outcomes than white women without a high school diploma on some measures. A model in which race and class effects simply add cannot produce this.

The wage gap. Decomposing pay by race and gender jointly produces different patterns than either analysis alone, and the mechanisms differ: occupational segregation operates differently, and the "motherhood penalty" documented for white women has a different shape for Black women, whose labour force participation has historically been higher.

Criminal justice. Sentencing research finds that the effects of race, gender and age interact rather than add, with young Black men receiving the most severe outcomes in ways not predicted by summing the main effects.

Health. The literature on weathering — Arline Geronimus's hypothesis that cumulative exposure to social and economic adversity produces early health deterioration — was developed specifically to explain Black women's health outcomes that neither race nor class models predicted alone.

And the methodological point. Statistically, intersectionality is a claim about interaction effects rather than main effects — that the effect of one variable depends on the level of another. This is testable, and a substantial quantitative literature now does it, using interaction terms, multilevel models and Evans and colleagues' MAIHDA approach (multilevel analysis of individual heterogeneity and discriminatory accuracy), which handles the combinatorial explosion of categories that worried the DeGraffenreid court.

The honest finding from that literature: interactions are frequently found, and they are frequently smaller than the main effects. Both halves matter — intersectional analysis is not always necessary, and where the mechanisms genuinely differ, additive models mislead.

The ancestry

Crenshaw named it. She did not invent it, and she has said so repeatedly.

Maria Stewart (1830s) and Sojourner Truth (1851) — the earliest statements.

Anna Julia Cooper, A Voice from the South (1892) (4.6.1): "Only the BLACK WOMAN can say when and where I enter, in the quiet, undisputed dignity of my womanhood, without violence and without suing or special patronage, then and there the whole Negro race enters with me."

Ida B. Wells (4.6.1) — her lynching analysis is explicitly about how race and sexuality operated together, through the construction of the Black male rapist and the simultaneous unprotectedness of Black women.

The Combahee River Collective Statement (1977) is the most complete pre-Crenshaw statement, and it deserves quoting:

"We are actively committed to struggling against racial, sexual, heterosexual, and class oppression, and see as our particular task the development of integrated analysis and practice based upon the fact that the major systems of oppression are interlocking. The synthesis of these oppressions creates the conditions of our lives."

"Interlocking" is the key word, and it predates "intersectional" by twelve years.

bell hooks, Ain't I a Woman (1981) and Feminist Theory: From Margin to Center (1984) — the sustained critique of feminism's construction from white middle-class experience.

Angela Davis, Women, Race and Class (1981) — the historical analysis, including the crucial account of how the enslaved woman's position was constructed outside the categories of both womanhood and labour.

Deborah King (1988) — "multiple jeopardy, multiple consciousness," which explicitly rejects the additive model a year before Crenshaw.

And Patricia Hill Collins's matrix of domination (1990) (5.6.1), which supplies the systemic framework: race, class, gender and sexuality as interlocking systems organised across four domains — structural, disciplinary, hegemonic and interpersonal.

Go deeper

The criticisms, taken seriously

Intersectionality has attracted criticism from several directions, and some of it is substantial.

The infinite regress problem. If race and gender intersect, so do class, sexuality, disability, age, religion, migration status, region, caste and body size. At what point does analysis become an unmanageable proliferation of ever-smaller categories? This is the DeGraffenreid court's "permutation and combination" worry, restated as a methodological rather than a legal objection.

The best answer: intersectionality is not a demand to include every possible category but a claim that the relevant axes must be determined empirically for the question at hand, and that the categories are relational rather than additive. Collins and Bilge's formulation: intersectionality is an "analytic sensibility," not a checklist.

Is it a theory, a method, or a heuristic? A genuine dispute. Some argue it is a research paradigm; others that it is a framework for generating hypotheses; others that it functions mainly as a critical orientation. Leslie McCall's influential 2005 essay distinguishes three approaches — anticategorical (deconstructing categories), intracategorical (studying neglected intersections in depth), and intercategorical (using existing categories provisionally to document relationships between groups) — and notes that most quantitative work is intercategorical, which some purists regard as a betrayal of the concept.

The identity drift. In popular usage, intersectionality has frequently become a claim about identity and about who may speak, rather than a structural analysis of how systems of power operate together. Crenshaw has objected to this repeatedly, insisting that her argument was about structures and legal categories, not about ranking identities.

Does it lose the specificity of each system? If race, class and gender are always analysed together, is anything lost in the analysis of, say, capital accumulation specifically? Some Marxist critics argue that intersectionality dissolves the distinct logics of different systems into a general category of "oppression."

And the political critique from both directions: from the right, that it fragments any possible common political subject; from parts of the left, that it substitutes a politics of recognition for a politics of redistribution (a version of Nancy Fraser's argument). Crenshaw's own position is that political intersectionality is precisely a diagnosis of why coalitions fail and a proposal for building better ones.

The honest assessment of the concept's status.

As an analysis of legal categories — the original argument — it is essentially uncontested. The single-axis structure of antidiscrimination law did produce the failures Crenshaw documented, and courts in several jurisdictions have since recognised compound and combined discrimination claims.

As an empirical claim about interaction effects — it is testable, has been tested, and is sometimes confirmed and sometimes not. The maternal mortality and sentencing findings are strong; not every intersection produces a significant interaction.

As a general theoretical framework — it is productive and it is contested, particularly on the regress problem and on whether it constitutes a theory.

And as a popular term — it has drifted substantially from its origin, in ways its author has publicly resisted.

A reader should hold all four separately, and be clear which one is being asserted in any given argument.

Where it is used

Law — antidiscrimination doctrine, human rights law, and the recognition of compound claims in several jurisdictions including the EU and Canada.

Health research — one of the most active areas, with a substantial methodological literature on how to operationalise intersectional analysis quantitatively.

Criminology — sentencing, policing and victimisation research.

Labour market analysis — decomposition of pay and employment gaps.

Social policy — the analysis of how welfare, migration and family policy interact for differently positioned groups.

And social movement research — the study of coalition formation and its failures, which is political intersectionality applied.

And the honest reflection this course owes

This course has now made the same argument five times: canons are institutional products (3.3.3), exclusion has an epistemic cost (4.5.5), reclassification removes work from a discipline (4.6.1), traditions outside the metropole arrive at conclusions decades early (4.6.3), and categories built from one standpoint misdescribe (5.6.1).

Intersectionality is the general form of that argument. It says: the categories a field uses were built by people in particular positions, they fit those positions, and people at the intersections of subordination fall outside them — not because they are unusual but because the categories were never constructed with them in view.

And the test of whether a reader has understood it is whether they can apply it to a case that does not involve race and gender. Try it on: class and disability. Migration status and age. Caste and gender in a labour market. Rural location and educational credential. In each case the question is not "which disadvantage is worse" but "does the standard analysis of either one describe what happens at their intersection?"

Usually it does not, and that is the whole claim.

"Intersectionality means multiple identities matter." It means categories are constituted differently at intersections, so that single-axis analysis misdescribes rather than merely underestimating.

"It means disadvantages add up." It is an explicit rejection of the additive model. Crenshaw's whole argument is that summing produces the wrong answer.

"Crenshaw invented the idea." She named it and gave it a legal and theoretical formulation. Stewart, Truth, Cooper, Wells, the Combahee River Collective, hooks, Davis and King precede her, and she says so.

"It's about ranking oppressions." It is about how systems operate together. Crenshaw has publicly and repeatedly objected to the identity-hierarchy reading.

"It can't be measured." It is a claim about interaction effects, and there is now a substantial quantitative literature, including purpose-built methods.

"Sojourner Truth said 'Ain't I a woman?'" The famous version is Frances Gage's 1863 reconstruction, in a dialect Truth is unlikely to have used. The argument is hers; the wording is probably not.

Why it matters

It identifies a structural defect in how categories are built, and the defect is general. Any framework whose paradigm case is someone subordinated on a single axis will fail for those subordinated on several — and will fail by producing no description at all, which is worse than an inaccurate one.

It converts a political observation into a testable claim. Interaction effects can be estimated. The maternal mortality finding, the sentencing findings and the wage decompositions are evidence, and evidence is what distinguishes a research programme from a position.

It explains coalition failure. Political intersectionality is a diagnosis of why movements that should be allies repeatedly are not, and why the price of belonging to each is silence about the other. That is a practical finding with practical consequences.

And it is the general form of this course's recurring argument. Categories are built by people in positions; they fit those positions; and the people they do not fit become invisible rather than merely disadvantaged. The discipline has now been shown this about class, about race, about gender, about empire and about its own canon. Intersectionality names the mechanism.

The origin: DeGraffenreid v. General Motors (1976), in which five Black women could not establish race discrimination (Black men were hired), could not establish sex discrimination (white women were hired), and were refused a combined claim — the court warning against "permutation and combination" opening "the hackneyed Pandora's box."

Crenshaw's 1989 argument: antidiscrimination law and social theory were single-axis, taking as paradigm the most privileged member of each subordinated group — Black men for race, white women for sex. Those subordinated on multiple axes therefore fall outside both frameworks entirely.

It is a rejection of the additive model. "The intersectional experience is greater than the sum of racism and sexism" is not a claim about severity but about construction: the category "woman" as it operated was built from women who were not racially subordinated, and protective stereotypes applied to white women were never applied to Black women. Sojourner Truth made the point in 1851 — though the famous wording is Frances Gage's 1863 reconstruction.

Three forms (1991): structural (different lived experience of violence and remedy at the intersection), political (situated within two movements with conflicting agendas, so that solidarity in each demands silence about the other), and representational.

The evidence: US maternal mortality disparities that persist across education and income, defeating additive models; interacting effects in sentencing; wage decompositions with differing mechanisms; and weathering in health research. Statistically it is a claim about interaction effects, testable and sometimes confirmed.

The ancestry: Stewart, Truth, Cooper (1892), Wells, the Combahee River Collective (1977) — "the major systems of oppression are interlocking" — hooks, Davis, King's multiple jeopardy (1988), and Collins's matrix of domination.

The criticisms: the infinite regress problem; the dispute over whether it is a theory, method or sensibility (McCall's anticategorical / intracategorical / intercategorical); the drift into an identity claim, which Crenshaw resists; and the concern that it dissolves the distinct logics of different systems.

Intersectionality — the analysis of how systems of subordination operate together, constituting categories differently at their intersections.

Single-axis framework — analysis that treats one dimension of subordination at a time, taking the most privileged member of the group as paradigm.

Additive model — the rejected view that disadvantages accumulate and can be summed.

Structural intersectionality — qualitatively different lived experience and remedy at the intersection.

Political intersectionality — location within two or more movements pursuing conflicting agendas.

Representational intersectionality — the elision of specific positions in cultural representation.

Interlocking systems — the Combahee River Collective's term, predating "intersectional."

Matrix of domination — Collins's framework: interlocking systems across structural, disciplinary, hegemonic and interpersonal domains.

Interaction effect — the statistical form of the intersectional claim: the effect of one variable depends on the level of another.

Anticategorical / intracategorical / intercategorical — McCall's three approaches to intersectional research.

Weathering — Geronimus's hypothesis of early health deterioration from cumulative adversity.

One — apply it away from race and gender. Take class and disability, or migration status and age, or rural location and credentials. Ask whether the standard analysis of either describes what happens at their intersection.

Two — find a single-axis category. Take any official classification or policy eligibility rule. Identify whose paradigm case it was built from, and who falls between its categories.

Three — test the additive prediction. Find published data broken down by two characteristics jointly. Check whether the joint outcome equals the sum of the separate effects. It frequently does not.

Four — trace a political intersection. Find a case where a movement suppressed an issue for fear of how it would be used against the group. Ask who paid for the silence.

Five — check a famous quotation. Look up both the Gage and the Robinson versions of Truth's 1851 speech. The exercise is about intersectionality's history and about how any quotation becomes canonical.