Of Groups, Intersections, and the People Who Inhabit Them

The powerful norms and rules that structured life in the American South during slavery and Jim Crow not only prohibited sex between black men and white women but tried to tightly regulate even the possibility of a black man’s sexual desire for a white woman. To rape a white woman, to be suspected of raping a white woman, to sexually proposition a white woman, sometimes to look at a white woman in a way that some observer interpreted as lustful—these were forbidden, and during the long era of lynching they were paradigmatically offenses punished by death. By contrast, those norms and rules only intermittently, and never as severely, restricted white men’s sexual interest in black women, up to and including rape. In some times and places it might be a matter of embarrassment, not to be publicly acknowledged—and the resulting born-into-slavery offspring were certainly not to be acknowledged—but it was not a hanging offense, and de facto almost never a criminal offense at all.

Was there a relationship between the racism of the era and sex? Or rape? It would take a singularly obtuse analysis to say “no; interracial sex was neither generally prohibited nor generally allowed; interracial rape was neither generally prohibited nor generally allowed; therefore, race and norms about sex or rape were uncorrelated.” The norms of white supremacy interacted with norms about sex and gender; racism was refracted through gender, and they manifested differently in the two (heterosexual) pairings. A contemporary academic scholar would be likely to express this as “racism was gendered” or “patriarchy was racialized,” or both, but that language is secondary to the basic thought that racist and sexist norms interacted in a way that means we shouldn’t just take some average and say “race didn’t matter.”

In statistics, when studying a phenomenon with a quantifiable dependent variable, we commonly find two independent variables that complicate each other. Suppose that, all else equal, higher income is associated with more conservative political views, and higher education is associated with more liberal political views, but higher education is also correlated with higher income. We improve the power of a statistical model that seeks to predict ideology by using as independent variables income, education, and an interaction between the two. Without including the interaction effect, the income and education effects might confound each other to the point where we doubted that either matters for ideology, when in fact both do. While interaction effects are easiest to understand—and to control for—in statistical studies, the underlying thought need not be restricted to easily quantifiable outcomes. Two or more causes or phenomena might both be important for understanding some outcome or state of affairs, but they interact with each other in shaping the outcome such that we won’t understand the outcome as well as we could if we look at each one separately.

None of this is controversial—certainly not in quantitative social science, and expressed at this level of generality, not in qualitative social science either.

For some reason, however, intersectionality—the analytical strategy of thinking about such phenomena as race and gender, or racism and sexism, as interacting—has become a wildly controversial concept, mostly among right-leaning non-academics who have fixated on an extraordinary caricature of it and elevated it into a religion, cult, or totalitarian ethos that somehow rules higher education. I imagine that we will devote some of the reply and discussion essays to that problem. I want to set it aside for now, in order to answer the question the symposium asks, and to argue that intersectionality rightly understood is an important and valuable intellectual tool for classical liberalism.

Intersectionality was first developed by the legal scholar Kimberlé Crenshaw in a 1989 article in the University of Chicago Legal Forum, “Demarginalizing the intersection of race and sex: A black feminist critique of antidiscrimination doctrine, feminist theory, and antiracist politics.”

She developed it further in a 1991 article in the Stanford Law Review, “Mapping the margins: Intersectionality, identity politics, and violence against women of color.” (I will also note, and the literature acknowledges, that my late colleague and collaborator Iris Marion Young developed closely related ideas using a different vocabulary in her 1990 Justice and the Politics of Difference.)

The two main targets of critique in these articles are, first, identity politics as it was then conceived, and, second, elements of U.S. antidiscrimination law, both of which she argued treated concerns about racism and racial discrimination and concerns about sexism and sex discrimination separately from one another. Doing so, she argued, tended to obscure intergroup differences. Feminism tended to become the politics of white middle-class women representing their concerns as those of women as such. Antiracist politics tended to become the politics of black men. Black women, and by extension women of color generally, were hidden from theoretical view: “this focus on otherwise-privileged group members creates a distorted analysis of racism and sexism because the operative conceptions of race and sex become grounded in experiences that actually represent only a subset of a much more complex phenomenon.”

It’s worth emphasizing a few features of intersectionality as an analytical tool that one would not know from the familiar caricature of it. One is that its key original targets of criticism were extant versions of identity politics and antidiscrimination law. Intersectionality is not itself an answer to any broad-brush question like “how racist or sexist is American society?” One could think—Crenshaw doesn’t, and I don’t, but one could—that the answer to that question was “not very,” and still think that intersectionality was an important tool for understanding how the remaining vestiges of racism and sexism manifest. Another is that intersectionality is not about the question “who is most oppressed?” and its method is not to count up different axes of oppression. Its central question is not how much?—a question that might call for such counting—but rather how?

Now, once one begins to think intersectionally, it becomes a natural question to ask of any given group’s identity politics, avowed representatives, and legal standing whether some subgroups’ interests or oppression are being obscured. And it is something of a truism that it will tend to be less powerful subgroups whose different experiences of injustice are less seen or heard. So an intersectional approach will tend to direct attention to the question “does this representation of the injustice one group suffers hide the different way that a less-powerful subgroup experiences the injustice?” That will mean, for example, that attention tends to get drawn to the question “does this particular feminist claim about injustice rely on a description of the lives of white middle-class women and obscure the very different way things play out for poor women or women of color?” But the point of that is not to stack the categories “woman,” “poor,” and “racial minority.” It is to refract each through the others.

In statistics, not all interaction terms turn out to have explanatory power. Not every dimension of social difference or social inequality interacts with every other dimension, either, and sometimes asking these questions won’t reveal relevant differences. But this is true for every methodological question of social analysis: the “who benefits?” of public choice theory, the “at what cost and at what margin?” of microeconomics, the “how will competing actors respond?” of game theory, and so on. Any of these questions can be distorting if it is taken to be the only question, and it can tempt the analyst who uses it to think it always generates answers that explain every situation. But they all sometimes direct our attention at questions we want answered, and that’s enough to understand them as being intellectually useful and valuable.

With that understanding in mind, what makes intersectionality useful and valuable to classical liberalism—and vice-versa? I should emphasize that I think classical liberals share other liberals’ reasons for wanting to know how, for example, racism and misogyny work and interact at a social and cultural level. But in the interest of having a symposium suitable to Cato Unbound, I want to single out some reasons why classical liberal social theorists in particular should be interested in asking intersectional questions, beyond the reasons that they share with liberalism more broadly understood.

At their worst, classical liberal ideas, slogans, and politics have been appropriated as defenses only of the liberty of white men, with “liberty” here meaning a freedom from state intervention in private and quasi-private sites of domination of others from the plantation to the marital bedroom. In the standard case, classical liberal political theory has defended formal equal liberty for all but implicitly relied on racialized nationalist understandings of the pre-political people and preferred formal treatments of rights to substantive political understandings of how state violence and power are used, and how they connect with social sources of power. But at their best, not only should classical liberal ideals be genuinely concerned with liberty for all; so should classical liberal social analysis be genuinely concerned with the social sources and social effects of unfreedom. For this, intersectionality offers essential insights.

Social injustice and spontaneous orders

I have argued elsewhere that the broad idea of social injustice that includes generalized racial and gendered social disadvantage is much more congenial to classical liberal social analysis than has been generally recognized. Smithian and Hayekian understandings of complex and emergent social orders encourage us to de-emphasize attention to actors making deliberate decisions to plan large-scale outcomes, in favor of observed outcomes that are real but possibly unintended by anyone involved. Emergent orders can have properties that their components do not have. It is possible for each of us to be led by an invisible hand to promote an end which was no part of our intention. And there is no reason to suppose that such ends are inevitably beneficial or salutary. Small-scale perversities, accidents, historical legacies, or practices can multiply and ossify into self-reinforcing large-scale patterns of exclusion and domination. Thomas Schelling’s famous model showing that white flight could be a case of a “complex system with collective results that bear no close relation to individual intent” has sometimes been taken as somehow exculpatory of American society, as if the only lesson that could be drawn from it is that one can’t infer individual-level racist attitudes from the fact of white flight. But if we are interested in social analysis rather than in treating racism as an individual sin and providing individual white people with a defense against the charge, this is very strange. As in Smith’s original discussion, our interest should be less in the souls of the individual actors than in the shape of the world they create—in the case of white flight, one in which both high-quality publicly provided goods and increases in wealth through homeownership are constantly receding from the reach of black households. This is not an injustice that is created de novo—a history of unjust de jure housing segregation shaped the initial dynamic, as did white homeowners’ suspicion that other white homeowners and buyers might harbor racist attitudes. But it amplifies, multiplies, and reshapes the unjust inheritances into a new pattern.

But precisely because emergent injustices don’t map onto obvious individual-level injustices (each homeowner has a right to sell their own house when they like and violates no one’s rights by doing so), they require deliberate social-level examination to pick out. Again, this is familiar. If you focus on each person pursuing their self-interest and never look at, so to speak, the wealth of nations, you would never notice the end that they have inadvertently promoted. The “spontaneous orders” F.A. Hayek examined through his scholarly career require the same ability to look at a phenomenon at two levels: language, the common law, the price system, evolved cultural rules—none of these can be read directly off individual actions or bilateral exchanges. So one has to have access to questions that direct intellectual attention. Understanding the existence of racism directs our attention to noticing white flight amidst the churn of tens of millions of residential property sales in a year. Knowing that race and gender interact gives us the ability to notice and to ask about patterns of exclusion and domination in which they do so, from labor markets to cultural stereotypes.

Social power and state power

Classical liberals are—or ought to be—particularly interested in the sources and causes of state power. All too often, though, that inquiry begins and ends with the concept of rent-seeking, the understanding that powerful economic actors seek to advance their economic interest through the strategic use of state regulation and expenditure. The phenomenon of the powerful seeking to use and expand state authority to entrench and enhance their domination is entirely general. This should sound like a truism:the powerful use their power to protect and expand their power. And yet all too often people think it is strange or counterintuitive to say that members of a dominant race, gender, religion, or orientation predictably do what Adam Smith taught us dominant economic interests do: expand state power to entrench existing power relations in another sphere. We astonishingly manage to forget that in the United States racism has been a source of statism, that state power has been expanded to enhance and entrench racial domination, from the organization of armed force to expropriate Indian lands and keep down slave revolts to the drug war, overpolicing, and mass incarceration.

Homophobia, of course, has also been a social cause of state power. While sodomy laws themselves preceded our contemporary sense of sexual orientation as an ongoing identity, in the twentieth century the existence of sodomy laws in the background empowered those in law enforcement and intelligence agencies to expand their power through blackmail, extortion, raids, and the threat of mass arrests. Individual gays and lesbians in government employment or the armed services, gay clubs and bars and social gatherings—all could be held hostage by any number of state actors with the threat of exposure and prosecution, in much the way that sex workers are vulnerable to police dominance to this day.

Classical liberals should want to know whether the intersection between race and homosexuality has generated especially aggravated kinds of state power from the outside—particularly invasive policing of black or mixed-race gay clubs. They should want to know how racism and sexism have interacted in the intrusive paternalism shown by many branches of the state in regulating of women of color’s reproductive lives and childraising.

They should also want to know whether access to or the promise of political power has aggravated in-group power differentials. Political scientist Cathy Cohen documented, for example, the ways in which post-Civil Rights Era African-Americans’ struggle for political power in a society marked by homophobia encouraged an internal silencing of black gay advocacy and expressions of identity, and a studied indifference to the advent of AIDS in black communities.

It’s a very common pattern. A subordinated group with access to a little bit of political power becomes internally vulnerable to the power of its representatives and intermediaries. (I discuss this, without the language of intersectionality, in chapter 3 of Rationalism, Pluralism, and Freedom.) These will typically be the relatively advantaged within the group: the relatively educated, the relatively wealthy, those with relatively high social standing or family status. In most settings we will find that these are, in addition, heterosexual men. The ability to speak for the group to the wider society makes these internal elites the proxy-holders for at least a portion of the wider society’s power over the group. It will be unsurprising if they use it to enhance and entrench their in-group power, and if they represent their elite subgroup’s interests when accessing political resources.

And so in addition to intersecting social group exclusion and domination sometimes being a source of state power, state power will sometimes be a source of intersectional disadvantage. Women, gays and lesbians, the poor, or dissidents will suffer additional in-group disadvantage from the ability of wealthy, educated, orthodox straight men in the group to act as political intermediaries. Even if classical liberal social theory is only interested in questions of state power and state abuses, intersectionality allows for interesting and important questions and analyses.


Sometimes critics of intersectional analysis offer what they suppose to be a clever faux-embrace of it: taken to its conclusion, it leads to complete individualism, because no one person has just the same set of life circumstances as any other, so the most intersectional way to think treats every person as an individual with no reference to social groups at all—or to any phenomena about social groups, such as racism.

The reductio is wrong: a tool of social analysis isn’t useful if it destroys the possibility of social analysis, which depends on the ability to group and aggregate. Economic analysis includes aggregating concepts like “consumers,” “the unemployed,” “the U.S. automotive industry,” “service workers,” “landlords,” and so on; political science has “voters,” “Republicans,” “military officers,” “protestors,” “the German government,” and so on. We never seek to understand the actions and decisions of 7 billion persons, one person at a time.

But the bad-faith reductio still points toward a valuable truth. When we observe the world, whether in the social sciences or outside them, we tend to treat aggregations as fixed blocs, dominant tendencies as stereotypes and rules, and groups as mutually exclusive and jointly exhaustive. We reify nations, religions, races, political parties, economic interests, or whatever. This is a particular problem when we are trying to understand intergroup dynamics that in part consist of just such reification: the stereotyping, objectification, and denial of agency or individuality that so often accompanies thinking about one or another disfavored group. How do we think about the social consequences of racism without reifying races?

The intersectional insight offers one answer. We don’t have to imagine away the categories that allow us to understand domination and exclusion; but we should remember that no one of them ever tells the whole story. No one is ever only their race, only their gender, only their religion, only their sexual orientation—even if any one of those might tell us important things about their social standing, their vulnerability, their legal rights, their political power. Each of those categories may have real explanatory power (“may,” always “may”) about a person’s life or a society’s structure. But there’s no one of them, not even any particular combination of them, that is ever the end of the inquiry. Always being willing to ask the intersectional questions—What other identity might matter? Does the general phenomenon manifest differently for a subgroup, particularly for a subordinate group? Once we’ve identified a group pattern, what various patterns might it be encompassing or hiding?—remind us never to treat any one of those identities or cleavages as the whole of a society, or the whole of a person.

Also from this issue

Lead Essay

  • Jacob T. Levy describes intersectionality as “the analytical strategy of thinking about such phenomena as race and gender, or racism and sexism, as interacting.” As such, he suggests that this approach is a promising one, and he defends it against some common criticisms. Those who view oppression of all kinds as a primarily state-mediated phenomenon should want to add it to their repertoire of useful ideas, he argues.

Response Essays

  • Phillip W. Magness warns that intersectionality theory amounts to little more than a qualitative approach to multivariate regression analysis—and a doubtful set of ideological commitments. For centuries, those now known as classical liberals fought against slavery and other forms of discrimination, and their successors today don’t need any such new and dubious theories to continue the struggle.

  • Fabio Rojas expresses ambivalence about intersectionality’s relationship to classical liberalism. On the one hand, there is a clear line of influence running from Marxism and critical theory to intersectionality’s key proponents and texts, and this influence brings with it hostility toward the market process. But on the other hand, he finds that intersectionality can offer new insights into how states perform oppression, and how marginalized voices can learn to speak back. There is no marriage here, but there should be a dialogue, he concludes.