Issue 13.2 | Spring 2016 / Guest edited by Soniya Munshi and Craig Willse

Legal Equality, Gay Numbers and the (After?)Math of Eugenics

Implicit in Foucault’s concept is the notion that the exact moment these modes of governmentality are reproducing the relations of rule, they are also providing the vocabulary for the contestations to those relations of rule.
—Grace Hong[1]

[LGBT-supportive policies] are linked to positive business-related outcomes including the corporate bottom line.
—Williams Institute[2]


In recent decades, a legal equality or “rights” politics that focuses on same-sex marriage, anti-discrimination laws, hate crime laws, and military service has become the most visible and resourced politics opposing heterosexism. The demands of this politics have sometimes become prominent electoral concerns and significant legal cases have made headlines. According to the advocates of this agenda, important victories have been won. The legal recognition of same-sex marriages and the repeal of “Don’t Ask, Don’t Tell” have been heralded as important achievements for equality. Many states, cities, and counties have passed anti-discrimination laws and hate crime laws that include sexual orientation as a protected category and some have added gender identity or expression as well. In 2009, the federal hate crimes statute was amended to include crimes based on sexual orientation and gender identity or expression.

To achieve these ends, advocates of the legal equality agenda have increasingly produced statistical data to support their arguments for legal equality. Gay numbers have proliferated. Demographers and policy reformers have produced reports that provide empirical support for the agenda of same-sex marriage recognition and military inclusion and that aim to describe gay and lesbian or sometimes LGBT populations. Perhaps the most significant contributor to this trend of generating empirical support for legal equality projects is the Williams Institute (WI) at UCLA Law School. Its website describes its founding:

Mr. Williams’ inaugural donation of $2.5 million to create the Williams Institute was the largest donation ever given to any academic institution in support of a gay and lesbian academic program in any discipline. As the institute has grown, Mr. Williams has given over $13 million to support the organization’s programs.

The WI describes itself as a “nonpartisan” think tank. It produces reports focused on marriage, military, and related subtopics such as tax policies impacting unmarried same-sex couples, and adoption for unmarried same-sex couples. Some examples of topics explored in recent[3] WI reports include:

  • how providing legal recognition of same-sex marriage will increase revenue to particular states,[4]
  • how same-sex couples pay more estate tax than different-sex couples because they cannot transfer wealth to a spouse untaxed before death,[5]
  • a study that suggests that 20 percent of gay veterans kicked out of the military would go back if they could and describes the negative impact (for the military) of replacing them when they have been kicked out, including raising concerns that the military is having to “lower its standards” by enlisting “convicted felons.”[6]

The WI is a leading organization in the production of empirical data about gay and lesbian people, is often cited by courts and lawmakers, and has also been a vocal advocate for increasing collection of data about gay and lesbian people, recommending that the US Census and other key data collection mechanisms include questions about sexual orientation.[7]

In this article, we examine how statistical methods are being employed to produce an image of a rights-deserving gay and lesbian or LGBT population. We use existing critical analysis of legal equality campaigns and critical analysis of statistical methods to understand how the two are operating together in the context of gay and lesbian legal equality campaigns. We briefly examine how the use of statistics for defining populations is implicated in the history of eugenics in order to ask how today’s uses of statistical data in reform campaigns might still operate to mobilize racist national technologies of control. We argue that the explosion of new empirical data about gay and lesbian or LGBT people is not discovering the truth about an existing population; rather, it is formulating that population in order to frame it as a “deserving” population in the contexts of US racial norms. The production of this knowledge promotes legal claims to a certain kind of national citizenship sought by this particular strain of equal rights advocacy. In this way, we seek to expose a relationship between statistical methods and contemporary legal equality claims that is not limited to the context of lesbian and gay or LGBT rights politics but is particularly visible in that advocacy.

Critiques of Legal Equality Approaches

Many scholars and activists have critiqued legal equality–centered responses to homophobia and transphobia, raising concerns about the political formations producing demands for legal reforms, as well as the reforms themselves. Legal equality framings often operate to pacify, neutralize, and redirect movements seeking transformative change, shifting demands for significant redistribution and restructuring into narrow demands for symbolic inclusion and state declarations of equality that do not alter material conditions of harm and violence. Legal reform often transforms only enough to stabilize and preserve the status quo, containing the threat of change represented by disruptive social movements and alternative ways of life by very slightly altering legal regimes but preserving harmful outcomes.[8] The concept of “discrimination” as it is framed by US legal equality regimes dehistoricizes and individualizes the harms of systems like white supremacy, settler colonialism, and heteropatriarchy by framing them as issues of individual prejudice and isolated bad acts. Such a framing makes it impossible to properly describe or remedy the harms of these systems, narrowing the scope of concern so severely that most conditions of harm and violence are not addressed. Because of the limits of legal framings, legal equality projects rarely meet the redistributive goals of populations in whose names they are created, and they can actually serve to obscure structural conditions of disparity.[9]

In the context of a growing racialized and gendered wealth divide in the US and globally, the expansion of criminalization and immigration enforcement systems and the “War on Terror,” critics charge that the cry for formal legal equality that has become synonymous with “LGBT rights” has produced victories that are little more than window dressing for racialized–gendered systems of violence and maldistribution that continue to shorten the lives of people impacted by homophobia and transphobia.[10] The dismantling of welfare programs, the expansion of criminal punishment systems, the criminalization of social movement work that demands transformative change, and the expansion of non-profitization have effectively shifted the political terrain such that the range of legible political demands only extends to those reforms that slightly tinker with and ultimately strengthen projects of militarism, criminalization, and deregulation of labor, environment, and capital.[11]

In this legal equality-centered politics, resistance to conditions of homophobia and transphobia has been narrowed in the last decades to an agenda that primarily seeks inclusion in systems and institutions that left movements have long identified as centers of state violence. Critiques of the ongoing criminalization of queer and trans life[12] are mostly ignored, crowded out by demands for increased criminal punishment in the form of hate crimes statutes, despite the lack of evidence that such laws prevent homophobic and transphobic violence.[13] Longstanding feminist, anti-racist and anti-colonial critiques of marriage have been silenced by a loudly articulated request for entrance into the institution of marriage with all its duties and privileges.[14] Critiques of the US military’s relentless violence, including its heteropatriarchal violence, are drowned out by celebration of the “right to serve” for “gay and lesbian Americans.”[15] In the context of this equal rights politics, the legible demands have become centered on removing formal barriers to entrance to the key institutions that organize US life and sustain its colonial, racial, and gendered violence. Critics have raised concerns not only about how this legal equality agenda rehabilitates and redeems those institutions, but also how it fails to address the ongoing harms of homophobia and transphobia. As the vulnerability of people facing homophobia and transphobia, especially those targeted by police and immigration enforcement, and those without property, employment, or health benefits expands, increasing formal state recognition through marriage, anti-discrimination law, and military service offers little respite for those facing the most dangerous manifestations of homophobic and transphobic violence.

Additionally, legal equality campaigns have been critiqued for the ways that they tend to produce narratives of deservingness that divide constituencies facing harm and violence. Rights claims frequently include messaging about how the marginalized group “deserves” rights, and that deservingness rests on portrayals of the group as “normal,” “hardworking,” “law-abiding citizens,” and other such tropes. These frames of deservingness reify key divides in US culture and politics that establish and justify the marginalization, criminalization, and abandonment of people of color, prisoners, people with disabilities, people on public assistance, and other demonized populations. Deservingness is constituted through talking points that distinguish the rights-seeking group from those who are implicitly understood as undeserving. In the context of equal rights campaigns, these portrayals have sought to establish deservingness by portraying propertied same-sex couples who “deserve” children because of their wealth, monogamy, patriotism, and heteronormativity. The legal rights sought through these campaigns tend to match up to the deserving-constituency frameworks established. The reforms make adjustments that are most likely to benefit those whose lives match the image of the deserving rights-seeking couple and are very unlikely to change or improve the circumstances of those facing brutal homophobia and transphobia outside of the circle of deservingness.

In sum, critics of legal equality campaigns have identified at least four key concerns. Legal equality campaigns tend to achieve what Critical Race Theorists have called “preservation through transformation,[16] transforming what the law says about a marginalized group but preserving the underlying status quo of maldistribution and state violence that group faces. Second, legal equality campaigns portray the problem as individual discrimination, obscuring the realities of systemic maldistribution and violence. Third, rights advocacy tends to produce narratives about institutions and systems in which advocates seek equality as good and fair systems that need only include the missing population so that they can join the exalted institution. Fourth, this advocacy produces binaries of deservingness and undeservingness in order to articulate belonging in exclusionary institutions and systems. The new rights tend to be crafted around these binaries, making them inaccessible for those in the worst circumstances. The mobilization of tropes of deservingess further demonize and endanger the most vulnerable members of the constituency.

Population and Normalization[17]

Some key concepts about how power works from the work of Michel Foucault are useful for unpacking the contemporary relationship between legal equality campaigns and the statistical methods. Foucault describes disciplinary and biopolitical modes of power as distinct from sovereignty, providing a framework that is particularly useful for examining the connections between how legal equality projects normalize the subject and how statistical methods normalize the population. Foucault describes sovereignty as power rooted in the “right to kill” (wielded against individual heretics or other disobedient people) or in “subtraction”—the power to take away. This subtractive power is wielded by the sovereign with the aim of obedience to the law and the maintenance of sovereign power itself.[18] Discipline, on the other hand, establishes norms of good behavior and ideas about proper and improper categories of subjects. Foucault famously traces the invention of certain categories of sexual subjects—the homosexual, the reproductive couple, the masturbating child—as he argues that relations of power produce these identities through an explosion of discourse about sexuality in the Victorian period.[19] Disciplinary practices congeal in certain institutional locations such as the school, the factory, and the clinic, where proper behavior is codified at the level of detail and subjects are formed to police ourselves and each other according to these norms.

Unlike sovereignty, and even discipline, biopolitics is concerned with population rather than individuals. Whereas sovereignty is defined by the right of the sovereign to kill, biopolitics is concerned with the distribution of life chances and the imperative to “make live,” to cultivate the life of the population.[20] Biopolitics develops as population grows and governments become concerned with counting birth and death rates, inventing public health initiatives and managing risks inherent in the population. Biopolitics emerges with new instrumental rationalities and new scientific methods for conceptualizing population, specifically statistics. As Grace Hong has described, “to create a population requires an apparatus that produces certain categories of statistical knowledge about that population.”[21] From an American law perspective, the rise of (often administrative) population-level interventions that we can understand as serving a caretaking function signal this mode of power. Immigration enforcement, social welfare programs, or multi-dimensional campaigns like “welfare reform,” the “War on Drugs” or the “War on Terror” that mobilize a range of legal and administrative technologies (e.g., education policy, criminal punishment systems, methods of recordkeeping, family law doctrines, public housing regulations, surveillance technologies) are all examples of population-level interventions. These interventions are mobilized in the name of promoting the life of the national population against perceived threats, and they operate by sorting and producing regularities rather than by individual targeting. This regulatory mode of power is not concerned with obedience to law, but rather has multiple and diffuse aims.[22]

Foucault suggests that the link between biopolitics and discipline is “the norm”:

In more general terms, we can say that there is one element that will circulate between the disciplinary and the regulatory, which will also be applied to the body and population alike, which will make it possible to control both the disciplinary order of the body and the aleatory events that occur in the biopolitical multiplicity. The element that circulates between the two is the norm. The norm is something that can be applied both to a body one wishes to discipline and a population one wishes to regularize.[23]

Norms of behavior that operate at the individual level and that are incorporated by subjects into their self-understanding are essential to discipline. Biopolitics mobilizes norms at the population level through sorting technologies that produce structured security and insecurity for various populations in the distribution of life chances.

Foucault’s description of biopolitics as a form of power concerned with cultivating life, “making live,” raises the question of how genocide, massacre, and other killing can occur in the context of this life-giving biopolitical power. Foucault identifies “state racism” to answer this question. He explains that this population-focused power concerned with promoting life always includes the identification of threats and drains to the population, and that the destruction or killing of these threats in order to preserve and promote the life of the population is always present in biopolitics. Achille Mbembe describes Foucault’s notion of racism in the context of biopolitics by saying:

This control presupposes the distribution of human species into groups, the subdivision of the population into subgroups, and the establishment of a biological caesura between the ones and the others…. In Foucault’s terms, … the function of racism is to regulate the distribution of death and to make possible the murderous functions of the state. It is, he says, “the condition for the acceptability of putting to death.”[24]

Importantly, Foucault explains that killing “do[es] not mean simply murder as such, but also every form of indirect murder: the fact of exposing someone to death, increasing the risk of death for some people, or quite simply, political death, expulsion, rejection, and so on.”[25] Thus, the function of biopolitical power is not the “right to kill” as in sovereignty, but the power to “make live and let die.”[26] This kind of power, and its specific concern with normalizing the population, relies heavily on the collection and analysis of standardized data, statistics, and statistical measure as a feature of biopolitics. Identifying and sorting the population and the environment in order to identify risks, threats, and drains becomes a central preoccupation of this kind of power.

Foucault explains that unlike sovereignty, which regards obedience to the law as its primary aim, governmentality is concerned with “the right manner of disposing things so as to lead to … an end which is ‘convenient’ to each of the things that are to be governed.” He writes, “… with government it is a question not of imposing law on men, but of disposing things: that is to say, of employing tactics rather than laws, and even of using laws themselves as tactics—to arrange things in such a way that, through a certain number of means, such and such ends may be achieved.”[27]

This understanding of laws as tactics in the context of decentralized practices of governance, rather than as determinative, is useful for conceptualizing the limitations of legal equality and inclusion claims. It also provides a way of understanding how accounting contributes to the distributions that occur through certain vectors of population or identity. First, recognizing the prominence of technologies that sort the population through purportedly neutral, banal, and scientifically grounded administrative apparatuses and seeing that those are the locations at which life and death are distributed helps us to recognize state declarations of equality in the form of discrimination laws or hate crime laws as window dressing to material arrangements that actually shorten lives. The belief that declaring illegal explicit exclusion from employment, military service, or marriage on the basis of sexual orientation is a meaningful remedy for the harm done by heteropatriarchy is unsupportable when we recognize the significance of population-level operations of power. Further, Foucault’s and Mbembe’s descriptions of the processes of sorting the population into life that must be cultivated and life that constitutes a threat or drain, draws our attention to the significance of normalization and of the production of purportedly universal categories of “us” in whose name national resources are mobilized for various forms of killing.

This analysis provides a critical framework from which to examine the intertwined natures of legal equality claims and empirical claims. It suggests that the gathering and dissemination of statistical data and the articulation of demands for legal equality share a grounding in notions of citizenship and population that produce and maintain racialized–gendered national norms. The assertion of these norms and their production are in fact a central goal and strategy of the statistical projects being undertaken to support legal equality claims for “LGBT rights.”

Social Development of Statistical Methodologies

In order to open up the questions that concern us about the use of statistical data to shore up arguments being made by legal equality advocates, it is useful to put our critiques of statistical methodology in the context of the existing critical conversations about how methods develop in and through social and political relations.[28] Here, we briefly review critical analyses of categorization, counting, and statistical methods; we explore some examples of the political expansion of these methods; and we introduce some ways of thinking about statistical methods, and the invention of the norm, specifically, as co-constitutive with processes of political organization.

It is important to understand that the collection and statistical analysis of data does not simply record or observe existing phenomena, but rather that analytic methods emerge in the context of social development.[29] Ian Hacking investigates categorization and study of people, saying, “[People] are moving targets because our investigations interact with them, and change them. And since they are changed, they are not quite the same kind of people as before. The target has moved…. Sometimes sciences create kinds of people that in a certain sense did not exist before.”[30] Fundamentally, counting a set of objects with multiple sources or contributing individuals requires that the objects be grouped into agreed-upon categories, which might be delineated differently depending on the interests of the counter. These categorical constructs, which emerge from, produce and reproduce social assumptions, are required for the meaningful concept of a number to be shared among people and arise alongside other technologies for sorting and managing populations and resources.

In his book Seeing Like a State, James C. Scott describes how the standardized collection of data constitutes and produces governing capacity for states and how the rise of standardized collection of data creates a new kind of “stateness” that characterizes modern nation-states. He looks at the evolution of those practices where local methods and measures are replaced by standardized ways of naming people, measuring land and crops, creating a standard national language, standardized ways of owning property, and the like. These processes of standardizing a certain way of seeing and inventorying things occurs both within colonizing countries that force the replacement of local customs with national standards on their own people and in processes of colonization where ways of organizing life through the terms of the colonizer—whether that is naming practices, land ownership schemes, family formation norms, racial categories or gender categories—are forced on colonized peoples.[31]

Scott offers the example of the standardization of measurement in France in the context of Napoleonic state-building to illustrate how stateness changes with the administration of standardized norms.[32] During this period, local practices of measurement were replaced by national standards. Prior to standardization, every village had its own pint or bushel size, its own rules about measurement, which were often manipulated by feudal lords to increase their ability to extract from peasants. The move to a standardized system of measurement facilitated trade and the extraction capacities of the central government. It was advocated by members of the Third Estate who were tired of being taxed unfairly by nobles who manipulated measures to extract higher rents. Yet it was also widely resisted by people who were used to their local measures and wanted to retain them. Local ways of measuring often fit local needs better than national measures. The significant tension surrounding the enforcement of national standard measures and the replacement of local practices of measurement often resulted in uneven implementation of national standards and outright resistance.

The idea of “equality” that was emerging during that time, which we see represented in the documents of the French Revolution and the Enlightenment more generally, was centrally about this new uniformity—about an idea of national citizenship that imagined:

a series of centralizing and rationalizing reforms that would transform France into a national community where the same codified laws, measures, customs and beliefs would everywhere prevail…. [It imagined] a national French citizen perambulating the kingdom and encountering exactly the same fair, equal conditions as the rest of his compatriots. … This simplification of measures, however, depended on that other revolutionary political simplification of the modern era: the concept of a uniform, homogeneous citizenship.”[33]

The fairness and equality called for during this period required new legal and administrative systems that aimed for uniformity and standardization that could produce a new legibility and a new ability to see and manage the country—resources and population—from the position of the central government.

Standardized measurement and counting procedures produce new kinds of governance that emerge with those regimes of knowledges and practices. In addition to illustrating this process of standardization and the production of new forms of stateness through the example of standardization of measures, Scott traces this process in the standardization of land tenure. Looking at a variety of contexts, including post-revolutionary France and Russia in the 1860s, he describes how land ownership changed as common land use was eliminated and new forms of land tenure were implemented. Common land use schemes typically involved villages determining the shared use of common lands between villagers in highly local, complex, and frequently changing arrangements. The elimination of the commons and the enforcement of “free hold estate” regimes where land belonged to a single owner rather than to whole villages made property relationships more legible to government, making it easier to extract revenue. Through the standardization of land tenure, a new relationship between individuals, populations and the government was established, in part because government gained new ways of counting, assessing, and knowing about land and therefore governing its use and users.

Enforcing standardized private-property regimes, methods of identity surveillance, family and gender norms, state-approved language and customs, and the like on indigenous people, of course, is central to processes of genocide and settlement that are ongoing in North America, the Pacific, Palestine, Australia, and elsewhere. The processes of producing “stateness” though colonization impacts its pace and methods, but Scott suggests that producing “stateness,” whether in a colonized territory or not, requires the forced replacement of local knowledge and practices with state norms. The processes that Scott describes, where local practices are eliminated and replaced by standardized practices and new ways of gathering data, can be understood as “state-building” practices that are ongoing, although the historical examples he uses regarding weights and measures and land ownership point to particular historical developments in this form of governance. Scott’s work helps expose how “the norm” emerges and circulates as both a statistical concept and a social concept and how the two co-constitute one another.

Ian Hacking’s work is useful for examining how the emerging standardized data collection and administration of categorical norms described by Scott further develops to produce the kind of governmentality described by Foucault as being characterized by state racism. Hacking’s work helps expose how math is an active force, not just measuring what exists but organizing social reality. Hacking argues that large-scale counting projects are launched to understand and alter some particular quantity, like maximizing profit or decreasing disease. He describes how the years between 1820 and 1840 were the period when we first see what he calls the “avalanche of printed numbers” which focused on surveying deviance.[34] During this period, the European reading public moved from non-numeracy to numeracy. The cholera epidemic and a purported “crime wave” in Europe were both reported and addressed with policies developed through the lens of statistical data. Reporting on these phenomena familiarized the public with statistics and numerical thinking as a way of conceiving of threats and dangers.

The identification of population-level risks in the form of threats and drains was a new way of understanding and constituting the nation. Statistical methods, essential for the interpretation of these population-based quantitative figures, produced new categories of people and events and new forms of knowledge and governance. Hacking suggests that the new legibility of the population through the advent of popular statistics transformed the social context by inventing and formulating new problems and subpopulations. Hacking writes, “enumeration demands kinds of things or people to count,”[35] asserting that “many of the categories we now use to describe people are byproducts of the needs of enumeration.” Hacking argues that the “human sciences … are driven by several engines of discovery, which are thought of as having to do with finding out the facts, but they are also engines for making up people.”[36]

We think of many kinds of people as objects of scientific inquiry. Sometimes to control them, as prostitutes, sometimes to help them, as potential suicides. Sometimes to organise and help, but at the same time keep ourselves safe, as the poor or the homeless. Sometimes to change them for their own good and the good of the public, as the obese. Sometimes just to admire, to understand, to encourage and perhaps even to emulate, as (sometimes) geniuses. We think of these kinds of people as definite classes defined by definite properties. As we get to know more about these properties, we will be able to control, help, change, or emulate them better.

The development of methods for gathering standardized data and the rise of certain counting methods enabled taxation, military conscription, and police work on a national scale and as a centralized practice. These methods produced new governing capacities for states—they created the forms of governance we now identify as states. The use of counting shifted with the invention of the norm to a use of calculation to seek averages, to assess risks at the population level, to imagine the population from a racialized perspective that seeks to identify and eliminate perceived threats and drains. Foucault, of course, describes the operation of biopolitical power and the role of state racism—the cultivation of the life of the population that requires the continual identification of populations marked as threats or drains such that the national population can be made to live while the perceived internal enemies can be left to die or killed in any number of ways. The assessment of risk at the population level is characteristic of biopolitics. Categories of deviancy and illness are co-constitutive with development of statistical methods to describe a population, and the move toward population management and control is what characterizes nation-state governing capacities.

Statistics and Eugenics

The process of producing categories of people for scientific examination, evaluation, and sorting that Hacking describes can be seen in the history of the creation of statistical methods. An examination of the key figures that developed statistical methods sheds light on the connection between population management, racialization, data collection, and statistical analysis. Quetelet, Galton, Pearson, Fisher, and other developers of statistics were leaders of European eugenics movements.[37] As discussed below, early eugenic movements varied widely in their specific goals and the means employed to achieve them, but promoting life and reproduction for some defined, desired populations and/or suppressing reproduction of undesired populations was a unifying aim of eugenics. The development of statistics is essential to define desired/undesired populations and perform eugenic projects.

The Gaussian Distribution

During the same period that the metric system was beginning to be enforced in France, Carl Friedrich Gauss, sometimes called “the Prince of Mathematicians,” invented a method of describing the behavior of a random variable whose values tend to fall around the average value. This is now known as the Gaussian distribution, normal distribution, or sometimes called the bell curve, and is represented in Figure 1.


Figure 1: Gaussian distribution.
A standard Gaussian distribution is shown such that a random variable following the distribution has probability proportional to the height of the curve of having any particular value. The mean (most probable value) and upper and lower tails (less probable values) are labeled.

In 1835, Adolphe Quetelet applied the idea of the Gaussian distribution to describe groups of people, calling these new descriptions of society “social physics.” Quetelet proposed that a Gaussian distribution could be used to analyze the distribution of people in a group and that the mean of this distribution represented the group “norm.” Quetelet was a teacher, mathematician, and astronomer and as such worked with the Gaussian distribution to estimate measurement errors in astronomical estimations.[38] He also took an interest in the quantitative study of human populations, particularly in rates of birth, death, and marriage, body shape and size, and “moral” behavior (such as rates of marriage, suicide, and crime).[39] Quetelet was familiar with astronomical measurement error and worked to promote standardization of measurement and categorization in both astronomy (i.e., time, distance, star type) and in “social physics” (i.e., height, weight, married, single, suicide, natural death). Quetelet strove to “perfect” the census and enable the study of “social physics.”[40] In one of the first efforts to quantitatively analyze human populations, Quetelet applied the Gaussian distribution to describe social observations, which controversially challenged the idea of free will and suggested that the behavior and physical attributes of human populations could be predicted.[41] Through his work, particularly his invention of the body mass index (BMI), which was motivated by the observation by actuaries that more death claims were reported for so-called obese policy holders, Quetelet gave the concept of the norm the status of an eugenic ideal that should be promoted by the state over other ways of being.[42] In his interpretation, deviation from the mean is considered as an error, so social deviance is devalued and discouraged.[43] For example, Quetelet observed that “regarding the height of men of one nation, the individual values group themselves symmetrically around the mean,”[44] following a Gaussian distribution.[45] He interpreted the mean of individual heights to define the ideal height for a group, where unusually tall or short individuals (individuals in the distribution tails)[46] were too tall or too short.[47] Quetelet went on to argue that differences in mean height between groups serve as evidence that race is rooted in fundamental differences between “peoples.”[48] Quetelet’s preference for rankable categories and his aversion to the unusual were extended to the interpretation of a probability distribution. Through Quetelet’s intervention, we can see that with the development of the norm “comes the concept of deviations or extremes.” The idea of the statistical norm “divides the total population into standard and nonstandard subpopulations.”[49]

In the late 1800s, prominent eugenicist and statistician Sir Francis Galton invoked a Gaussian distribution to model social classes defined by earned income and argue that they could be attributed to differences in “genetic worth.”[50] Galton imposed his social hierarchical ideas onto distributions of measurable characteristics in a different way than Quetelet did. He favored one ‘high’ extreme over the rest of the distribution.[51] Galton aimed to alter the genetic make-up of the population to increase “intelligence,” height, physical capacity, whiteness, and the like, thus “improving” the population. Galton coined the word “eugenics” to describe this strategy. He advocated for eugenic public policies to encourage some people (high-earning, white, intelligent) to reproduce, such as state-funded incentives for marriage between individuals from high-earning families, and prevent others (poor, disabled, criminalized) from reproducing. Through this work, Galton became widely regarded as the founder of the British eugenics movement.[52] Eugenic policies were promoted in the name of public health and welfare—the life of the population deemed to be the national population would be cultivated, and other kinds of life deemed to be threatening to the national population would be extinguished or encouraged to die off.

To support his eugenic agenda, Galton used a statistical approach to categorize and quantify human characteristics. For instance, Galton assumed that intelligence could be quantified and that quantified intelligence follows a Gaussian distribution.[53] In that system, Galton valued individuals with unusually high intelligence over individuals with average or unusually low intelligence. In Figure 1, this can be interpreted as valuing individuals with quantified intelligence in the upper tail of the distribution and devaluing individuals with quantified intelligence near the mean or in the lower tail. The dominant social interpretation of the Gaussian distribution shifted, indicated by a renaming to the now commonly used “normal distribution.”[54] Rather than depicting an ideal mean and undesirable deviance around that mean, as Quetelet had suggested, the mean began to be considered mediocre and undesirable; one tail was considered extremely unacceptable, and the other tail was considered the preferred state. Galton’s aspirational approach to the normal distribution—his desire to use statistical measure to intervene in human characteristics and change the population “for the better”—is clear in his literal reference to a distribution mean as the “mediocre point.”

This paradigm shift in the interpretation of a probability distribution, from valuing the mean and devaluing the tails, to valuing one tail and devaluing the other tail and common values, illustrates the key role played by political and social interests in interpretations of statistical data. Despite the divergent values they placed on different portions of the normal curve, both Quetelet and Galton used the same fundamental strategy of promoting racist agendas in their interpretations of a probability distribution.[55]

Regression Towards the Mean

The social reinterpretation of the Gaussian distribution was just one part of the new quantitative technologies and perspectives emerging with the increasing momentum of the eugenics movement in Europe. Galton and other eugenicists proposed to improve the citizenry of a country, defining improvement as increased intelligence, height, heterosexuality, wealth, whiteness, physical capability, and the like in the population, achieved by altering the genetics of the population as a whole.[56] The eugenic movement called for selective breeding programs where individuals deemed less desirable (people with physical impairments, people of color, people with “deviant” sexuality or gender, poor people, people deemed insane) would be sterilized and individuals determined to be more desirable (white people, rich people, people deemed intelligent, people perceived to be able-bodied, heterosexual people) would be encouraged to reproduce and provided with support from the state.[57] However, eugenic strategies were not limited to selective breeding and sterilization programs, but included anti-union organizing (since unions advocated equal pay to workers deemed to be eugenically unequal), immigration restrictions, and atrophy of welfare.[58] All of this was justified in the name of public health and wellbeing—the life of the population deemed to be the national population would be cultivated, and others kinds of life deemed to be threatening to the national population would be extinguished.

To support eugenic projects, Galton used statistics to explain the heritability of variable traits through genetically related families.[59] Of particular note, Galton’s eugenics agenda inspired the work on trait inheritance in which he first described the statistical concept of “regression towards the mean.”[60] Galton found that the seeds of pea plants whose parents had particularly large seeds are often smaller than the seeds of their parents. Galton then considered a similar phenomenon in humans, finding that the children of unusually tall individuals are typically shorter than their parents and, conversely, that the children of unusually short individuals are typically taller than their parents.[61] He plotted average parent height against average child height, found a line to best fit this data, and described the best-fit line mathematically in an effort to explain and predict height.[62] For these analyses he is credited with inventing regression, a fundamental technique used widely in statistics today to help understand relationships and correlations.[63]


Figure 2: Gaussian distributed random components to height.
Based on the example in the text, the non-heritable random height component for a parent is unusually large (shown in the vertical line labeled parent). Assuming that the parent and child share the same genetic background, the probability that the child of the parent is taller by chance is the area under the curve to the right of the parent, and the probability that the child of the parent is shorter by chance is the area under the curve to the left of the parent.

Galton specifically used the term “regression towards the mean,” similar to his description of the mean as the “mediocre point,” because he believed that the upper tail of a Gaussian distribution (measuring, for example, seed size, height, intelligence, or physical capacity) is the point of aspiration. His observation that unusually tall individuals are likely to have children shorter than themselves represented a literal “regression towards the mean.” This language has persisted in statistics today in “regression analysis,” a linguistic marker of the eugenic origin of this class of statistical methods. Galton’s theory of genetic regression towards the mean over generations, it turned out, was flawed. However, his work was successful in promoting eugenics, regardless of the fact that it was later proved to be inaccurate.[64] The deployment of purportedly scientific or data-based arguments was effective regardless of whether those arguments were internally consistent.[65]

To support and facilitate the eugenics movement, new statistical techniques were developed by Galton, Karl Pearson, Ronald Fisher, and others both to analyze complex new genetic data and to predict the parameters of a eugenic program needed to obtain the desired results.[66] The eugenics movement in the United States advocated social reforms that many would today describe as explicitly racist, xenophobic, and anti-poor.[67] For example, the movement promoted legislation limiting the immigration of people from supposedly genetically inferior groups[68] and legislation calling for the sterilization of supposedly genetically inferior individuals.[69] However, social reformers of many kinds mobilized ideas being made popular by eugenics-focused scientists in order to argue that the changes they sought would improve the population and promote the good life. For example, statistically supported eugenic arguments famously received support from white feminists like Margaret Sanger working to improve access to contraceptives.[70]

Shifting Eugenic Interventions

Historians of statistics differ in their views about the relationship between eugenics and statistics. Some argue that it is essential to understand that the methods were shaped by their eugenic aims. According to Donald Mackenzie, “eugenics did not merely motivate [the] statistical work [of Galton, Pearson, and Fisher] but affected its content. The shape of the science they developed was partially determined by eugenic objectives. … [E]ugenics was the dog that wagged the tail of population genetics and evolutionary theory, not the other way round.”[71] Francisco Louçã, however, departs from this view, arguing that statistics is the “emancipated heir” of eugenics—that their paths “converged and diverged,” suggesting that statistical methods need not be approached with caution or suspicion related to their eugenic origins.[72] Tukufu Zuberi agrees with Louçã that eugenics and statistics have diverged, but Zuberi asserts that statistical genetics still “supports the use of race as a biological indicator of social difference.”[73]

The relationship between statistical methods and the public assertion of a eugenics agenda has shifted significantly since the early development of statistical methods by eugenics leaders. As statistical reasoning and methodology were further developed, it was found that in many cases the eugenic programs promoted by early developers could not achieve their stated goals of altering the characteristics of a population.[74] Over time, especially after World War II, when eugenics became associated with Nazism and genocide, the term “eugenics” and the idea of controlling the genetics of a population using mass sterilization and controlled breeding programs lost popularity.[75] However, reform programs that aim to cultivate national populations according to racial, economic, and sexual norms remain central to contemporary governance and continue to draw on fields of study that are either functionally eugenicist or have been influenced by eugenics.[76] In the 1950s in the United States, eugenics scholars[77] increasingly incorporated ideas about environmental influences, particularly childhood experiences, into their understanding of how to shape the population to meet norms. Rather than promoting the use of state-enforced breeding programs, these scholars produced studies that encouraged individuals to choose their co-parent based on their genetic potential and to provide the “correct” home environment for children, with the goal being to produce normative children.[78] During this same time period, the moral model of disability was replaced by the geneticized medical model and race became increasingly geneticized.[79] These shifts supported the idea that individual decisions regarding family formation were essential to producing genetically superior, i.e., norm-abiding children.[80]

Scholars in many fields have critically examined how programs central to the US administrative state operate to establish and enforce racialized gender norms and to forward projects of colonization and population control. They have pointed to many programs and legal and administrative schemes, including Indian boarding schools, welfare family caps, marriage promotion programs, curricula in public education, and immigration regulations that tie immigration status to family, health statuses, criminal history and class.[81] Through norm-enforcing programs aimed at behavior modification and those aimed at changing the biological make-up of the population more explicitly, state administrative capacity that relies on the collection of statistical data using racialized–gendered classification systems are mobilized to produce and/or protect a desired national population. Membership in the national population is distributed and capacities to reproduce are provided or terminated in ways that are significantly racialized and gendered.[82] These kinds of programs are decreasingly described as eugenics by their advocates in the period following World War II, but it would be naïve to assume that a change of terms and certain developments in methodology establish a clean break from the intertwining of eugenics with statistical methods and the racialized–gendered national population control projects they serve. Since the relationship between eugenics and statistics was obscured and buried, many of the statistical techniques developed for eugenics continue to be used without re-examination of the assumptions that undergird them. We are particularly interested in how methods originally designed for population management and normalization are still often used for those same purposes in new contexts.

Relying on the analytical frameworks developed by Ian Hacking, Mitchell Dean, and Michel Foucault, and more broadly in disability studies and women of color feminism, we can see how the intertwining of eugenics and the initial development of statistical methods continues to have effects today. Statistics connect intimate life behaviors and capacities, like those involved in reproduction, to the imperatives of “populations.” Processes that identify and promote the life of a national population by collecting standardized data that constructs who is inside that population and who constitutes a threat to that population are still central to the maintenance and growth of governing capacities, and still operate through the deployment of statistical data. Even if contemporary statistics has moved away from its eugenic origins, the demands of biopolitics continue to influence the scholarly field and the uses of statistics in political and legal discourses.

LGBT rights advocacy demonstrates the connections between demands for formal legal equality, the production of statistical data, and the promotion of narratives of deservingness and undeservingness rooted in racial nationalism. However, the concerns we raise about the mobilization of these investments are not limited to that particular strain of advocacy, but rather haunt many reform agendas. The interwoven strategies of legal equality reforms and the mobilization of statistical data to portray a rights-deserving population are ubiquitous in US social reform today. Social reform advocacy takes “the nation” as its target, and uses these methods to construct that target.

The Deployment of Empirical Data for Legal Equality

Gay and lesbian or LGBT legal equality discourse has been specifically critiqued for articulating a rights-centered politics that manifests significant problems and limitations. While maintaining neoliberal multiculturalism, it has articulated a white, wealthy, able-bodied, gender-conforming citizen as the purported universal subject of gay rights, seeking reforms that fail to disrupt and often support or exacerbate arrangements of maldistribution and state violence. It has participated in narratives of deservingness and undeservingness in order to shore up claims for “equality” that have made it complicit in and a site of reproduction of anti-immigrant, anti-poor, racist, and ableist logics.[83] In recent years, its efforts have increasingly relied on the production of data that articulates and categorizes sexual orientation (and sometimes gender identity) within the confines of a racial/national project that supports neoliberal arrangements, including the expansion of criminalization, privatization, austerity, and warfare. This politics of inclusion and recognition endeavors to make propertied/professional white lesbian and gay people (and to a lesser degree propertied/professional white trans people) into junior partners in white supremacy, settler colonialism, and heteropatriarchy.[84] “Gay and lesbian equality” advocacy often employs very conservative contemporary political frameworks: calls to law and order, calls to manage the harms of capitalism inside the marital family as poor relief programs are dismantled, and calls to invade and occupy purportedly “backward” countries portrayed as more homophobic and sexist than the US.[85]

The use of statistical analysis is essential to producing an idea of a LGBT population that meets racial national norms and that should be protected from national enemies alongside other proper citizens. The project of identifying and defining populations, or sorting the population, as Scott might say, produces narrow and exclusive categories. In the context of rights deservingness, and of seeking affirmation in American legal structures, that project is a racial one. Its reliance on statistical data roots it to a methodology designed to promote valued populations and diminish those cast as threats. Looking briefly at some of the ways the Williams Institute, a preeminent producer of such data, conceptualizes lesbian and gay (and sometimes LGBT) populations in order to support various legal equality claims illustrates the relationship between legal equality claims, statistical methods, and the production and circulation of racial–national norms.

In the following analysis we examine the implementation of statistical methods in several studies published by the WI. Our analysis of the problems within the WI’s deployment of statistical methods aims not to suggest that they should do their statistical analysis “right,” but rather to demonstrate that the desire for quantitative data to back up the LGBT legal equality advocacy agenda is so strong, and the deployment of numbers so requisite in policy work directed to the state, that the methods need not even be employed faithfully or consistently in order to be produced as persuasive evidence. We observe some examples of flaws in the WI’s work to show how advocacy goals, and the kinds of LGBT populations those goals aim to portray, govern the deployment of statistical methods. Not only are statistical methods intertwined with racialized–gendered population management projects in a way that raises questions about their use in advocacy, but the effects of those projects on knowledge formation can be so strong that they overwhelm even the consistent application of these methods.

First, we examine how the WI’s use of census data over-represents people in marriages or marriage-like relationships in accord with the central prioritization of same-sex marriage advocacy in the WI’s work. Next, we critically examine the WI’s estimate of the number of LGBT people in the United States. We then consider a particular study in the WI’s series of analyses estimating the economic impact of legalizing same-sex marriage, again focusing on assumptions made and examining how those assumptions might be convenient to the political aims of the WI’s legal equality agenda. Finally, we consider how results are reported (or excluded) across studies by the WI, noting how this influences reader interpretation.

Who Gets Counted?

Many studies published by the WI rely on US census-style data, that is, household surveys with individual information regarding categorical race and ethnicity, binary gender, identification of “head of household,” relationship to the “head of household,” birth date, and sometimes additional information. When considering census-style data, the WI typically limits their analysis to “same-sex couples,” which are defined as pairs of cohabiting individuals, where one is the “head of household” and the other defines their relationship to the “head of household” as either “married” or “unmarried partners,” and both mark the same gender box.[86] This method of data collection and classification establishes a narrow definition that many people who are targeted by homophobia and/or transphobia and/or identify as LGBT fall outside of if they do not organize their households or family lives according to the norms the WI is deploying to conceptualize the population. This group excludes individuals not in a cohabiting partnered relationships, couples living in a larger extended family structure (when neither couple member is the “head of household”), and individuals cohabiting with a partner using a different gender marker. In excluding these groups, many gender variant and trans people, people of color, and poor people are systematically excluded from study, and owning-class White same-sex couples are over-represented.[87]) Specifically, because of market forces at play in couple formation, poor people and people of color are less likely to be in a cohabiting relationship.[88] Individuals belonging to these same groups are also more likely to live in larger family structures, making them less likely to be named “head of household.”[89] Even if gender variant and transgender individuals are “heads of households” in cohabiting sexual relationships, they will only be counted if their partner marks the same gender category. In addition to this disproportionate representation, this study design is only capable of seeing and counting people who are married or in marriage-like relationships, which can cause an overrepresentation of higher-income people since marriage is correlated with higher wages.[90]

The WI’s focus on producing statistical data to support the campaign to legalize same-sex marriage is an incentive to design studies that depict same-sex couples as married or in marriage-like relationships, and to exaggerate the value of marriage to populations facing homophobia and transphobia and erase people who do not fit the WI definition of same-sex couples. Indeed, a notable portion of the WI’s funding is provided by donors for whom legal marriage equality has been a top priority, such as the Gill Foundation, the Arcus Foundation, and the Wellspring Foundation.[91] The WI’s prioritization of marriage is reflected in their 48 policy studies concerning the economic impact of extending marriage to same-sex couples (which, in almost every study, predict a net economic gain for the state), as compared to four policy studies published on HIV/AIDS and three on immigration.[92] The focus on marriage inclusion and neglect of other key survival issues aligns with the legal equality agenda that the WI’s data support.

A few studies published by the WI are based on data outside the census style. For example, in a study meant to explore which groups are over- and under-represented by the same-sex couple definition outlined above, an online survey was used to collect data through a third-party survey company.[93] The survey company enlists volunteers to take online surveys about a variety of subjects. People polled in this way are more likely to have easy internet access and time and energy to spend taking surveys, probably resulting in an underrepresentation of poor people, single parents, people of color, and people with less education access.[94] Perhaps because they were aware of these trends, the survey was targeted to an audience over-representing people of color. However, after all the data was collected, it was statistically re-weighted to resemble national demographic proportions, resulting in the down-weighting of survey responses from people of color.[95] The WI’s efforts to count a lesbian and gay or LGBT population constructs that population in ways that produce and enforce racial, gender, class, and family formation norms in order to produce data that fits legal equality arguments centered in racialized–gendered images of national citizenship.

Defining and Counting an LGBT Population

In 2011, the WI published a study entitled “How Many People Are Lesbian, Gay, Bisexual and Transgender?,” boldly declaring an estimate of the LGBT population in the United States.[96] The estimate is a composite figure drawn from eleven surveys performed between 2002 and 2011 by a variety of agencies across the United States and internationally that included questions regarding self-identified sexual orientation and/or gender identity. The surveys were administered using diverse methods (mail-in survey, phone interview, etc.) in a variety of geographic locations.

Several things about the study stand out. First, membership in the LGBT population is determined by self-identification, without regard to who may and may not self-identify in the categorical LGBT groups (typically constructed around a white middle-class norm and lacking other identifying terms that may have more resonance in populations of color, among indigenous people, and among people in street economies). Additionally, self-identification is likely to produce a lowered count of people engaged in same-sex sexual behavior or desire or gender nonconforming behavior or desire, as is acknowledged, but not examined, by the study.[97] This sort of population count may be useful to support arguments about the economic impact of legalizing gay marriage, but it is unclear that population counts of self-identified LGBT individuals are useful in the struggles against discrimination or health discrepancies, the other reported uses of this data. For instance, in addressing homophobia and transphobia-related health disparities, merely counting the number of self-identified LGBT individuals may not be as useful as collecting information from people about sexual practices, access to health care, experiences with health care providers and insurance companies, risk behaviors of various kinds, and other key factors. In fact, those people facing barriers to health care related to same-sex desire or practice or gender nonconforming desire or practice who do not self-identify as LGBT may be those who are least reached by interventions aimed at addressing homophobia and transphobia as barriers to health information and health care.

The estimate of the number of transgender individuals included in this report raises additional concerns, not least of which are how the surveys compiled to create the estimate define trans identity.[98] The estimate of the trans population is arrived at by compiling data from two prior surveys that used varying language to define transgender to survey participants. Both of these surveys, though in different ways, generally used self-identification as the indication of trans identity.[99] The WI authors characterize these surveys as using “questions that implied a transition or at least discordance between sex at birth and current gender presentation,” (emphasis added).[100] The authors then compare the estimate they have arrived at with an estimate from another study in an effort to use the consistency between the studies to suggest accuracy.[101] The study to which they compared their estimate articulated trans identity as “actually tak[ing] steps to transition from one gender to another.” This approach to defining a trans population utilizes key tropes of transphobia, including a reduction of trans identity to certain body modification practices and an erasure of trans people who do not engage in them. The authors’ characterization of these survey results in a way that emphasizes “transition” and identifies those who have not engaged in “transition” as meeting a somehow lesser criteria, alongside their uncritical invocation of an estimate that utilizes medical criteria for trans identity, has several concerning implications. It affirms transphobic understandings of trans identity that emphasize medical authority and exclude people who do not desire or cannot afford medical treatment from membership in the trans population. These exclusions are highly racialized and classed given the fact that gender confirming health care for trans people is mostly excluded from Medicaid programs nationally and the racial wealth divide means people of color have less access than white people.[102] This is particularly concerning given that these estimates of the population are explicitly being used to shape advocacy efforts. Additionally, the report’s comparison of its own estimate with the estimate arrived at by the survey using medical criteria suggests a willingness to under-represent or a failure to recognize the problem with these results if they appear consistent despite using significantly different criteria for defining the category. These issues indicate both a willingness to employ and rely on statistical data regardless of its obvious inaccuracy, and a lack of concern for key issues in trans politics, particularly those impacting low-income trans people and trans people of color.

These problems can also been seen in how, in interpreting some of the survey results, inconsistent assumptions are made that cause trans identity to be collapsed into LGB identities. The study attempts to estimate the proportion of the general population that is transgender. To create this estimate, it is assumed that all transgender people identify as LGB (so will be implicitly counted in surveys used to estimate the LGB population proportion). Many trans people who are not bisexual, lesbian, or gay may answer surveys in ways that do not identify them as such. These people will not be counted in the LGB estimate. Starting from the problematic assumption that transgender people are all included in LGB, the study then estimates the percentage of the total population that is transgender as the product of the estimated proportion of the total population which is LGB and the estimated proportion of the LGB population which is transgender.[103] In doing so, transgender people who do not identify as LGB are erased from this count, which also lowers the count of trans people in the general population. All of these awkward and inappropriate assumptions may seem reasonable when attempting to estimate the number of transgender individuals forced through a binary gender framework. Notably, using this method, the total population estimated percentage of transgender individuals is five times lower than the same percentage estimated directly in a different survey.[104]

The implications of the WI’s methods of defining the transgender population and its relationship to the LGB population could be the subject of an entire article, but a few implications are worth noting here. In addition to citing survey data that relies on transphobic definitions of trans identity that center medical authority rather than self-identity,[105] “How Many People Are Lesbian, Gay, Bisexual and Transgender?” conflates transgender identity with LGB identity in ways that directly disserve trans and gender nonconforming populations. Trans and gender nonconforming people have been consistently marginalized in lesbian and gay rights politics, sometimes through explicit exclusion and sometimes because resources are consistently devoted to issues impacting lesbian and gay people, while issues, often of great urgency, impacting trans and gender nonconforming people are ignored.[106] Conflating “T” with “LGB” supports the erasure of the specificity of harm and violence faced by trans people, whose struggles with a range of legal systems and administrative programs (prisons, foster care, homeless shelters, hospitals, immigration, Medicaid) that enforce binary gender norms often suggest quite different political priorities than the “marriage equality”-centered agenda promoted by the WI. Making many trans and gender nonconforming lives illegible and folding trans and gender nonconforming people into a politics centered on an exceptionally narrow understanding of what constitutes an LGB population and the interventions that population might seek perpetuates the constitutive racialized and classed transphobia of lesbian and gay rights politics.

Finally, as mentioned earlier, the inconsistency in research methods, sample sizes (number of individuals surveyed), and social contexts across the surveys contributing to the cumulative study “How Many People Are Lesbian, Gay, Bisexual and Transgender?” may dramatically affect the resulting estimates. For a rigorous analysis of diverse estimates, these factors must be taken into account. However, in the published study, equal weight is given to each of these surveys in the cumulative estimates, resulting in somewhat arbitrary figures. Despite this lack of rigor, because the desire for such numbers is so strong, these estimates were announced to much fanfare.[107] Their persuasive value is directly related to the perception of them as “scientific,” yet they fail to measure up even to the problematic statistical methods they purport to employ. These failures, however, point to the underlying investments of the counting project, the populations it cultivates and disposes of, and the racialized and gendered norms it produces and reproduces.

Alternate Estimate of the Gay Male US Population

To further illustrate the process by which an estimate of the number of LGBT people in the US such as the one engaged by the WI “makes up people,” to use Hacking’s words, we present a similar estimate made with alternate data sources more closely tied to sexual behavior. The exercise of coming up with a different estimate, using a different set of criteria for gay identity and different available data sets, illustrates how such projects invent, rather than discover and count, a population. As of August 2011, an estimated 2.08 million residents of the United States had profiles on, an online cruising site targeted towards men who want to have sex with men.[108] This is more than half the WI estimate of 4.03 million total gay men residing in the United States[109] and continues to grow by about 8,000 individuals each week.110 If one could learn what proportion of men who want to have sex with men also use, the estimated number of users in the United States could be divided by that proportion to get an estimate of the total number of men who want to have sex with men in the United States. That proportion is unknown. However, without differing from the strategies of assumption and speculation often employed by WI studies, we can manufacture an estimate. According to a survey of 609 men during Atlanta gay pride in 2002, 34 percent of the total sample reported having met a sexual partner through the internet.[110] If we ignore issues of sample bias and chronological inconsistency,[111] and speculate that half of men who find male sexual partners online use specifically,[112] we arrive at a total estimate of 12.24 million men desiring sex with men residing in the United States, more than three times the WI estimate. If we were to assume that, given the existence of many competing online websites for men who want to have sex with men, the proportion using might be even smaller, we would reach an even larger estimate of the number of men who want to have sex with men in the US.

To be clear, we do not propose to have accurately estimated the number of gay men or men who want to have sex with men in the United States. In fact, our estimate aims to expose the dangers of such endeavors. However, by using a different approach which, like the WI study, takes existing data and extrapolates from it to produce a particular gay population, we reveal the subjectivity of such a process. Obviously, while membership in is likely correlated with sexual behavior or desire, perhaps more so than self-identification, it does not define sexuality. This method is also subject to sample bias since users do not uniformly represent the population of gay men or men who want to have sex with men,[113] so certain groups will be over and under-represented based on these data. However, this back-of-the-envelope calculation can be contrasted with the WI estimate to illustrate problematic assumptions required to categorize LGBT and non-LGBT individuals and to extrapolate from limited data to describe a larger group. It is perhaps especially useful as a foil because, while the WI’s marriage-focused approach to producing statistics about LGBT people has often enforced family norms by over-representing people in marriages or marriage-like same-sex relationships and under-representing single people and people living in extended families, our use of data centers sexual practice or desire, imposing a different definition of gay identity that is also arbitrary and narrow. The contrast between the two approaches exposes how the purportedly objective task of counting populations is actually a task of defining populations and proposing interventions—the task of normalization or, in Hacking’s term, of “making up people.”

The Economic Impact of Marriage for Rhode Island

The WI has published many studies examining the economic impact of legalizing same-sex marriage, almost always finding it to be beneficial.[114] Predicting the economic ramifications of legalizing same-sex marriage requires many assumptions and estimations. These studies typically use methods that consider those populations most benefited by existing social and legal structures that distribute wealth and life chances, like legal marriage, in ways that rely on “family values” rhetoric often used to support racist, anti-poor agendas.[115] A useful example is the WI study of the economic impact of legalizing gay marriage in Rhode Island.[116] In this study, census-identified “married” or “unmarried partner” same-sex cohabiting couples are considered, and it is found that legalizing same-sex marriage will increase revenue to the state due to increased taxation of these couples, increased taxable wedding spending of same-sex couples, and decreased public benefits distributed when these individuals combine their incomes. Many problematic assumptions are made to produce statistical support for these assertions.

As explored earlier, the census-based methodology utilized by the WI produces a likely over-representation of wealthier same-sex couples. By using data that disproportionately represents couples with more disposable and taxable income, estimates of increased tax revenue to the State are likely be inflated. As a part of the calculations, estimates were made for changes in public benefits distribution by the State due to legalizing gay marriage. In these estimates, same-sex couples are assumed to receive public benefits at the same rate as different-sex couples, as assumption that discounts the economic impact of homophobia. In fact, a different WI study shows that same-sex couples experience poverty and receive public benefits at higher rates than their different-sex counterparts.[117] In estimating reduced rates of public benefits towards legally married same-sex couples, an implicit assumption is made that individuals in same-sex couples receiving public benefits are less likely to be eligible for those benefits when legally married. The assertion seems to rest on an idea that individuals will somehow become wealthier by marrying. It is difficult to understand the basis for this assumption. Perhaps it is assumed that poor people will marry people of greater means? However, research suggests that people are more likely to marry others with similar socioeconomic status.[118] The assertion that legalizing same-sex marriage in Rhode Island would decrease public benefits reliance is especially concerning because of how it aligns with anti-poor rhetoric in the US and eugenic ideas of using marriage policy and family law to eliminate undesirable populations. Attacks on welfare and other poverty programs in the 1990s and 2000s have used gendered-racialized images of poverty, like the trope of the “welfare queen,” to assert a notion that poverty is a result of moral laxity and is best solved by eliminating benefits programs and promoting marriage. Such assertions have deep roots in the use of marriage promotion of various kinds as a part of social control programs aimed at people of color in the US, particularly Black people.[119] The desire of same-sex marriage advocates to frame marriage as a cure for poverty has disturbing overlap with right-wing “family values” politics that has targeted people of color as in need of state intervention to promote marriage and was taken up in recent years under George W. Bush’s administration and continued by Barack Obama’s through the implementation of “marriage promotion” policies in social welfare programs.[120] In addition to all of these concerns, this advocacy echoes the eugenic principle that statistics can be used to measure a population and develop policy reforms that will improve the population by eliminating poor people cast as “dependent” and “draining.” In this case, legalizing gay marriage is proposed to decrease the number of individuals receiving public benefits, transforming them into members of married couples who somehow become more economically secure.

Other problematic assumptions are visible in this report as well. The study assumes that a tax structure that would cause legally married individuals to pay more taxes than unmarried individuals will not affect marriage rates, despite citing a study showing that women are significantly less likely to marry when that would cause them to incur a tax penalty.[121] Additionally, the spending of wedding guests is estimated with remarkable precision, but based on a series of arbitrary assumptions possibly inflating the final estimate,[122] so that the precision is not meaningful, revealing its use as a rhetorical device rather than an accurate estimate. Finally, potential harm to same-sex couples is ignored when wedding spending (whether by reducing savings or by increasing debt) is valued as increased revenue for the state. Given the significance of American consumer debt, unemployment, and the rising cost of weddings, these arguments raise concerns about the alignment of wedding spending arguments for legalizing same-sex marriage with a concern for the well-being of same-sex couples.[123] The combined arguments that legalizing same-sex marriage will increase consumption and tax revenue and decrease public benefits reliance relies on harmful framings of same-sex couples as possessing inflated wealth and of marriage as a solution to poverty.

Selective Representation of Results

The presentation of the WI studies represents important choices about which results to emphasize, which to exclude, and how to frame the results. Consistently across the WI studies, error bars are omitted from plots,[124] resulting in plots where small, statistically insignificant differences appear important. For example, the rates of same-sex couples identifying as “husband/wife” and “unmarried partner” are compared between states with different degrees of legal recognition of same-sex couples (legal marriage, legal civil unions or domestic partnerships (CU/DP), and no legal status).[125] The text in the study describing these data states that the difference in rates for partnered couples is not statistically significant[126] between states with and without legal CU/DP recognition. However, in a plot the raw difference is depicted without error bars that should be used to indicate the uncertainty.[127] If the error bars were included, it would be clear that any perceived difference is not statistically significant. The lax visual representation leads the reader to conclude that legal CU/DP recognition is correlated with higher rates of same-sex couples identifying as “husband/wife,” which is not supported by the data. The study appears to use this non-significant difference to argue that legal recognition of same-sex marriage will increase identification as “husband/wife” on the US census, and thus help facilitate counting of married same-sex couples. This sort of selective representation of non-significant results is used to persuade the readers of a particular point, regardless of the empirical evidence supporting that point.

More evidence of potentially misleading choices in visual representation of data can be seen in the WI’s landmark study estimating the number of LGBT individuals in the United States.[128] In that study, Figure 5 shows a bar plot comparing the estimated numbers of lesbian and bisexual women, gay and bisexual men, and transgender individuals. Before examining the plot, note that dividing individuals into those three separate categories is practically impossible since those identities are not mutually exclusive and are inconsistently defined, so some contrived categorization must be used. The plot shows the estimated number of individuals in each group. Using these estimates, it appears that transgender individuals make up about 8.7 percent of the LGBT population.[129] This result is not included in the published paper, though it may be interesting because it is much higher than the survey-based estimates.[130] Additionally, the plot representing these data appears to scale the percentage of lesbian and bisexual women by the total number of women, the percentage of gay and bisexual men by the total number of men, but the percentage of transgender people by the total number of people. These are clearly not comparable numbers, giving the transgender category roughly half the visual impact of the others, resulting in a visually powerful image where the proportion of the bar graph area representing transgender individuals is well under 8.7 percent of the total plot, leading readers to believe that transgender individuals make up a smaller percentage of the LGBT population than the data being described actually suggest.

Just as a result can be over-emphasized and interpreted (like plotting non-significant differences without error bars), some findings can be downplayed or ignored when they are inconsistent with the study objective. This is demonstrated in the previously mentioned study exploring same-sex couple identity by the lack of investigation of the evident finding that African Americans in same-sex couples are half as likely to be married or in a legal CU/DP as whites, which is a statistically significant difference.[131] In contrast to the previous non-significant result discussed, this result is mentioned in passing in the text, but the reader must turn to the appendix to see that the difference is statistically significant and no visual figure is provided to illustrate the point. Further investigation of this significant finding might indicate that the structures of marriage and legal partnerships are not as useful or beneficial for African Americans and better serve whites, perhaps because white people are more likely to have wealth, immigration status, private health insurance, and other benefits that can be shared through marriage. However, no more attention is given to the result.


The mobilization of statistical data in the promotion of same-sex marriage provides a site where the rhetorical significance of empirical claims is clear. Regardless of the consistency or verifiability of the data produced, even within its own methodology, its persuasive value is significant. Given the WI’s apparent success in mobilizing its data to support its projects, its methodological errors are worth noting—not because our critique is limited to these errors, but because the errors expose the ways data operates to produce an image of a population that matches the imperatives of biopolitics. Through survey-based studies, the WI uses statistical methods that were developed for population control to justify increased legal benefits and recognition for a population it generally casts as white, able-bodied, owning-class and possessing legal immigration status. At times these methods are implemented inaccurately, clarifying the significance of their role in producing a persuasive rhetoric. Placing demands for legal equality within this framework forms the politics of equality in ways that support the broader racialized (upward) redistribution projects of neoliberal reform.[132]

All uses of statistical methods rely on making certain assumptions, and the representation of all results based in such methods is inevitably rhetorical.[133] We are not suggesting that such methods must be abandoned, but rather that an awareness of their origins and an understanding of the persuasive role of “hard science” might lead us to examine how those assumptions and representations operate in contemporary advocacy projects that certainly do not consider themselves white supremacist projects or eugenics projects.

This analysis of the WI data neither attempts to be an exhaustive study of its work, nor to suggest that its practices are unusual. The link between legal rights claims and the production of statistical data, and the concerns we suggest about the production of racial-national norms that these strategies achieve, are not limited to lesbian and gay rights discourse nor to this particular think tank. Such strategies are visible across US social movements where non-profitization, among other forces, have pushed various strains of work toward “legible” demands that fit within existing economic and political arrangements.[134] However, lesbian and gay rights discourse is a particularly interesting location for examining these relationships because of the current visibility of quests for formal legal equality in that realm and the relatively sudden emergence of a proliferation of statistical data to support those quests. Rhetoric mobilized to support lesbian and gay rights has also become a site of articulation of key trends in neoliberal politics (pro-militarism, increased immigration control, dismantling of poverty alleviation, marriage promotion, reduced taxation for the wealthy, increased criminalization of people of color).

An understanding of how legal equality rhetoric aligns with the rehabilitation, legitimation, and expansion of racialized–gendered violence such as austerity, criminalization, and militarization provides a space to question what equality means in the context of the normalization of populations. The limitations of legal equality claims and their relationship to the collection of standardized data are interesting when considered alongside Scott’s discussion of the meaning of “equality” in the context of French Revolutionary fervor for standardized weights and measures. Scott’s analysis suggests that the “equality” articulated in nation-building documents of the mid-1700s in France and the US might be understood as an equality that was about producing a uniform relationship between certain white, male, propertied citizens and the government. Uniform weights and measures and individualized property ownership gave governments new powers to inventory, regulate, and administer and offered the promise (whether realized or not) of a new, rationalized relationship between white propertied men and the government (as opposed to the arbitrary extraction of wealth by nobles). That capacity to manage and control the population, at the level of population, expanded with the advent of eugenics and statistical methods. Eugenics rearticulated the logics that undergirded the racialized property statuses that founded the United States through the framework of scientific reason, both conceptualizing and implementing new forms of population control.

How do these genealogies still structure the frameworks of “equality” demanded through law and our ways of knowing about the population? How might this understanding of “equality” expose the ways that statistical methods and legal equality demands collaborate to sort the population and to promote the lives of those deemed worthy and diminish the lives of those framed as “threats” and “drains”?

The production of data about the gay and lesbian population (or sometimes the LGBT population) invents and describes the population in whose name rights demands are articulated. Decades of critical thinking from women of color feminism, queer theory, disability studies, and Critical Race Theory show the ways that rights claims rely on a false universalism while actually producing a narrowly articulated subject who can possess rights. Considering these insights in conjunction with the work of scholars like Scott, Foucault, and Hacking exposes the links between statistical methods, the production of normalized national populations, and the distribution of life chances. Returning to Foucault, we can begin to assess how both legal equality frameworks and the production of statistical data to support them produce images of a white settler national population that must be cultivated and protected in relation to “threats” and “drains” that must be eliminated, abandoned, or extinguished. The eugenics projects that motivated key developers of statistical methods, such as Galton, aimed to identify elements of the population that are favored and disfavored so that they can be cultivated or eradicated. The effects of these projects and methods remain with us. Examples of how lesbian and gay rights discourse develops in relation to these genealogies abound. When lesbian and gay rights discourse depicts gay and lesbian parents as “good parents” using statistical data and with the goal of winning recognition and benefits for same-sex couples, it articulates assertions about parenting that support the very racial–national norms that ensure that Black parents, parents in prison, Native parents, poor parents, and parents with disabilities will be targeted by the child welfare system.[135] When lesbian and gay rights discourse depicts lesbian and gay couples as American workers/property owners who deserve equal access to regressive tax policies, it mobilizes the same rhetorics that have produced significant growth in the wealth divide in the US in the last four decades.[136] When lesbian and gay rights discourse depicts the police and prisons as forces that must be mobilized to save and rescue gay and lesbian people from violence through the passage of hate crime laws, it participates in expanding a system that targets people of color and poor people with homophobic and transphobic violence every day.[137]

Legal equality arguments tend to present existing legal structures as generally fair and neutral but for a singular exclusion and to construct the excluded group as a population that deserves inclusion. Contemporary political and economic conditions have seen the emergence of gay and lesbian rights advocacy that makes legal equality arguments and increasingly backs them up by producing a statistical picture of the population. This work constructs desirable and undesirable populations, those deserving a chance at life and reproduction and those whose exile, imprisonment, or death is acceptable or even important for the survival of the nation.

While the framings currently articulated by the most visible and well-funded lesbian and gay rights discourse makes the links between legal rights, statistical methods, and racial–national normalization relatively obvious, the methodological and strategic questions that such an analysis raises are not limited to the most blatant sites of contradiction. All resistance projects must struggle with the problem of how conceptualizing a population in order to articulate a claim involves creating an image of constitutive others who are cast as threats and drains.[138] The queer and trans racial and economic justice–focused activism that operates as an alternative to the gay and lesbian rights framework must also struggle with these questions. This alternative politics also produces studies that support reform projects that, though different from the marriage/military/hate crimes reforms, still imagine a population that likely has a set of constitutive outsiders. Many formations that have sought to resist capitalism, white supremacy, ableism, heteropatriarchy, and colonialism have run up against this problem. They have produced divisions within their constituencies, required forms of violence and surveillance to manage people, and ended up disappointingly reproducing governmental functions that look all too similar to those the movements had originally opposed.

We engage this analysis not from a desire to be singular about method or strategy, but instead from an interest in tracing the relationships between certain ways of knowing and ways of governing. We are interested in tracking incentives and investments that travel with particular methods and strategies. Such relationships and incentives may root statistical methods and legal equality demands to dominant arrangements. While knowledge is implicated with power in potentially problematic ways, knowledge in the form of an active analysis of these implications is also necessary to transform those arrangements.


We thank Gary Gates and the Williams Institute at UCLA School of Law, for generously sharing the survey data analyzed here, and the individuals participating in that survey for sharing their information and experiences. We thank Britt Rusert, Craig Willse, Urvashi Vaid, Janet Jakobsen, and Soniya Munshi for editorial advice, and research assistants for citation work.

Figure 1: Gaussian distribution

A standard Gaussian distribution is shown such that a random variable following the distribution has probability proportional to the height of the curve of having any particular value. The mean (most probable value) and upper and lower tails (less probable values) are labeled.

Figure 2: Gaussian distributed random components to height

Based on the example in the text, the non-heritable random height component for a parent is unusually large (shown in the vertical line labeled parent). Assuming that the parent and child share the same genetic background, the probability that the child of the parent is taller by chance is the area under the curve to the right of the parent, and the probability that the child of the parent is shorter by chance is the area under the curve to the left of the parent.

Pages: 1 2 3 4 All Pages

Next page

  1. Grace Hong, The Ruptures of American Capital: Women of Color Feminism and the Culture of Immigrant Labor (Minneapolis: University of Minnesota Press, 2006), 78. [Return to text]
  2. M.V. Lee Badgett, Laura E. Durso, Angeliki Kastanis, and Christy Mallory, “The Business Impact of LGBT-Supportive Workplace Policies,” UCLA: The Williams Institute, May 2013. [Return to text]
  3. This article was originally drafted in 2010-2011, but publication of the collection of articles it is part of was delayed. We have sought to update the information included here where needed and we believe the arguments remain relevant, but our choices of which WI studies to analyze is representative of the time period in which we were writing. Similarly, our decision to use as the example web service in our alternative estimate of men who have sex with men was based on the popularity of that site at the time we were writing. [Return to text]
  4. See, for example, Jody Herman, Craig Konnoth, and M.V. Badgett, “The Impact on Rhode Island’s Budget of Allowing Same-Sex Couples to Marry,” UCLA: The Williams Institute, February 2011; Angeliki Kastanis, M.V. Lee Badgett, and Jody L. Herman, “Estimating the Economic Boost of Marriage Equality in Iowa: Sales Tax,” UCLA: The Williams Institute, December 2011; Brad Sears, Christopher Ramos, and M.V. Lee Badgett, “The Impact of Extending Marriage to Same-Sex Couples on the New Jersey Budget,” UCLA: The Williams Institute, December 2009; Niraj Choksi, “Gay Marriages Could Generate Hundreds of Millions in First Year of Legalization for 11 States, Studies Find,” Washington Post, August 14, 2014, [Return to text]
  5. Naomi Goldberg and M.V. Lee Badgett, “Tax Implications for Same-Sex Couples,” UCLA: The Williams Institute, April 2009; Michael D. Steinberger, “Federal Estate Tax Disadvantages for Same-Sex Couples,” UCLA: The Williams Institute, November 2009. [Return to text]
  6. Gary J. Gates, “Effects of ‘Don’t Ask, Don’t Tell’ on Retention Among Lesbian, Gay and Bisexual Military Personnel,” UCLA: The Williams Institute, March 2007; Gates, testimony on “Don’t Ask, Don’t Tell,” US House of Representatives, Armed Services Committee, Military Personnel Subcommittee, July 18, 2008. [Return to text]
  7. For example, on March 22, 2011, the WI announced that its executive director, Brad Sears, had testified on that date before the California Senate Committee on Government Organization about SB 416. “If passed, questions regarding sexual orientation, gender identity and gender expression, domestic partnership status, and the gender of a spouse would be added as voluntary demographic information in statewide surveys conducted or funded by the state…. Sears stated that the ‘collection of data on sexual orientation and gender identity is necessary because these are relevant demographic characteristics.’ He also said that ‘in the absence of such data, policymakers, businesses, and the public risk making decisions based on myths and stereotypes about LGBT people. The availability of this data would provide much needed information for state and local governments, businesses, social service agencies, community organizations, researchers, and the public.’” The Williams Institute is by no means alone in advocating for increasing government data collection about LGBT populations. Other national organizations such as the National Gay and Lesbian Task Force are actively advocating for a range of government data collection tools to add questions about LGBT identities. See, for example, “Demographic research on lesbians and gays emerges from shadows” or “Task Force launches next phase of Queer the Census campaign.” [Return to text]
  8. Reva Siegel, “Why Equal Protection No Longer Protects: The Evolving Forms of Status-Enforcing State Action,” Stanford Law Review 49.5 (1997): 1111–48; Angela P. Harris, “From Stonewall to the Suburbs? Toward a Political Economy of Sexuality,” William and Mary Bill of Rights Journal 14 (2006): 1539–82; Dean Spade, Normal Life: Administrative Violence, Critical Trans Politics and the Limits of Law (New York: South End Press, 2011). [Return to text]
  9. Alan David Freeman, “Legitimizing Racial Discrimination through Antidiscrimination Law: A Critical Review of Supreme Court Doctrine,” in Critical Race Theory: The Key Writings that Formed the Movement, ed. Gary Peller, Kimberlé Crenshaw, Neil Gotanda, and Kendall Thomas (New York: The New Press, 1995), 29–45. [Return to text]
  10. Anna M. Agathangelou, D. Morgan Bassichis, and Tamara L. Spira, “Intimate Investments: Homonormativity, Global Lockdown, and the Seductions of Empire,” Radical History Review 100 (Winter 2008): 120–143; Spade, Normal Life. [Return to text]
  11. Dylan Rodríguez, “The Political Logic of the Non-Profit Industrial Complex,” The Revolution Will Not Be Funded: Beyond the Non-Profit Industrial Complex, ed. INCITE! Women of Color Against Violence (Massachusetts: South End Press, 2007), 21–40. Reprinted in this issue. [Return to text]
  12. Ginia Bellafante, “Arrests by the Fashion Police,” The New York Times, April 5, 2013; Joey L. Mogul, Andrea J. Ritchie, and Kay Whitlock, Queer (In)justice: The Criminalization of LGBT People in the United States (Boston: Beacon Press, 2011); Sean Strub, “Think Having HIV Is Not a Crime? Think Again,” Huffington Post, October 29, 2013; Nicole Pasulka, “The Case of CeCe McDonald: Murder—Or Self-Defense Against a Hate Crime?,” Mother Jones, May 22, 2013; Eric A. Stanley and Nat Smith, eds., Captive Genders: Trans Embodiment and the Prison Industrial Complex (Oakland, CA: AK Press, 2011). [Return to text]
  13. Ryan Conrad, ed., Against Equality: Prisons Will Not Protect You (Maine: Against Equality Press, 2012). [Return to text]
  14. Conrad, Against Equality; Mark Rifkin, When Did Indians Become Straight?: Kinship, the History of Sexuality and Native Sovereignty (New York: Oxford University Press, 2011). [Return to text]
  15. Conrad, Against Equality. [Return to text]
  16. Siegel, “Why Equal Protection No Longer Protects,” n7; Angela P. Harris, “From Stonewall to the Suburbs?,” n7. [Return to text]
  17. Portions of this section are adapted from Dean Spade, “Laws as Tactics,” 21 Colum. J. Gender & L. 442 (2011). [Return to text]
  18. Michel Foucault, “Governmentality,” The Foucault Effect: Studies in Governmentality, ed. Graham Burchell, Colin Gordon, and Peter Miller (Chicago: University of Chicago Press, 1991), 103. [Return to text]
  19. Foucault, The History of Sexuality, Volume I: An Introduction (New York: Pantheon Books, 1978). [Return to text]
  20. Foucault, “Society Must Be Defended”: Lectures at the Collège of France 1975–1976, ed. Mauro Bertani, Alessandro Fontana, François Ewald, and Arnold I. Davidson, trans. David Macey (New York: Picador, 2003), 241. [Return to text]
  21. Hong, Ruptures of American Capital. [Return to text]
  22. “Government is defined as a right manner of disposing things so as to lead not to the form of the common good,… but to an end which is ‘convenient’ for each of the things that are to be governed. This implies a plurality of specific aims: for instance, government will have to ensure that the greatest possible quantity of wealth is produced, that the people are provided with sufficient means of subsistence, that the population is enabled to multiply, etc. There is a whole series of specific finalities, then, which become the objective of government as such.” Foucault, “Governmentality,” 95. [Return to text]
  23. Foucault, “Society Must Be Defended,” 253. [Return to text]
  24. Achille Mbembe, “Necropolitics,” Public Culture 15.1 (Winter 2003): 11–40, 17. [Return to text]
  25. Foucault, “Society Must Be Defended,” 256. [Return to text]
  26. Foucault, “Society Must Be Defended,” 241. [Return to text]
  27. Foucault, “Governmentality,” 95. [Return to text]
  28. See Miranda Joseph, Debt to Society: Accounting for Life Under Capitalism (Minneapolis: University of Minnesota Press, 2014): “I take accounting to be, like law, a set of rules and procedures that can and should be subject to critique not only as an instrument of established dominant powers but also as a performative force, a socially formative force” (30). [Return to text]
  29. See also Alain Desrosières, The Politics of Large Numbers: A History of Statistical Reasoning, trans. Camille Naish (Cambridge: Harvard University Press, 1998); Mary Poovey, The History of the Modern Fact: Problems of Knowledge in the Sciences of Wealth and Society (Chicago: University of Chicago Press, 1998) (“[I]n most texts that purport to describe the material world, even the numbers are interpretive, for they embody theoretical assumptions about what should be counted, how one should understand material reality, and how quantification contributes to systematic knowledge about the world…. [Yet] numbers have come to seem preinterpretive or somehow noninterpretive… [xii]); and Kim TallBear, Native American DNA: Tribal Belonging and the False Promise of Genetic Science (Minneapolis: University of Minnesota Press, 2013) (“Rather than being discrete categories where one determines the other in a linear model of cause and effect, ‘science’ and ‘society’ are mutually constitutive—meaning one loops back in to reinforce, shape, or disrupt the actions of the other, although it should be understood that, because power is held unevenly, such multidirectional influences do not happen evenly” [11]). [Return to text]
  30. Ian Hacking, “Making up People,” London Book Review 28.16–17 (2006): 23–26. See also Catherine Bliss, Race Decoded: The Genomic Fight for Social Justice (Stanford, CA: Stanford University Press, 2012) (examining how contemporary genomic science constructs new racial meanings and “new avenues for identity formation” [19]); and Britt Russert and Charmaine D.M. Royal, “Grassroots Marketing in a Global Era: More Lessons from BiDil,” Journal of Law, Medicine and Ethics, 39: 79–90 (examining the deployment of racial categories in pharmacogenomics and the marketing of pharmaceuticals). [Return to text]
  31. Andrea Smith, Conquest: Sexual Violence and American Indian Genocide (Massachusetts: South End Press, 2005); Scott Lauria Morgensen, “Homonationalism: Theorizing Settler Colonialism within Queer Modernities,” GLQ: A Journal of Lesbian and Gay Studies 16 (2010): 105–131, 116; Rifkin, When Did Indians Become Straight; J. Kēhaulani Kauanui, Hawaiian Blood: Colonialism and the Politics of Sovereignty and Indigeneity (Durham, NC: Duke University Press, 2008). [Return to text]
  32. See also Desrosières, The Politics of Large Numbers: “Creating a political space involves and makes possible the creation of a space of common measurement, within which things may be compared, because the categories and encoding procedures are identical” (9). [Return to text]
  33. James C. Scott, Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed, Yale University Press (New Haven: 1998) 32. [Return to text]
  34. Hacking, “Biopower and the Avalanche of Printed Numbers,” Humanities in Society 5 (1982), 289. [Return to text]
  35. Ibid., 280. [Return to text]
  36. In the article “Making up People,” Hacking identifies several engines of discovery which he says are also engines for making up people. The engines he lists are: 1. Count, 2. Quantify, 3. Create norms, 4. Correlate, 5. Medicalise, 6. Biologise 7. Geneticise, 8. Normalize, 9. Bureaucratise, 10. Reclaim our identity. [Return to text]
  37. Lennard J. Davis, “Constructing Normalcy: The Bell Curve, the Novel and the Invention of the Disabled Body in the Nineteenth Century,” The Disability Studies Reader, ed. Lennard J. Davis (New York: Routledge, 1997), 9–28. Garabed Eknoyan, “Adolphe Quetelet (1796–1874)—The Average Man and Indices of Obesity,” Nephrology Dialysis Transplantation 23 (2008): 47–51. [Return to text]
  38. Frank H. Hankins, Adolphe Quetelet as Statistician, diss., Columbia University, 1908. [Return to text]
  39. Ibid.; Eknoyan, “Adolphe Quetelet.” [Return to text]
  40. Hankins, Adolphe Quetelet. [Return to text]
  41. Hacking, The Taming of Chance (Cambridge: Cambridge University Press, 1990). [Return to text]
  42. Eknoyan, “Adolphe Quetelet.” [Return to text]
  43. Francisco Louçã, “Emancipation through Interaction—How Eugenics and Statistics Converged and Diverged,” Journal of the History of Biology 42 (2009): 649–684; Hacking, “Making up People”; Hankins, Adolphe Quetelet. [Return to text]
  44. Quetelet, Lettres à S.A.R. le duc régnant de Saxe-Coburg et Gotha, sur la théorie des probabilités, appliquée aux sciences morales et politiques (1846), viii, cited in “Adolphe Quetelet,” International Encyclopedia of Social Sciences (1968), [Return to text]
  45. Louçã, “Emancipation through Interaction.” [Return to text]
  46. See Figure 1. [Return to text]
  47. Davis, “Constructing Normalcy.” [Return to text]
  48. Quetelet wrote, “Every people presents its mean and the different variations from this mean in numbers which may be calculated a priori. This mean varies among different peoples and sometimes within the limits of the same country, where two peoples of different origins may be mixed together.” Qtd. in Hankins, Adolphe Quetelet. [Return to text]
  49. Davis, “Constructing Normalcy.” [Return to text]
  50. Donald A. MacKenzie, Statistics in Britain: 1865–1930 (Edinburgh: Edinburgh University Press, 1981), 15–21. [Return to text]
  51. “Galton cast his attention on the differences between individuals, on the variability of their attributes, and on what he would later define as natural aptitudes, whereas Quetelet was interested in the average man and not in the relative distribution of nonaverage men” (Desrosières, The Politics of Large Numbers, 113). [Return to text]
  52. MacKenzie, Statistics in Britain, 51–72. Interestingly, along with his work in statistics, Galton also invented the practice of police fingerprinting, exposing the perpetual link between interests identifying and promoting notions of normality and producing categories of racialized criminality. Davis, “Constructing Normalcy.” [Return to text]
  53. Davis, “Constructing Normalcy.” [Return to text]
  54. Ibid. [Return to text]
  55. “For Galton, it was not Quetelet’s “hommemoyen” but the outliers that should concern science: furthermore, for Quetelet, deviations from the norm were pathological, whereas for Galton they were the necessary condition for excellence.” Ian Hacking, The Taming of Chance (Cambridge: Cambridge University Press, 1990). Cited by Louçã at 658. [Return to text]
  56. Louçã, “Emancipation through Interaction.” [Return to text]
  57. Ibid. [Return to text]
  58. Celeste M. Condit, The Meanings of the Gene: Public Debates about Heredity (Madison: University of Wisconsin Press, 1999). [Return to text]
  59. Louçã, “Emancipation through Interaction.” [Return to text]
  60. Francis Galton, “Regression towards Mediocrity in Hereditary Stature,” The Journal of the Anthropological Institute of Great Britain and Ireland 15 (1886): 246–263. [Return to text]
  61. Galton, “Regression towards Mediocrity.” [Return to text]
  62. Ibid. [Return to text]
  63. MacKenzie, Statistics in Britain, 51–72. [Return to text]
  64. Interestingly, Galton mistakenly attributed his observations to a remote ancestral influence. That is, Galton theorized that children of tall parents are sometimes shorter due to inherited shortness from their grandparents, great-grandparents, and so on. Today statisticians explain observed regression towards the mean simply by the shape of a Gaussian distribution. In the case of height inheritance, consider a model where height is explained partially by inherited genetics and partially by Gaussian distributed random variation (Fig. 2). Individuals who are unusually tall are likely to have both a genetic background and random variation contributing to their height. The children of these individuals will share their genetic background, so their genetic contribution to height will give them a tendency to be tall. However, the random component of their height is likely to increase their height less than that of their parents. Since the parents’ random height component was unusually large, the children’s are likely to be smaller, because choosing a more extreme outlier is unlikely by definition. As an illustration, consider a Gaussian distribution (Fig. 2) and one random Gaussian-distributed number happens to be in the upper tail. The chance that the next random number will be greater than the first is proportional to the area under the Gaussian curve to the right of the first number, which is small. Notably, although Galton explained regression towards the mean inaccurately, his ancestral-input theory did support his eugenic agenda (the idea that desired characteristics could be determined precisely through generations of selective breeding), unlike the more scientifically rigorous explanation of random chance (desired characteristics are subject to non-heritable variation, so cannot be totally determined through selective breeding). [Return to text]
  65. Condit, Meanings of the Gene, 46–62; MacKenzie, Statistics in Britain, 51–72. [Return to text]
  66. Louçã, “Emancipation through Interaction.” [Return to text]
  67. The American Eugenic Society defined a partial list of “eugenic problems” which specifically included feeblemindedness, criminality, epilepsy, prostitution, rebelliousness, manic depression, nomadism, ethnicity, inferior races, birth defects, moral perversion, schizophrenia, racial hygiene, homosexuality, immigration, poverty, and feminism. Stephen Jones, “Zoology 61: Teaching Eugenics at WSU,” Washington State Magazine (2007). [Return to text]
  68. Eugenicists were key proponents of the Immigration Act of 1924, which severely restricted immigration, particularly of Jews and Southern and Eastern Europeans. Condit, Meanings of the Gene, 3–24. [Return to text]
  69. Between 1907 and 1931, 30 states passed laws calling for sterilization of disabled people, resulting in the sterilization of at least 30,000 individuals. Tukufu Zuberi, Thicker than Blood: How Racial Statistics Lie (Minneapolis: University of Minnesota Press, 2003), 58–79; Condit, Meanings of the Gene, 3–24. [Return to text]
  70. Condit, Meanings of the Gene, 27–45. [Return to text]
  71. Mackenzie, Donald. Statistics in Britain, 1865–1930: The Social Construction of Scientific Knowledge (Edinburgh: Edinburgh University Press, 1981) at p. 12, cited in Louçã, “Emancipation through Interaction,” 650. [Return to text]
  72. Louçã, 682. [Return to text]
  73. Tukufu Zuberi, Thicker than Blood. [Return to text]
  74. Louçã, “Emancipation through Interaction.” [Return to text]
  75. Condit, Meanings of the Gene, 65–81. [Return to text]
  76. Specifically, in 1957 “crypto-eugenics” was promoted in the Eugenics Society as a strategy to subtly continue to support eugenics through partnerships with other organizations, particularly those working towards accessible birth control. In 1989 the group changed its name to the Galton Institute ( and began focusing on funding birth control in poor countries. MacKenzie, Statistics in Britain, 45–46. [Return to text]
  77. Even though eugenics was losing popularity after World War II, academic programs and institutes devoted to eugenics continued to exist. Specifically, eugenics courses were taught in public land-grant universities until 1972. Leland L. Glenna, Margaret A. Gollnick, and Stephen S. Jones, “Eugenic Opportunity Structures: Teaching Genetic Engineering at US Land-Grant Universities Since 1911,” Social Studies of Science (2007) 37: 281–296. [Return to text]
  78. Condit, Meanings of the Gene, 63–96. [Return to text]
  79. Eli Clare, Exile and Pride: Disability, Queerness, and Liberation (Cambridge: South End Press, 1999). [Return to text]
  80. Condit, Meanings of the Gene, 63–96; Troy Duster, Backdoor to Eugenics (New York: Routledge, 2003), 3–20. [Return to text]
  81. See, for example, Mae M. Ngai, Impossible Subjects: Illegal Aliens and the Making of Modern America (Princeton: Princeton University Press, 2004); Smith, Conquest; Loretta Ross, “The Color of Choice: White Supremacy and Reproductive Justice,” in The Color of Violence: The INCITE! Anthology, ed. INCITE! Women of Color Against Violence (Massachusetts: South End Press, 2006), 54–65; Gabriel Chin, “Regulating Race: Asian Exclusion and the Administrative State,” Harvard Civil Rights–Civil Liberties Review 37.1 (2002). [Return to text]
  82. See, for example, Smith, Conquest; Kenneth J. Neubeck and Noel A. Cazenave, Welfare Racism: Playing the Race Card Against America’s Poor (New York: Routledge, 2001); Loretta Ross, “The Color of Choice”; Condit, Meanings of the Gene; Duster, Backdoor to Eugenics; Zuberi, Thicker than Blood. [Return to text]
  83. Mattilda Bernstein Sycamore, ed., That’s Revolting! Queer Strategies for Resisting Assimilation (Brooklyn: Soft Skull Press, 2004); Ian Barnard, “Fuck Community, or Why I Support Gay-Bashing,” States of Rage: Emotional Eruption, Violence, and Social Change, ed. Renée R. Curry and Terry L. Allison (New York: New York University Press, 1996), 74–88; Cathy J. Cohen, “Punks, Bulldaggers, and Welfare Queens: The Radical Potential of Queer Politics?” GLQ: A Journal of Lesbian and Gay Studies 3.4 (1997), 437–465; Ruthann Robson, “Assimilation, Marriage, and Lesbian Liberation,” Temple Law Review 75 (2002), 709; Anna M. Agathangelou, D. Morgan Bassichis, and Tamara L. Spira, “Intimate Investments: Homonormativity, Global Lockdown, and the Seductions of Empire,” Radical History Review 100 (2008), 120; Christina B. Hanhardt, “Butterflies, Whistles, and Fists: Gay Safe Streets Patrols and the ‘New Gay Ghetto’ 1976–1981,” Radical History Review 100 (2008), 61; “Is Gay Marriage Racist?” A Conversation with Marlon M. Bailey, Priya Kandaswamy, and Mattie Udora Richardson in Sycamore, That’s Revolting!; Kenyon Farrow, “Is Gay Marriage Anti-Black?,” March 5, 2004; Chandan Reddy, “Time for Rights? Loving, Gay Marriage and the Limits of Legal Justice,” Fordham Law Journal 76 (2008), 2849. [Return to text]
  84. Frank B. Wilderson III, “The Prison Slave as Hegemony’s (Silent) Scandal,” Warfare in the American Homeland: Policing and Prison in a Penal Democracy, ed. Joy James (Durham: Duke University Press, 2007), 23–34. [Return to text]
  85. Jasbir Puar and Amit Rai, “Monster, Terrorist, Fag: The War on Terrorism and the Production of Docile Patriots,” Social Text 20.3 (Fall 2002): 117–148; Jasbir Puar, Terrorist Assemblages: Homonationalism in Queer Times (Durham: Duke University Press, 2007); Katherine Franke, “Dating the State: The Moral Hazards of Winning Gay Rights,” Columbia Human Rights Law Review 44.1 (2012); Sarah Schulman, “Israel and ‘Pinkwashing’,” The New York Times, November 22, 2011; Pinkwatching Kit, Pinkwatching Israel, May 24, 2012; Tom W. Smith, “Cross-National Differences in Attitudes about Homosexuality,” April 2011 (a WI-supported study finding that “ex-socialist” countries are more homophobic). [Return to text]
  86. Gates, “Same-Sex Spouses and Unmarried Partners in the American Community Survey, 2008,” UCLA: The Williams Institute, 2009; Gates, “Same-Sex Couples in US Census Bureau Data: Who Gets Counted and Why,” UCLA: The Williams Institute, 2010; Herman et al., “Impact on Rhode Island’s Budget.” [Return to text]
  87. We re-examined the data presented in Gates, “Same-Sex Couples in US Census Bureau Data: Who Gets Counted and Why,” UCLA: The Williams Institute (2010), which includes demographic information for LGBT individuals in cohabiting sexual relationships and their status as counted or not counted using the definition above. Comparing individuals who are counted and not counted, we found that counted individuals are statistically significantly (p = .008) higher income than people not counted.
    – We used a one-sided Student’s t-test to test the null hypothesis that the average income for counted individuals (people who answered the census with “husband or wife” or “unmarried partner” [q810] and also one partner of which is “head of household” [q805]) is lower than uncounted individuals (people who answered census with “other nonrelative” or “housemate or roommate” or where neither partner is “head of household”). We considered income with the available binned discretized data (q462) and assumed that the income of each individual was the mid-point of their bracket and people making “$250,000 or more” were making $300,000.
    African Americans are statistically significantly (p = .005) less likely to identify as “husband or wife” or “unmarried partner” than Whites.
    – We used a one-sided Fisher exact test against the null hypothesis that the odds ratio is more than 1.0 on a contingency table comparing race marker (q485, “White” and pooled “Black” and “African American”) and relationship (q810, pooled “husband or wife” and “unmarried partner” and pooled “other nonrelative” and “housemate or roommate”).
    African Americans are nearly statistically significantly (p = .06) less likely to be counted than Whites. If there was more data available from African American individuals, the statistical power of the test would increase and the difference might become statistically significant.
    – We used a one-sided Fisher exact test against the null hypothesis that the odds ratio is more than 1.0 on a contingency table comparing race marker (q485, “White” and pooled “Black” and “African American”) and counted and uncounted status (as defined above [Return to text]
  88. A 2002 survey of US residents 15–44 years of age found higher marriage rates for White people than Black people and found that poor people have lower marriage rates than people who are not poor. The survey found that 45.4 percent of Latina women, 37.2 percent of White women, and 25.8 percent of African American women were currently married. Similarly, 42.7 percent of Latino men, 44.4 percent of White men, and 31.5 percent of African American men were currently married. It also found that 35.8 percent of women and 39.5 percent of men at or under the poverty line were married, as opposed to 60.7 percent of women and 52.0 percent of men at or over one and a half times the poverty line. Centers for Disease Control and Prevention, “Fertility, Family Planning, and Reproductive Health of U.S. Women: Data from the 2002 National Survey of Family Growth,” series 23, number 25 (December 2005), table 46; Centers for Disease Control and Prevention, “Fertility, Contraception, and Fatherhood: Data on Men and Women from Cycle 6 (2002) of the National Survey of Family Growth,” series 23, number 26 (May 2006), table 29. See also, “How Your Race Affects the Messages you Get,” OkTrends blog,, October 5, 2009; Hongyu Wang and Grace Kao, “Does Higher Socioeconomic Status Increase Contact between Minorities and Whites? An Examination of Interracial Romantic Relationships among Adolescents,” Social Science Quarterly 88 (2007): 146–166. [Return to text]
  89. “Preparing for Investments along the University Corridor: Income, Race, and Family Structure in the University Corridor,” Institute on Race and Poverty,, last visited 2011. [Return to text]
  90. Hal R. Varian, “Analyzing the Marriage Gap,” The New York Times, July 29, 2004; Kate Antonovics and Robert Town, “Are All the Good Men Married? Uncovering the Sources of the Marital Wage Premium,” The American Economic Review, May 2004; Hyunbae Chun and Injae Lee, “Why Do Married Men Earn More: Productivity or Marriage Selection?,” Economic Inquiry 39.2 (New York: Oxford University Press, 2001), 307–19; but see Ronald Mincy, Jennifer Hill, and Marilyn Sinkewicz, “Marriage: Cause or Mere Indicator of Future Earnings Growth?,” Journal of Policy Analysis and Management 28.3 (2009): 417–39 (finding that marriage promotion activities targeting low-income people and supported by research suggesting that marriage raises wages may be ineffective). [Return to text]
  91. Personal correspondence with Urvashi Vaid, former executive director of the Arcus Foundation and board member of the Gill Foundation, February 18, 2013 and June 2, 2013. [Return to text]
  92. These counts of studies are based on what studies had been posted to the WI website through the end of October, 2015. [Return to text]
  93. Gates, “Same-Sex Couples in US Census Bureau Data.” [Return to text]
  94. It is well-documented that internet access is not uniform over individuals with different ethnicities, ages, incomes, or education backgrounds. Karen Mossberger, Caroline J. Tolbert, and Michelle Gilbert, “Race, Place and Information Technology,” Urban Affairs Review 41 (2006): 583–620; Paul G. Harwood and Wayne V. McIntosh, “Virtual Distance and America’s Changing Sense of Community,” in Democracy Online: The Prospects for Political Renewal through the Internet, ed. Peter M. Shane (Routledge: Psychology Press, 2004): 209–224; “Internet and American Life Project: Demographics of Internet Users,” Pew Research Center; Steven P. Martin and John P. Robinson, “The Income Digital Divide: Trends and Predictions for Levels of Internet Use,” Social Problems 54 (2007): 1–22. [Return to text]
  95. Gates, “Same-Sex Couples in US Census Bureau Data.” [Return to text]
  96. Gates, “How Many People are Lesbian, Gay, Bisexual and Transgender?,” UCLA: The Williams Institute, April 2011. [Return to text]
  97. The bulk of the WI’s studies use census data and study cohabiting same-sex couples as a proxy for LGB populations. Even given the limits of self-identification data, this study could be used to contrast the different populations studied under each definition of LGB. However, no such analysis is presented, perhaps because it might raise challenges regarding sample bias. [Return to text]
  98. Clearly, from our perspective, the entire project of defining trans identity is problematic and concerning. The term transgender, transsexual, and trans are used with an enormous variety of meanings by various individuals, regional groups, subcultural groups, and language groups around the US. Other terms are also used extensively to indicate behavior or identification that exceeds or violates gender norms and makes those behaving or identifying in particular ways vulnerable. Transgender advocacy and scholarship has extensively examined debates about line-drawing around trans identities and the dangers and harms that come with rigid or narrow definitions that usually operate to exclude or deny many of those facing the worst violence of coercive gender systems and that often affirm authority and control for knowledges and institutions dominated by white professionals. Dean Spade, “Documenting Gender,” Hastings Law Journal 59 (2008): 731; Pooja Gehi and Gabriel Arkles, “Unraveling Injustice: Race and Class Impact of Medicaid Exclusions of Transition-Related Health Care for Transgender People,” Sexuality Research and Social Policy: Journal of NSRC 5.1 (March 2008): 7; Riki Anne Wilchins, Read My Lips: Sexual Subversion and the End of Gender (1997; repr. New York: Riverdale, 2013), 51; Franklin Romeo, “Beyond a Medical Model: Advocating for a New Conception of Gender Identity in the Law,” Columbia Human Rights Law Review 713 (2005). [Return to text]
  99. The study, “Transgender Health in Massachusetts” asked this question: “Some people describe themselves as transgender when they experience a different gender identity from their sex at birth. For example, a person born into a male body, but who feels female or lives as a woman. Do you consider yourself to be transgender?” The California LGBT Tobacco Survey asked the question this way: “We are also interested in speaking with adults who consider themselves to be transgender, or transsexual in any way. By this, I mean people who have a gender identity or presentation that is different from what society says you should have for your birth sex. Would you include yourself in this group?” K.J. Conron, G. Scott, G.S. Stowell, and S. Landers, “Transgender Health in Massachusetts: Results from a Household Probability Sample of Adults,” forthcoming; Field Research Corporation, “Lesbians, Gays, Bisexuals, and Transgender: Tobacco Use Survey, 2004,” California Department of Health Services (2004). Clearly, these questions are different in important ways. The first question much more closely adheres to an understanding of trans identity that is focused on transition from one distinct binary gender category to another, and treats binary sex as a natural fact rather than a social construct. Certainly, many gender nonconforming people whom we might imagine should be “counted” if the data is to be used to consider barriers and harms facing people because of gender nonconformity would not describe themselves in the narrow terms of being female but born in a male body or vice versa. However, a detailed analysis of the assumptions in these questions is beyond the scope of this article. Rather, we primarily mean to point to the difference between these questions and an approach that defines trans identity as requiring desire for and access to specific medical care. [Return to text]
  100. Gates, “How Many People…?,” 5. [Return to text]
  101. Although the estimates they compare vary by an order of magnitude. [Return to text]
  102. Dean Spade with Gabriel Arkles, Phil Duran, Pooja Gehi, and Huy Nguyen, “Medicaid Policy and Gender-Confirming Health Care for Trans People: An Interview with Advocates,” Seattle Journal for Social Justice 8 (Spring/Summer 2010): 497; Spade, “Documenting Gender”; Gehi and Arkles, “Unraveling Injustice.” [Return to text]
  103. For those inclined towards mathematical equations, this reasoning can be represented as P(T) = P(LGB) P(T|LGB). This equation is only correct if all T are included in LGB. If there are T outside LGB, the equation would need to sum over all possible conditions (for example non-LGB). [Return to text]
  104. The direct estimate of the percentage of transgender individuals in the Massachusetts Behavioral Risk Factor Surveillance Survey is 0.5 percent, while the estimate arrived at by combining data from the California LGBT Tobacco Survey and the California Health Interview Survey using the logic described is 0.1 percent. [Return to text]
  105. For a more detailed analysis of why evidence of medical treatment should not be used to define trans identity and the significant harm that is caused by the use of such criteria in numerous policies and programs, see Spade, “Documenting Gender,” 731; Gehi and Arkles, “Unraveling Injustice.” [Return to text]
  106. Shannon P. Minter, “Do Transsexuals Dream of Gay Rights? Getting Real About Transgender Inclusion,” in Transgender Rights, ed. Paisley Currah, Richard M. Juang, and Shannon P. Minter (Minneapolis: University of Minnesota Press, 2006), 141–170; Sylvia Rivera, “Queens in Exile, the Forgotten Ones,” in Genderqueer: Voices from Beyond the Sexual Binary, ed. Joan Nestle, Riki Wilchins, and Clare Howell (Los Angeles: Alyson Books, 2002), 67–85; Dean Spade, “Fighting to Win,” in Sycamore, That’s Revolting!, 31–38. [Return to text]
  107. The estimate of nine million LGBT people in the US contained in the WI’s study, “How Many People are Lesbian, Gay, Bisexual and Transgender?,” for example, has been cited so frequently that the WI does not track the number of citations. A Google search for “9 million” and “gay” yields over two million citations. Email between Cathy Renna, communications consultant to the WI, and Alex West, research assistant to Dean Spade, September 12, 2011. [Return to text]
  108. As of August 2011, reports having over 6.5 million user profiles ( According to the internet use research company Alexa, 32 percent of the traffic at originates from the United States ( Assuming that traffic is uniform per user across countries, we take 32 percent of the total 6.5 million profiles to get an estimated 2.08 million profiles by US residents. [Return to text]
  109. Gates, “How Many People…?” [Return to text]
  110. Jonathan Elford, “The Internet and Gay Men,” Social Research Briefs 1 (2003); E. Benotsch, S. Kalichman, and M. Cage, “Men Who Have Met Sex Partners via the Internet: Prevalence, Predictors, and Implications for HIV Prevention,” Archives of Sexual Behavior 31 (2002): 177–183. [Return to text]
  111. At the time of the survey, was a new website used primarily by people in and around Boston. [Return to text]
  112. Note that, in fact, this is an arbitrary decision, so labeling it as “optimistic” is somewhat nonsensical. However, this kind of reasoning is common in WI study estimates. [Return to text]
  113. Even among the general internet population, receives more visitors who are aged 25–54, making more than $60,000/year, African American, White, or Latino, and fewer visitors who are older than 55, making less than $60,000/year, or Asian ( The representation of transgender people on is not clear. This is in addition to sample bias inherent in considering internet users. [Return to text]
  114. Of 48 such studies published as of October 2015, almost all predicted that he state will cumulatively gain money if legal marriage were extended to same-sex couples. [Return to text]
  115. A press release from the WI dated May 28, 2011 suggested another explicit use of the organization’s empirical methods to support an agenda that explicitly endorses capitalism and derides alternatives. The press release was entitled “New Study Shows Vast Majority of Countries Have Become More Accepting of Homosexuality; Trend Slower or Reversed in Russia and Other Ex-Socialist Countries.” [Return to text]
  116. Herman et al., “Impact on Rhode Island’s Budget.” [Return to text]
  117. Randy Albelda, M.V. Lee Badgett, Alyssa Schneebaum, and Gary J. Gates, “Poverty in the Lesbian, Gay, and Bisexual Community,” UCLA: The Williams Institute, 2009. [Return to text]
  118. It is clear that complex market forces based on class background, race, religion, education, and other factors greatly influence dating and marriage partner choice, usually resulting in homogamous couples. “How Your Race Affects the Messages You Get,” OkTrends; Wang and Kao, “Does Higher Socioeconomic Status Increase Contact between Minorities and Whites?”; Wendy Manning and Pamela J. Smock, “First Comes Cohabitation and then Comes Marriage?” Journal of Family Issues 23 (2002): 1065–87; Matthijs Kalmijn, “Intermarriage and Homogamy: Causes, Patterns, Trends,” Annual Review of Sociology 24 (1998): 395–421; Debra L. Blackwell and Daniel T. Lichter, “Mate Selection among Married and Cohabiting Couples,” Journal of Family Issues 21(2000): 275–302. [Return to text]
  119. Gwendolyn Mink, “The Lady and the Tramp: Gender, Race and the Origins of the American Welfare State,” in Women, the State and Welfare, ed. Linda Gordon (Madison: University of Wisconsin Press, 1990), 92–122; Holloway Sparks, “Queens Teens and Model Mothers: Race, Gender and the Discourse of Welfare Reform,” in Race and the Politics of Welfare Reform, ed. Sanford F. Schram, Joe Soss, and Richard Fording (Ann Arbor: University of Michigan Press, 2003), 188–189; Daniel Patrick Moynihan, The Negro Family: The Case for National Action (Washington, D.C.: Office of Policy Planning and Research, US Department of Labor, 1965); Personal Responsibility and Work Opportunity Reconciliation Act of 1996, Pub. L. No. 104–193, 101 (1996); Neubeck and Cazenave, Welfare Racism. [Return to text]
  120. See Robert Rector and Melissa Pardue, “Understanding the President’s Healthy Marriage Initiative,” Heritage Foundation, March 26, 2004; Robert Pear and David D. Kirkpatrick, “Bush Plans $1.5 Billion Drive For Promotion of Marriage,” The New York Times, January 14, 2004; Phoebe G. Silag, “To Have, To Hold, To Receive Public Assistance: TANF and Marriage Promotion Policies,” Journal of Gender, Race, and Justice 7 (2003): 413, 419 (describing West Virginia’s $100 bonus to public assistance recipients who are married); Sarah Olson, “Marriage Promotion, Reproductive Injustice, and the War Against Poor Women of Color,” Dollars & Sense, January–February 2005, 14 (describing marriage promotion programs that provid[e] extra cash bonuses to recipients who get married, [deduct] money from welfare checks when mothers are living with men who are not the fathers of their children, [and] increase[e] monthly welfare checks for married couples”). [Return to text]
  121. James Alm and Leslie A. Whittington, “For Love or Money? The Impact of Income Taxes on Marriage,” Economica 66 (1999): 309–310. [Return to text]
  122. First, it is assumed that weddings in Rhode Island will attract the average of number of out-of-town wedding guests observed in same-sex weddings in Massachusetts. Then it is assumed that half the number of guests will rent hotel rooms at the average rate in Rhode Island. In a different part of the study, it is assumed that wedding spending for same-sex couples is one quarter that of different-sex couples. These assumptions are somewhat arbitrary and raise questions such as: Do same-sex wedding guests stay in luxury hotels and business-rate hotels included in the average hotel figure? Does that estimate inflate the estimated amount spent? Do same-sex wedding guests ever stay with friends or family or share hotel rooms among more than two people? Does that estimate overestimate hotel revenues? How much do same-sex couples spend on weddings? Is a quarter of different-sex couples spending an overestimate? Will the states that recognize same-sex marriage later attract as many weddings as the first few states to do so, or did people flock to the first states to recognize same-sex marriage? These sorts of questionable assumptions are used consistently to set up an estimation concluding that legalizing same-sex marriage will economically benefit the state. [Return to text]
  123. Our Wedding Put Us in Debt!,”, April 27, 2004; “Average Wedding Debt ‘Takes Almost 3 Years to Pay Off’,”, May 3, 2011; Ellie Omahoney, “Wedding Debt Outlives Marriage by Five Years,” Marie Claire UK, July 29, 2009; G.E. Miller, “What Does the Average Wedding Cost,” 20 Something Finance, March 14, 2015. [Return to text]
  124. See, for example: Gates, “Same-Sex Spouses and Unmarried Partners in the American Community Survey, 2008,”; Gates, “Same-Sex Couples in US Census Bureau Data”; Gates, “How Many People…?”; Herman et al., “Impact on Rhode Island’s Budget.” [Return to text]
  125. Gates, “Same-Sex Couples in US Census Bureau Data.” [Return to text]
  126. A finding that is statistically significant is unlikely to occur by random chance, as specified by some threshold (usually 95 percent). A finding that is not statistically significant is likely to be due to random chance, rather than some systematic trend. [Return to text]
  127. Gates, “Same-Sex Couples in US Census Bureau Data,” fig. 7. [Return to text]
  128. Gates, “How Many People…?” [Return to text]
  129. With an estimated 4,007,834 lesbian and bisexual woman, 4,030,946 gay and bisexual men, and 697,529 transgender individuals, the proportion of transgender people in the whole LGBT population is 697,529 /(4,007,834 + 4,030,946) = 0.087 = 8.7% [Return to text]
  130. The WI study reports that “[t]he 2003 California LGBT Tobacco Survey found that 3.2% of LGBT individuals identified as transgender.” [Return to text]
  131. Recall that when we examined the data reported in “Same-Sex Couples in US Census Bureau Data,” we found that African Americans are statistically significantly less likely to identify as “husband or wife” or “unmarried partner” than Whites. (See note 85.) [Return to text]
  132. See Lisa Duggan’s discussion of neoliberalism as being characterized by the “upward distribution of wealth,” and her useful analysis of neoliberal gay rights politics. Duggan, The Twilight of Equality? Neoliberalism, Cultural Politics, and the Attack on Democracy (Boston: Beacon Press, 2004). [Return to text]
  133. It is worth noting that the WI publishes its own studies, eliminating any peer review or editing, which is typically required for academic publishing. Self-publishing frees the WI from the need to substantiate its analyses, but has not prevented its studies from making a significant impact in lesbian and gay rights quests. [Return to text]
  134. Dylan Rodríguez, “Political Logic of the Non-Profit Industrial Complex.” [Return to text]
  135. According to one Williams Institute study, “[s]ame-sex couples raising adopted children are older, more educated, and have more economic resources than other adoptive parents.” Notably, this study makes use of same-sex couples identified using census data, which, as discussed earlier, is likely to over-represent wealthier individuals. Though this study includes a discussion of possible sources of error, we still see invisibilizing of marginalized populations and a disregard for accurate statistical methodology when convenient to the political aims of the study. Gary Gates, M.V. Lee Badgett, Jennifer Ehrle Macomber, and Kate Chambers, “Adoption and Foster Care by Gay and Lesbian Parents in the United States” (2007), 16. For further reading on how child welfare systems target certain populations for family disruption, see, for example: Dorothy Roberts, Shattered Bonds: The Color of Child Welfare (New York: Civitas Books, 2002); Jane Jeong Trenka, Julia Chinyere Oparah, and Sun Yung Shin, eds., Outsiders Within: Writing on Transracial Adoption (Massachusetts: South End Press, 2006), 59–73. [Return to text]
  136. See, for example: Michael D. Steinberger, “Federal Estate Tax Disadvantages for Same-Sex Couples,” UCLA: Williams Institute, 2009; United for a Fair Economy, Estate Tax Campaign; Beverly I. Moran, “Capitalism and the Tax System: A Search for Social Justice,” Southern Methodist University Law Review 61 (2008); Michael A. Livingston, sity LA SeKalven at 50: Progressive Taxation, ‘Globalization,’ and the New Millenium,” Florida Tax Review 4 (2000). [Return to text]
  137. See, for example, Rebecca Stotzer, “Comparison of Hate Crime Rates Across Protected and Unprotected Groups,” UCLA: The Williams Institute, 2007; Morgan Bassichis, “‘It’s War in Here’: A Report on the Treatment of Transgender and Intersex People in New York State Men’s Prisons” (New York: Sylvia Rivera Law Project, 2007); Alexander L. Lee, “Gendered Crime and Punishment: Strategies to Protect Transgender, Gender Variant and Intersex People in America’s Prisons (Parts 1 and 2),” GIC TIP Journal (Gender Identity Center of Colorado: Summer 2004 [part I] and Fall 2004 [part II]); Alex Coolman, Lamar Glover, and Kara Gotsch, “Still in Danger: The Ongoing Threat of Sexual Violence Against Transgender Prisoners” (Los Angeles: Stop Prisoner Rape and the ACLU National Prison Project, 2005); Joey L. Mogul, Andrea J. Ritchie, and Kay Whitlock, Queer (In)Justice: The Criminalization of LGBT People in the United States (Boston: Beacon Press, 2011); Katherine Whitlock, In a Time of Broken Bones: A Call to Dialogue on Hate Violence and the Limitations of Hate Crime Laws (Philadelphia: American Friends Service Committee, 2001). [Return to text]
  138. See Miranda Joseph, Debt to Society: Accounting for Life Under Capitalism (Minneapolis: University of Minnesota Press, 2014), discussing the violence of quantification while also cautioning against the dismissal of statistical knowledge in efforts for social justice (xx). [Return to text]