The Epistemological Problem of White Fragility Theory

One flaw in the theory of white fragility is that it relies crucially on the paradigm of implicit bias. Papers like this one, which provides a reanalysis of an influential study on implicit bias and finds “a pattern of behavior consistent with a pro-Black behavioral bias, rather than the anti-Black bias suggested in the original study,” show that the science is far from conclusive on the validity of implicit bias as a cause of systemic discrimination. Robin DiAngelo, the sociologist responsible for bringing the notion of white fragility into mainstream progressive discourse, may be mistaken in her belief that the implicit biases of white people are a central force in perpetuating institutional racism in America. If so, white fragility survives only as a rhetorical device to invalidate heterodox opinions at a stroke, in the same way critics of capitalism dismiss critiques of communism as bourgeois.

However, DiAngelo’s theory is worse than that. Not only is her central premise of implicit bias dubious, but so is the research methodology that underlies her manifold claims about the implications of white racial illiteracy and white fragility. Di Angelo’s thesis confuses objectivity with neutrality; dismissing objectivity as an ideological obstacle to knowledge (“there is no objective, neutral reality”); relies on anecdotal observations; suffers from a complete absence of rigorous hypothesis testing and quantitative measurement; and blithely ignores the principle of falsifiability, which distinguishes science from pseudoscience.

Finally, her thesis is, as stated in her book, White Fragility: Why It’s So Hard for White People to Talk about Racism, “unapologetically rooted in identity politics” and this is motivated by a subversive campaign that seeks the “decolonization” of modern universities and an interrogation of whiteness as manifested in the “presumed neutrality” of Western Enlightenment epistemology, which, she claims, privileges “particular forms of knowledge over others”—written over oral, history over memory and rationalism over wisdom—to the detriment of anyone who is not white.

In short, DiAngelo relinquishes any concern with objective measurement, despite the fact that she is “quite comfortable generalizing” about white people because “social life is patterned and predictable in measurable ways.”

DiAngelo contends that the “racialized” position of white people in American society rests on social, political, economic and cultural dominance. Ensconced in privilege, white people exhibit white fragility when aroused by any challenge to their privilege. In white fragility, there is the root of implicit bias. To dismantle racism, it is necessary to dismantle the ideological foundations of society itself, which entails a tenacious and ongoing effort to dissect the implicit biases of people who hold the keys to power, i.e. white people. Her work is methodically devoted to discerning and exposing as many examples as she can find of implicit bias, and explaining how these implicit biases underlie and perpetuate the institutional construct of racism.

Di Angelo has extracted her thesis primarily from an exhaustive reading of scholarship in her profession; a selective, one-dimensional, and sometimes egregiously inaccurate reading of history that leads one to surmise that progressive icon Howard Zinn is her favorite historian (in chapter 2, she writes that the US economy at the time of the nation’s formation was based in part on the annexation of Mexican lands, when, in fact, the Texas and New Mexico territories came into the possession of the United States after the Mexican War of the 1840s, and were part of a negotiated purchase following a war that started when Mexico fired the first shot). She also bases her work on her clinical experience with inter-group and intra-group dialogues on racism in formal settings—focus groups, case studies, workshops, seminars and talks—in which she has served as a mediator, facilitator or speaker. In these settings, she has frequently encountered resistance and defensiveness from white people, after confronting them on their racial biases. In her seminal paper, she describes this resistance as white fragility, a condition in which “even a minimum amount of racial stress becomes intolerable, triggering a range of defensive moves.”

DiAngelo posits that white fragility stems from a misunderstanding of what racism is and how it works. Racism, according to DiAngelo, is inextricably tied to the powerful grip that white people have on the levers of institutional control, a grip that will remain ironclad until white people learn to let go of their biases and allow DiAngelo and her ilk to explain to them how everything they say and do is racist, i.e. functions as a scaffold of socialization on which white supremacy survives, against the gravitational pull of social justice activists who seek to bring about its collapse.

DiAngelo’s work offers a synthesis of ideas inherited (implicitly) from Rousseau, Marx, Freud and Foucault (among others), as applied to contemporary scholarship in Whiteness Studies. Her view of how white fragility is integral to institutional racism seems innocuous, even intuitive, when one considers the data showing wide discrepancies between whites and blacks in unemployment rates, earnings and measures such as wealth, poverty, educational attainment and homeownership. Moreover, white males largely hold sway in terms of leadership and influence in politics, Silicon Valley, business, university faculties and the media. When the odds are heavily slanted in favor of white people, and white males in particular, it seems straightforward to suggest that conditioned prejudices, to which we all succumb, affect social outcomes in ways that benefit white people at the expense of people of color. Ending racism is not about eradicating overt prejudice, but about deconstructing the unconscious biases of white people, who hold the cards and thus necessarily retard progress in the eradication of racial inequality.

Clear as the data are on outcomes, however, the hows and whys have long remained a contentious source of debate. While DiAngelo and her ilk may be inclined to attribute this contentiousness to white fragility, even black scholars such as Thomas Sowell and Glenn Loury do not invariably toe the party line on structural definitions of racism. Moreover, the scholarship on racial inequalities is so vast as to make one suspect that a consensus on cause and effect has never been—nor can ever be—reached.

Consider a 2004 Quarterly Journal of Economics paper by Steve Levitt and Roland G. Fryer Jr., two John Bates Clarke medal-winning economists. They document a conspicuous uptrend in parents giving distinctively black names to black children since the Black Power movement of the late 1960s. They also conclude that distinctively black names are strong predictors of socioeconomic status, but do not have a causal impact on adult life outcomes, a result they are able to reconcile with findings from audit studies that “repeatedly have found that resumes with traditional names are substantially more likely to lead to job interviews than are identical resumes with distinctively minority-sounding names.”

DiAngelo might be inclined to argue—not unreasonably given research on resume callbacks—that white people, who presumably constitute the majority of recruiters, implicitly use black names as a signal to identify supposedly unqualified candidates, leading to socioeconomic outcomes that sustain racial inequality and racial segregation. Indeed, in her book, Is Everyone Really Equal: An Introduction to Key Concepts in Social Justice Education, she explicitly argues this point by citing research on resume callbacks and concluding that the human resource workers who screened these resumes “were likely not aware they were discriminating, and would probably have vigorously (and sincerely) denied any suggestion to the contrary.” This may be true, but if the Levitt-Fryer findings are any indication, it does not necessarily follow that implicit bias per se contributes to poor life outcomes for black job candidates who have black names. DiAngelo (and in this case, her co-author), however, simply invokes this research as an example of socialization that explains how systemic disparities remain in place.

Levitt and Fryer scrupulously examine a nuanced dataset, use indexing and plots to illustrate trends in the data, develop equations to quantify relationships between black names and socioeconomic status, employ regression techniques to conduct an empirical test of their model and then interpret the results ex post (as opposed to performing an ex ante imposition of theory on the data) by analyzing the consistency of observed trends and suggesting four alternative models that might explain those trends. They also test hypotheses as to the correlation between distinctively black names and adult life outcomes, explaining the challenges of their estimation techniques and the caveats associated with their conclusions. Finally, they place their paper within an emerging literature on how black culture relates to socioeconomic outcomes, without offering sweeping generalizations about the role of black culture in socioeconomic outcomes. Most impressively, they exhibit restraint in interpreting their results, and use rigorous techniques to analyze a vast dataset. Levitt and Fryer convey results that are insightful and suggestive, but by no means definitive, and indicate that a great deal more work must be done.

DiAngelo, by contrast, is sure that she has found the answers, and is not shy about telling us what she believes has caused these outcomes and keeps them in place. She does call attention to many alleged aspects of white racial illiteracy that are worthy of study, but with an aura of omniscience and ideological rectitude that converts her concerns not into hypotheses to be examined and tested, but into doctrines that must be inculcated.

In one paper, for example, she describes a morning session in an “antiracist” work group as having “gone well” because the white trainer “finished an in-depth presentation on white privilege” and “the group listened attentively and no challenges were raised.” But when a white woman raises objections to a black trainer’s presentation on the impact of racism on people of color and argues that an example the black trainer uses to illustrate her point shortchanges the complexity of the topic, the white trainer intervenes to point out why the white woman is being “racially problematic.” When a black man comes to the defense of the white woman, a black woman in class accuses him of “acting on his internalized racial oppression by ‘rescuing’ the white woman,” and the class session spirals out of control. Thus, when debate erupts, the session becomes “problematic,” a view that stands in stark contrast to the usual presumption in academic seminars that debate is a healthy indicator of interest in a topic.

DiAngelo’s work is indefatigably concerned with instilling principles of antiracist practice in white people (according to a structuralist definition of racism). She explores such themes as how (1) reliance on personal experience in inter-group dialogues on racism limits the capacity of white people to validate the damaging effects of racism on people of color; (2) whites who grew up poor often confuse class privilege with racial privilege; (3) “opinion discourse” in “equity-oriented” classrooms reflects not simply relativism (“that’s just your opinion”) or critical thinking skills that are uncultivated (“a facile response to new ideas”), but “functions as a specific legitimization of existing relations of power”; (4) the values of white parents who identify as antiracist are inconsistent with their practical application of these values; (5) “white silence” in inter-racial discussions, regardless of whether it is a function of personality (e.g. introversion) or humility (e.g. wanting to listen rather than talk) or politeness (e.g. not wanting to offend anyone) often “functions to maintain white power and privilege”; and (6) university hiring processes reproduce whiteness because they do not (among several other alleged failings) “politicize traditional canonic fields”— she recommends that, in job descriptions, candidates should be required to demonstrate an ability to situate knowledge in their field in a social (cultural, historical) context.

In general, DiAngelo argues that societal forces such as segregation, ideological attachments to ideas about individualism and universalism, a deep-rooted sense of entitlement to racial comfort, racial arrogance, racial belonging and racial othering are fundamental causes of racism, and that, collectively, they function organically as an interrelated set of ongoing dynamic norms and practices that underlie a society characterized by racism, i.e. the internalization of supremacy among whites and the internalization of coercion among people of color.

These are extraordinary claims: not simply because they are so sensational, but because they are so all-encompassing. They are rooted in a comprehensive theoretical realignment of how we are supposed to understand not only the nature of racism and its connection to social outcomes, but the entire moral, economic, political and cultural ethos of American society. Instead of being a constitutional republic founded intellectually on majoritarian constraints against the dangers of factionalism, but with adequate protections for natural rights (albeit imperfectly applied), America was founded, DiAngelo argues, on nothing more than economically and politically exploitative practices of “slavery, colonization, and genocide” (even though debates among the founders reveal complex divisions between slavery and antislavery advocates from which emerged a constitution that provided no explicit sanction to “property in man,” i.e. slavery, as Sean Wilentz has shown.) As a result, all white people have been born into an inherently racist society, in which all socioeconomic disparities are directly related to institutional racism.

But if we are to take these claims seriously, then natural questions to ask include: what is her methodology? Which construct validity models guide her investigation? How does she state and present her hypotheses? What tests does she run? What data does she collect? What criteria does she use to evaluate and interpret her results?

In the social sciences, the answer to these questions often involves the use of data and statistical analysis. Statistics is crucial because it provides techniques suited to the rigorous examination of data on a scale applicable to questions about systematic trends and patterns in society. Quantitative data analysis is fraught with perils, but disciplined quantitative data analysis is usually unavoidable when trying to measure the accuracy and precision of general claims about systematic patterns in the social, economic, political and cultural dimensions of society. Data do not give us answers. Data give us something to investigate, and there are few more organized ways of parsing reams of data in a way that facilitates measurable inferences, which are sufficiently transparent for critical evaluation—although we must still be cautious about making generalizations. Statistics provides tools to develop quantitative measures of trends and patterns in social life.

This is critical. As Levitt and Fryer note, “[a] primary obstacle to the study of culture has been the lack of quantitative measures.” Which is why it is astonishing to discover that DiAngelo makes sweeping generalizations about cultural and other dimensions of society without ever making use of statistical analysis. She has published three books and several papers in peer-reviewed journals. She has participated in multiple talks (see here, here and here). She has written articles for Medium and The Good Men Project. Nonetheless, in all her work, one finds a complete absence of statistical analysis that aims to quantitatively measure how white fragility contributes to institutional racism.

In fairness, in defining racism in chapter 7 of her book What Does It Mean to Be White? as a “form of oppression in which one racial group dominates others,” she does cite statistics on racial discrepancies in health care, wealth and representation in the criminal justice system; and, in chapter 2 of her book White Fragility, she cites data on the demographic composition of Congress, governors, college professors and teachers, in her attempt to define racism as a social construct. But nowhere in her work does one find a study in which she employs statistical techniques to quantitatively examine a well-defined hypothesis as to the causal relationship between disparities in outcomes and a specific measure of implicit bias, which can be tested for accuracy, precision and statistical significance.

Simply put, DiAngelo infers systemic oppression from data on racial disparities. As she states in What Does It Mean to Be White?, “In the United States the dominant group is white, therefore racism is white racial and cultural prejudice and discrimination, supported intentionally or unintentionally by institutional power and authority, and used to the advantage of whites and the disadvantage of people of color.” But saying there is a link is not the same as proving it. For one who insists on sustained, rigorous, statistical hypothesis testing, her structuralist leap of logic is not obvious. It is a hypothesis, not a foregone conclusion. Unfortunately, one finds not one example in her work of rigorous, statistical hypothesis testing that stands the test of time.

The absence of statistics in DiAngelo’s work is also astonishing considering how she handles an objection she frequently encounters in interactions with white people: that she is prone to generalization. She argues that this objection stems from the pervasive ideology of individualism in American culture. This belief, she argues, presents an obstacle to understanding racism because it cultivates the erroneous perception that the contingencies and vicissitudes of experience for each individual are sufficiently unique to make one immune to racialized conditioning. Given their belief in the singularity of experience, white people refuse to accept the proposition that they can be pigeon-holed as racists, who reinforce white supremacy by virtue of their complacency about a status quo characterized by racial privilege.

Although she relegates this objection to ideology, and explains it away as another manifestation of white fragility, she nonetheless agrees in her book, White Fragility: “As a sociologist, I am quite comfortable generalizing; social life is patterned and predictable in measurable ways.” There is, of course, a jarring insouciance in this claim, given that social scientists are typically wary of generalizations. Social life has observable patterns, which can be detected using techniques conducive to measurable predictions. But these patterns are often exceedingly hard to measure, and, given the organic and dynamic nature of social forces, can rarely be predicted with a high degree of accuracy. Measurement is invariably a delicate matter in scientific research. It gives rise to much disagreement as to how questions should be asked and addressed and how results are interpreted. But if one is inclined to generalize about social patterns, especially when dealing with issues as broad, complex and controversial as those DiAngelo identifies in her work, the effort not infrequently requires an extraordinary effort at data collection and analysis.

Statistics is the science of data collection and analysis. Given the scale and scope of DiAngelo’s claims about how white fragility sustains racism as a social construct, her methodological approach should include statistical techniques. She should collect and analyze data on a scale and scope equal, or at least proportional, to the scale and scope of her claims. But she does not.

How, then, does she know what she claims to know?

As someone who has spent decades in the field of multicultural education, and earnestly admits that she is a racist because she is white, DiAngelo incessantly calls for humility in our thinking about race and racism, and does not excuse herself from the epistemic impotence that all white people have on matters of race and racism. As a corollary, she also claims in many of her talks (see at 14:15 here; see also at 11:30 here) that if you haven’t spent many years studying and struggling with the problems of race and racism, then your opinions are necessarily very limited and uninformed. In her book White Fragility, she reiterates the same point: “[w]e must be willing to consider that unless we have devoted intentional and ongoing study, our opinions (on racism) are necessarily uninformed, even ignorant.”

Aside from being tautological, and failing to consider the possibility that one can come to the wrong conclusions even after devoting many years of one’s life to intentional and ongoing study, the statement is fair. For example, one does not expect to be hired to teach economics at a university without having earned a PhD in economics, or at least having gained professional experience in a field that involves extensive thinking about how markets work. But Di Angelo’s statement also draws attention to a fundamental concern about her epistemology.

Epistemology is a branch of philosophy that deals with claims of knowledge. How do we know what we know? This question has preoccupied philosophers for centuries. According to Socratic dictum, wisdom consists of the recognition of how much one does not know. Thus, if there is an epistemological starting point for social science research, it is an appeal for caution about the inferences one can draw about whole populations. In short, one should be wary of generalization.

That does not mean we should consider ourselves impossibly handicapped by the depth, breadth and complexity of a topic. But it does mean that methodological rigor, coupled with caution, is imperative.

Social science methodology can be divided into qualitative and quantitative techniques. On the quantitative side, statistics is widely used to measure central tendency and dispersion in a sample population. Given samples of data for variables such as income, demographics, educational attainment and homeownership, statistical analysis also involves the formulation of hypotheses about relationships between variables, and provides a set of tools to rigorously measure the relationship between a dependent variable and an independent variable, or set of independent variables. Finally, statistics provides a set of criteria by which to test the significance of these relationships.

For example, one could develop a model that attempts to explain white denial of implicit bias as a function of white belief in individualism. This would involve a careful effort to define and identify a measure of implicit bias, which, if the inconclusive state of research on microaggressions and the Implicit Association Test is any indication, would not be easy. Then, if one focuses on a simple one-variable logistic regression model, one could construct an indicator variable that controls for whether white individuals do or do not believe in individualism, which would require an equally careful effort to define individualism, perhaps by administering a survey, using a double-blind experimental design, with appropriately worded questions about beliefs in individualism posed to a sufficiently large and randomly selected sample of white people. The data could then be used to test the regression model that formalizes the hypothesis about whether a belief in individualism predicts a denial of implicit bias, which potentially can then be developed into another model that examines how implicit bias explains the movement and variation of some measure of racial inequality. Of course, this simplistic scenario ignores potential problems that might arise from the omission of additional relevant variables and linear relationships between the omitted variables, among other potential statistical complications (e.g. heteroskedasticity if a simple linear model were indeed developed to estimate the effect of implicit bias on a measure of racial inequality).

In short, statistics provides a framework for data analysis that allows a scientist to go beyond the realm of personal intuition derived from qualitative distillations of anecdotal observations. This is of utmost importance when studying the prevalence of racism and fragility in the entire white population of the United States. Regardless, without statistics, DiAngelo claims to have observed consistent, frequent and general patterns in her work. She draws on these patterns to develop a litany of examples, which allegedly demonstrate the racialized experience of white people, its relevance to racial inequality, and how it instills, influences and sustains the implicit biases that underlie white complacency about the institutional construct of racism (i.e. white supremacy).

But, in failing to make use of quantitative techniques that are customized for making inferences about populations (i.e. generalizing), she falls short of her claim to intellectual humility by failing to formulate her claims as testable, falsifiable hypotheses (something statistics allows one to do). When Karl Popper identified the concept of falsifiability as a foundational pillar of scientific inquiry, he was demarcating science from pseudoscience: “The dominance of the critical spirit in Einstein, and its total absence in Marx, Freud and Adler, struck Popper as being of fundamental importance: the pioneers of psychoanalysis, he came to think, couched their theories in terms which made them amenable only to confirmation, while Einstein’s theory, crucially, had testable implications which, if false, would have falsified the theory itself.”

Similarly, the theory of white fragility, and its practical application in DiAngelo’s work, are couched entirely in terms amenable to its confirmation rather than its falsifiability. This is reflected most starkly in the discovery that there is not one example in her research of rigorous hypothesis testing using statistical techniques. Her claims are based exclusively on qualitative assessments of her observations of white people interacting with people of color in group settings. Objections are invariably explained away as manifestations of white fragility. In the same way that critics of capitalism dismiss critiques of communism as bourgeois, she treats her results as non-falsifiable.

It is not as if qualitative research is necessarily always inferior, or as if there is something categorically wrong with DiAngelo relying on intuition drawn from her years of experience in multicultural education. Moreover, a host of qualitative research techniques are regularly employed in social science research. But it is hard to take seriously any research that consistently and frequently makes comprehensive claims about patterns in social life, without ever once having made competent use of hypothesis testing to conduct an analysis of data empirically relevant to a claim about patterns in social life.

One main concern is that qualitative research is limited to small sample sizes. Results are parochial, necessarily and unavoidably related to the context in which the study is undertaken, while interpretation of results is often unduly influenced by the theoretic and thematic discourse that the researcher brings to the table (i.e. bias associated with sociopolitical theories and values espoused by researchers). While researchers can glean useful insights from the variety of perspectives offered by participants in focus groups, workshops and intra-group dialogues, it is rare if not impossible for researchers to draw inferences about systemic patterns in society from small sample studies that are inherently susceptible to sampling bias, self-selection bias and the Hawthorne effect. Brainstorming is one thing. Systematic data collection and hypothesis testing is another.

DiAngelo’s critical shortcoming is that she does not formulate her claims into hypotheses that can be tested using statistical techniques. DiAngelo’s reliance upon qualitative research to deduce systematic patterns of white fragility in American life is akin to Freud diagnosing psychological conditions as traceable to Oedipal anxieties in the unconscious mind. Like Freud deriving psychoanalytic insights into the unconscious mind from the faulty premise of the Oedipus complex, DiAngelo derives psychological insights into white fragility from the questionable premise of implicit bias. Similarly, just as Freud derived insights from clinical observations of patients, DiAngelo derives insights about white fragility from her observations as a mediator in group discussions of racism.

In failing to make use of statistics, DiAngelo’s theory of white fragility is presented as non-falsifiable. As such, it cannot be distinguished from pseudoscience. DiAngelo offers us no way of evaluating whether she commits a type 1 or type 2 error when she makes one of her innumerable assertions about how the reactions of white people, when confronted with their alleged racialization and privilege, exhibit white fragility.

A type 1 error is a false positive, i.e. claiming a result (e.g. a white person who draws on personal experience in inter-group dialogues on racism invalidates the damaging effects of racism on people of color) is statistically significant when, in fact, the null hypothesis is true (e.g. a white person who draws on personal experience in inter-group dialogues on racism does not invalidate the damaging effects of racism on people of color). A type 2 error is a false negative, i.e. claiming a result is not statistically significant when, in fact, it is (i.e. the null hypothesis is false).

In one paper, DiAngelo states: “Human objectivity is not actually possible, but as long as we construct the world as if it is, and then ascribe it only to ourselves, we keep White experience and people centered and people of color in the margins.” Unfortunately, DiAngelo confuses objectivity with neutrality. Indeed, in another paper, she argues that modern universities are obsequiously reliant on the orthodoxy of “White European enlightenment epistemology,” which confers privilege on written knowledge over oral knowledge, history over memory and rationalism over wisdom. She then asserts incoherently that the validation and elevation of “positivistic White Eurocentric knowledge” over “non-White, Indigenous and non-European knowledges” is a normative construct in which universities seek knowledge not for knowledge’s sake, but for the sake of (white European colonial) power, which thus requires the “decolonization” of the academy. Knowledge, even “positivistic White Eurocentric knowledge,” is never neutral and always normative.

Regardless of what one thinks about the possibility of neutrality, objectivity is something different. To insist on objectivity is to insist on the gathering of facts, data and other evidence in pursuit of an argument, which can be tested in accordance with rigorous standards of measurement. If objectivity is impossible, so is measurement. You can’t measure social outcomes in a way that achieves consensus (except, I suppose, by coercion). Truth becomes relative. Battles over ideas become battles over power.

DiAngelo might agree that truth comes down to power, but, if so, her theory of white fragility collapses. As she writes in her seminal paper on white fragility, “Whiteness Studies begin with the premise that racism and white privilege exist in both traditional and modern forms, and rather than work to prove its existence, work to reveal it.” With that as her starting point, she reveals her central concern with revelation over proof, and thus begs the question. But she nonetheless goes about collecting innumerable anecdotes that supposedly reveal privilege and fragility. This comports with her view on the impossibility of objectivity, but it also means that she does not measure social patterns in a way that can achieve consensus. Given the complete absence of rigorous hypothesis testing in her work, which would allow one to objectively assess the validity of her manifold claims about the implications of white racial illiteracy, her entire theory of white fragility rests on a methodological house of sand.