The Egalitarian Fallacy: Are Group Differences Compatible with Political Liberalism?

Many people greet evidence of biologically based race and sex differences with extreme skepticism, even hostility. We argue that some of the vehemence with which many intellectuals in the West resist claims about group differences is rooted in the tacit assumption that accepting evidence for group differences in socially valued traits would undermine our reasons to treat people with respect. We call this the egalitarian fallacy. We first explain the fallacy and then give evidence that self-described liberals in the United States are especially likely to commit it when they reason about topics like race and sex. We then argue that people should not be as worried as they often are about research that finds psychological differences between men and women, or between people of different racial or ethnic groups. We conclude that if moral equality is believed to rest on biological identity, ethnically diverse societies are in trouble.


The Taboo of Group Differences
We will focus on negative reactions to scientific findings concerning group differences, especially when these reactions are motivated by ideological commitments. The egalitarian fallacy involves a mental shortcut, or "heuristic," that is so ingrained in modern Westerners that many of us don't think about it any more than we think about the grammatical structure of sentences when we utter them. Like grammatical errors, we recognize when someone has violated the taboo against making claims about group differences, but unlike grammatical errors, we often respond to violations of the group differences taboo with hostility and denial rather than mere skepticism.
As mentioned above, the egalitarian fallacy consists in rejecting claims about group differences because of their perceived moral implications. People who commit the kind of fallacy we are describing seem tacitly to endorse the view that different populations do not differ biologically on socially valued traits, because such differences would be cosmically unjust (we have therefore called this view "cosmic egalitarianism"). Examples of intellectuals who commit the egalitarian fallacy are so numerous that we could fill an encyclopedia. But we wish to focus on a few recent cases, and then review some experimental evidence that strongly suggests these cases are not unusual.
Before developing these examples, it is worth noting two things. First, some scholars have advanced biased theories to support invidious norms and policies. Many of the elites in the antebellum South, for example, were attracted to theories that Blacks were incapable of self-governance and were "designed" to be manual laborers because those theories (supposedly) provided a justification for slavery. Second, such bigoted theories and policies likely linger in the imaginations of modern liberals, who fear that contemporary scholars are behaving just as deplorably as those who used armchair theories about racial minorities or women to justify treating them as social subordinates.
Whatever its origins, in this paper we simply want to show that the egalitarian fallacy is widespread, that it is especially prominent among those who fall on the political left, and that the political implications of group differences may be less scary than many people believe. Jason Richwine's 2009 doctoral dissertation at Harvard University argued that US immigration policy should be designed to favor people with high skills and intelligence (measured by IQ). Since IQ is correlated with important life outcomes ranging from educational attainment and earned income to family stability and incarceration (Beaver et al. 2013;Haier 2016;Jones 2016;Plomin 2018), this is a defensible position, even if there are plenty of arguments on the other side (Heine 2017). But a few years later, when Richwine was working for a conservative think tank, journalists obtained a copy of his dissertation and began to insult his integrity and question his motives. In his dissertation, Richwine argued that Hispanic immigrants 1 who had entered the country to work in low-skilled occupations had a lower average IQ than some previous immigrant groups, and that the IQ gap did not narrow much over time. He did not argue for a race-based immigration policy, but he did argue in favor of selecting members of all groups who tended to have a higher IQ (which might lead to more immigration from some groups than others).
Richwine's opponents did not attempt to challenge his data, but instead denounced him as a bad person. The media firestorm that ensued from the discovery of his dissertation was fierce (Richwine 2013). After repeated accusations of racism by journalists, Richwine's employer, a conservative think tank called The Heritage Foundation, fired him.
Writing for The Economist, Will Wilkinson praised Heritage for firing Richwine. Wilkinson justified his view with the argument that: If Mr. Richwine's view "turns out to be correct", what we are to do is to acknowledge that the racists were right all alongthat racism has, to some extent, a valid scientific basis (Wilkinson 2013).
The argument exemplifies the egalitarian fallacy to the extent that it moves from an empirical conjecture about biological differences to a moral conclusion that such differences justify treating some racial groups better than others. Yet the idea that if there are average differences between groups we are justified in elevating all members of one group, and subordinating all members of other groups, is an implausible moral position that we will criticize below.
Another response to Richwine came from a writer for Scientific American, a mainstream popular science magazine. Writing about race and IQ, John Horgan erroneously equated the finding that there are average differences between groups with the view that some groups exhibit "genetic inferiority" relative to other groups. There are at least two things this could mean. On one reading, to say a group is "genetically inferior" would mean that, on average, they score lower than other groups on some genetically mediated trait. Someone might say, on this interpretation, that East Asians are "genetically inferior" to West Africans because they are shorter on average, and height is heavily influenced by genes.
But this is not a natural way of talking. People sometimes say things like "Kasparov is a superior chess player," by which they mean that he's better than most players. However, it would be arresting to hear someone say Kasparov is genetically superior to other people, or that Russian Jews are genetically superior to other groups simply because Russian Jews tend to excel in chess.
The second reading of Horgan's phrase, and a more natural interpretation of it, is the kind of thing a demagogue would say: that group A is, all things considered, simply better than group B. But this is not a plausible interpretation of what Richwine was saying, and it is not a plausible implication of his view. To say that genetically mediated traits are differentially distributed across human populations does not imply that one group is superior to another, all things considered. On the second interpretation of Horgan's statement, which is often invoked by critics of research into group differences, "inferiority" is a moral judgment about the relative worth of a group. Historically, it might even be taken to imply that some groups should govern others, since they arein some sensebetter than others. Yet mainstream proponents of the view that humans exhibit biodiversitypeople like Charles Murray, Jason Richwine and Nicholas Wadeare extraordinarily careful to separate scientific findings from moral judgments. As Nicholas Wade put the point in a recent book on racial differences, "the notion that any race has the right to dominate others or is superior in any absolute sense can be firmly rejected as a matter of principle and, being rooted in [moral] principle, is unassailable by science" (2014, p. 8).
In response to the Richwine affair, Horgan proposed either that states ban research into race differences, or that Institutional Review Boards (IRBs) reject proposals to do such research. According to Horgan, "IRBs…should reject proposed research that will promote racial theories of intelligence, because the harm of such research-which fosters racism even if not motivated by racism-far outweighs any alleged benefits" (2013).
Horgan doesn't give any evidence of doing a serious cost-benefit analysis, such that the harm of revealing genetic differences will outweigh any benefits associated with it. He just feels as though certain scientific conclusions will produce more harm than good, and that research bearing on these conclusions should be either legally banned or socially banished. It is clear that Horgan is committing the egalitarian fallacy, and he is especially worried about groups perceived to be oppressed. This is suggested by the fact that Horgan follows up his argument against research on race differences by directing his readers to studies on stereotype threatstudies that have not stood up to replication attempts (Jussim 2015) which attempt to explain achievement gaps by appealing to the terror that some racial groups allegedly feel about potentially confirming invidious stereotypes. In other words, Horgan spurns research that suggests a partly biological cause for poor performance among groups of people who aren't especially well off, but uncritically accepts the (unlikely) conclusions of research that shows the opposite.
It is also worth noting that many people who assail the scientists who study or write about potential group differences misrepresent the literature. Consider much of the reaction to Herrnstein and Murray's The Bell Curve (Herrnstein and Murray 1994). Critics grossly mischaracterized it as a hateful, racist manifesto that contended that all group differences are genetically determined and that any social interventions to help the least well off are doomed to failure (Winegard and Winegard 2017). In one particularly egregious but representative review, Bob Herbet, of the New York Times, wrote, "Murray can protest all he wants, his book is just a genteel way of calling somebody a n**ger." (Hebert 1994) Stephen J. Gould, quite explicitly committed the egalitarian fallacy, contending that the message of The Bell Curve must be resisted or it would "cut off all possibility of proper nurturance for everybody's intelligence" (1994).
The truth is very different. Herrnstein and Murray (like many others who grapple with these issues) were judicious, asserting: "It seems highly likely to us that both genes and environment have something to do with racial differences. What might that mix be? We are resolutely agnostic on that issue; as far as we can determine, the evidence does not yet justify an estimate." (p. 311). It is true that they were largely pessimistic about current policy proposals, but the book was hardly the grotesque tome of racial animus its critics depicted. And nothing in the book suggested that they believed that blacks should be treated differently from whites or that some children, simply because of their race, should be ignored or shuffled into pedestrian school systems that don't "nurture" intelligence. Furthermore, those who study such topics do not assert that genes ineluctably lead to traits that produce socioeconomic disparities. Rather, they cautiously note that some unknown portion of group disparities in traits and social outcomes are likely caused by genetic differences.
We do not believe that most of those who exaggerate the claims in books like The Bell Curve or A Troublesome Inheritance are acting in bad faith. Rather, they are truly concerned about the nature of the material and misread it because they believe that it threatens vulnerable groups. But their exaggerations actually make the material more dangerous and distort the message that some prominent geneticists believe: Genes powerfully influence our psychological traits (Plomin and von Stumm 2018), including those that may help explain some group differences (Reich 2018).
Horgan's response to the Richwine affair arose almost two decades after Jared Diamond wrote a book in which he claimed geography rather than genes explains why some groups have created more prosperous societies than others. Remarkably, along the way, Diamond claims that he considers it likely that New Guineans (who live a tribal lifestyle) are smarter than Europeans. According to Diamond, "modern 'Stone Age' peoples are on the average probably more intelligent, not less intelligent, than industrialized peoples.…From the very beginning of my work with New Guineans, they impressed me as being on the average more intelligent…than the average European or American is" (1999, p. 19-20).
The critical success of Diamond's book seems to have come in part from his attempt to show that different environments alone can explain achievement gaps. But he can't help but throw in an offhand and unsupported assertion that if anything, the group seen as powerless is probably more intelligent on average than the group seen as powerful. Interestingly, Europeans had little contact with people in New Guinea until very recently, so there is no history of mistreatment. But Diamond's statement perfectly matches the modern tendency of those who call themselves "progressive" or "liberal" (in the American context) to divide the world into oppressor and oppressed, and to explain all achievement gaps as the product of oppression (Haidt 2012). More importantly, the fact that Diamond's readers rewarded him with praise rather than scorn (despite his comment about New Guineans being smarter than Europeans) strongly suggests that they didn't notice the egalitarian fallacy had been violatedpresumably because it was violated in the direction of a relatively powerless group rather than a powerful one.
People who have made claims about average sex differences in recent years have also been fired from jobs and harassed online. A prominent recent case is James Damore, a software engineer fired from Google in the summer of 2017. Damore was a young coder who attended the company's diversity training program, and was asked for feedback on Google's diversity policy. In response, he crafted a lengthy and wellsourced critique of the assumption implicit in Google's policy that men and women are psychologically identical (Damore 2017). Damore said that although he supported modest efforts at recruiting more women into software and engineering, he doubted Google would attain an equal number of men and women without seriously compromising hiring standards. He also made the case that it was Google's culture of intolerance toward people who they disagree with politically that prevented executives from seeing why he might be right. Much of the science was vindicated by researchers on sex differences (Jussim et al. 2017). Even Cordelia Fine, arguably the most influential skeptic of scientific research into sex differences, argued that Damore's summary of the evidence on sex differences was "more accurate and nuanced than what you sometimes find in the popular literature" (quoted in Lewis 2017).
The Damore case was instructive. First, the response by journalists was brutal. Many accused Damore of sexism, others called for Damore to be fired, and Google acceded to public pressure by firing him just a few days after his internal memo was leaked. Second, many commentators erroneously claimed that Damore said women were "inferior" to men or that women were bad programmers. This was their translation of his actual claim that small differences in the mean aptitudes and interests of men and women would translate into unequal sex ratios in different occupations. The reflexive reaction of anger and the unconscious translation of Damore's scientific generalizations into moral termsmale/female differences imply that one group is "superior" to or "better than" the other groupstrongly suggests the egalitarian fallacy was at work. And since it was applied selectivelymany of Damore's fiercest critics do not call for even numbers of men and women in female-dominated professions like pediatric medicine or social workit is plausible to describe many of Damore's critics as committing the egalitarian fallacy.
Six months after Damore was fired, the National Labor Relations Board of the United States ruled that Damore's memo was sexist and harmful, and thus that Damore may have violated anti-discrimination laws by summarizing what he believed to be the best available science. Here is what the NLRB said (Verbruggen 2017): Statements about immutable traits linked to sex-such as women's heightened neuroticism and men's prevalence at the top of the IQ distribution-were discriminatory and constituted sexual harassment, notwithstanding effort to cloak comments with "scientific" references and analysis, and notwithstanding "not all women" disclaimers. Moreover, those statements were likely to cause serious dissension and disruption in the workplace. Indeed, the memorandum did cause extreme discord, which the Charging Party exacerbated by deliberately expanding its audience. Numerous employees complained to the Employer that the memorandum was discriminatory against women, deeply offensive, and made them feel unsafe at work. Moreover, the Charging Party reasonably should have known that the memorandum would likely be disseminated further, even beyond the workplace. Once the memorandum was shared publicly, at least two female engineering candidates withdrew from consideration and explicitly named the memo as their reason for doing so. Thus, while much of the Charging Party's memorandum was likely protected, the statements regarding biological differences between the sexes were so harmful, discriminatory, and disruptive as to be unprotected.
Notice the leap here from scientific claims about biological differences (whether or not they are true) to the moral conclusion that these claims are illegitimate (even if they are true) because they make people feel "unsafe at work." The word "safety" on college campuses and in the workplace in the United States is important because laws crafted in the 1960s to protect racial minorities from violence, and women from sexual harassment, specify that everyone has a right to an "equal and safe" environment at work or in school. But the meaning of "safety" has since morphed from a right not to be physically assaulted to a right not to feel threatened by ideas you disagree with.
The response to Damore's memo by many journalists, and by the NLRB of the US Government, strongly suggests that many influential commentators committed the egalitarian fallacy. We do not want to multiply cases, but it is worth noting that similar episodes occurred when Harvard President Larry Summers was pressured to resign in 2006 for saying small sex difference might lead to differential success in different fields, and when the Nobel Laureate, James Watson, was fired in 2007 from a prestigious research lab for saying that Africans have, on average, a lower IQ than some other groups, and that Western aid to Africa would probably not close all achievement gaps.
In our experience, those who are most apt to reject evidence in favor of group differences often argue in favor of the (logically separate) conclusion that the relevant groups don't actually existthat there can be no differences between men and women, or Asians and Africans, because these categories are illegitimate (e.g. Haslanger 2000; Kitcher 2007). Yet just because categories like race and sex (or tables and chairs, cars and trucks, cities and countries) are partially socially constructed, it does not follow that there are no good reasons for categorizing the world in this way, especially if these categories have some explanatory and predictive power (Sesardic 2010).
We now want to turn to experimental evidence that indicates the egalitarian fallacy is selectively applied by self-described liberals (or "progressives") in the United States. We then spell out why the existence of group differences may not be as worrisome as some people fear. In fact, we think, hostility toward people who conduct research into group differences, or toward people who summarize that research publicly, may be producing more harm than the research itself (Anomaly 2017;Carl 2018;Cofnas 2020). 2

Testing the Egalitarian Fallacy
We have recently tested these observations ). First, we devised a measure of what we call "equalitarianism," which consisted of 18 propositions (e.g., "Racism is everywhere, even though people say they are not racist," and "People often use biology to justify unjust policies that create inequalities"). Then we examined responses to a variety of questions, predicting that self-described liberals (those who indicated they were above the halfway point on a scale that went from 1 = very conservative to 7 = very liberal) would believe that "victim's groups" (e.g. Blacks, Women, Muslims) are treated more unfairly than "privileged groups," and that such beliefs could be gauged by scores on the equalitarian measure. We found large effect sizes. For example, the correlation coefficient was .44 for Blacks, .39 for women, and .39 for Muslims.
More informatively, we documented a consistent pattern of liberal bias against test results that seemed to suggest that victim's groups perform worse than privileged groups. Bias is notoriously difficult to measure, but fairly easy to understand. One way to define bias is that it occurs when someone changes their response when extraneous information is introduced (Kahan 2016). Below, we briefly review two examples of liberal bias in how information about group differences is interpreted.
In the first case, we gave participants a vignette about a gene variant that was discovered by a researcher that was said to explain variation in intelligence. It also said that the researcher believed the gene might explain intelligence differences between Blacks and Whites. In one condition, the vignette said Blacks outperform Whites on IQ tests, and in the other, it said that Whites outperform Blacks. Then we asked participants several questions about the vignette, including how plausible the researcher's arguments were, which were combined into one measure called "argument credibility." Liberals rated the results as much less credible when Whites were said to outperform Blacks than when Blacks were said to outperform Whites.
In the second case, we gave participants a vignette about a new college entrance exam. The vignette said the exam had strong predictive validity but that either men or women did better on it (men perform better = privileged group outperforms victim group). As predicted, liberals rated the vignette in which men did better as more sexist and less fair than the vignette in which women did better.
These results are consistent with the cosmic egalitarian thesis. When confronted with evidence for group differences, liberals tend to consider the evidence less credible when it suggests that groups perceived as more powerful score better on socially valued traits than groups perceived as less powerful. As we have argued, this seems to be because they believe that if such differences exist, then it would be morally justifiable to treat some groups badly.

Moral and Political Implications
We have identified a psychological bias that often results in a strong visceral rejection of claims about group differences. We now want to discuss why we think the moral and political consequences of discovering group differences may not be as scary as many seem to think.
Peter Singer has argued that empirical findings about genetically based group differences shouldn't be taken to undermine the principle that people are moral equals (2011, p. 17). On Singer's view, all sentient creatures have interests that are equally worth protecting (2011, p. 20). What these interests are depends on what kind of creature we are (pigs have different interests than people) but also what properties each of us has (people with severe mental impairments don't have the same interests as intellectuals, though they do have an interest in avoiding pain and frustration).
Singer goes on to argue that, aside from average differences between groups, if we could reliably rank individuals based on intelligence "enslaving those who score below a certain line on an intelligence test would not… be compatible with equal consideration. Intelligence has nothing to do with many important interests people have" (2011, p. 21). We think Singer is too quick to dismiss the importance of intelligence in formulating and achieving our goals, which is a central interest in human life. But we agree that when we think carefully about the moral basis of rights, we are not committed to the idea that every individual must possess an equal amount of whatever qualities give us moral rights. We just need enough.
Moreover, we think this is a characteristic feature of political liberalism in its classical formulation, even if self-described liberals in modern America might reject this view. For example, the right to control our body, or the corresponding obligation others have to ask permission before attempting to get us to do things for them, is not a scalar concept, such that for every additional IQ point or ounce of empathy I have, others have correspondingly higher moral obligations to accede to my demands. Instead, on our view, moral standing is a threshold concept, and most people meet the threshold to be given equal moral rights. 3 Of course, moral standing may have multiple thresholds, so that most people have greater moral standing than most chimpanzees, and chimpanzees have greater standing than butterflies. But between the thresholds, and focusing on people (rather than animals), most moral theoriesincluding classical liberalism of the kind espoused by John Locke or Thomas Jeffersonare not committed to the idea that every individual within a group, or the average score between groups, must be the same for each person to have equal basic rights like the right to free speech, bodily autonomy, or religious liberty.
Nevertheless, it is worth distinguishing moral standing, which is a threshold concept, from moral status, which is a scalar concept. According to Allen Buchanan, "a being has moral standing if it counts morally, in its own right… Moral status, in contrast, is a comparative notion. Two beings can both have moral standing, but one may be of a higher moral status" (2009, p. 346). Buchanan argues that all creatures with sufficient moral status have equal moral standing, or equal moral rights, even if different individuals may have different moral status.
We do not want to take a stand on the precise combination of traits that bestow some creatures with moral standing, or different amounts of moral status. Usually sentience is taken to be sufficient for having minimal moral standing, because creatures capable of suffering have an interest in not suffering. And rational capacities are often taken to be necessary for having the level of moral standing we accord to people, perhaps because rationality is required for crafting and carrying out complex plans. Nevertheless, differences between people in traits like empathy and intelligence may confer varying degrees of moral status.
If this is true, we might worry that differences between individuals in morally relevant traits can be so large that different people have different moral standing. For example, Albert Einstein or Eleanor Roosevelt may be thought to be more valuable and to have different interests than someone with a mental handicap that prevents them from solving problems, forming plans, or having deep emotional relationships. This wouldn't imply that the mentally handicapped person has no moral standing, but it may justify treating them quite differently than we treat Roosevelt or Einstein.
Moving from individuals to groups, we might worry that if average group differences were such that all members of group X had more of (morally important) trait T than all members of group Y, then Xs would be morally superior to Ys in the way chimpanzees have greater moral standing than mosquitoes. This doesn't follow though. If all women had more compassion than all men, but men had enough compassion to respond to moral claims appropriately, this wouldn't automatically mean women have greater moral standing than men. But if the differences were big enough, and ranged across enough morally relevant traits, at some point a threshold might be crossed so that we could declare Xs morally superior to Ys not only with respect to status, but also with respect to standing.
The best evidence is that average biological differences between different racial groups, or between men and women, are such that for any given trait like intelligence or empathy, it is not the case that all members of one group score higher on that trait (or the total set of morally relevant traits) than all members of another group (Crow 2002). This suggests that, in thinking about race and sex, no one group has superior moral standing to another, even if some members of any particular group may be considered to have greater moral status than some members of another group.
Turning from moral standing and moral status to political rights, we might worry that if there are average group differences in socially valued traits, this will license assigning different legal rights to different groups. However, the principle of equality outlined above should commit us to assigning ordinary people similar legal protections, even if statistical discrimination can be perfectly rational.
Why might statistical discrimination be rational? If information is costly to gather, and if average group differences are large enough to generate reliable predictions, it might be perfectly rational to make generalizations about members of groups. For example, if we are choosing teammates from different populations (including, say, people of a different race, sex, or age), and we have no other information available, we will rationally use past experience with different groups to guide current judgments. And we will often be right. Moreover, we will often be morally justified in engaging in statistical discrimination, or profiling (Sesardic 2018). This is true when we expose ourselves or others to unreasonable risk by refusing to use information that could help prevent serious harms. For example, when hiring a babysitter for the night or a nanny to take care of young children, it is reasonable to assume a 50 year-old woman is less likely to abuse the children than a 25 year-old man. It would be wrong to never hire men for jobs as nannies or babysitters.
But it is neither irrational nor immoral to be more wary of men in their 20s than women in their 50s, given that much more violent crime is done by men in their 20s.
Still, to the extent that individually rational generalizations about groups conflict with our commitment to treat one another as people with similar moral standing, we may have reasons to endorse laws that prohibit certain forms of statistical discrimination. On this view, equal protection laws may be a way of solving a moral collective action problem. If each of us sometimes acts in ways that are inductively rational but morally objectionable according to principles we endorse, we may have reasons to support a policy of equality under the law to minimize the likelihood that we'll rely on generalizations too much in certain areas of life.
Suppose you're an employer sifting through hundreds of applications for a temporary job. You look at resumes, recommendation letters, and so on. But you have very little time and you have strong evidence that members of certain groups tend to be best at the kind of job you're selecting for. You therefore discount all members of other groups in order to minimize search costs. If we each recognize that we'll be tempted to act in this way, but we'd be better off on moral grounds if employers didn't act this way, we might have good reasons to endorse legal restrictions that impose costs on employers who engage in certain kinds of statistical discrimination. 4 We are not arguing in favor of such laws, but pointing out that there could be a rationale for antidiscrimination laws that classical liberals might endorse.
In sum, in the absence of perfect information and large differences in morally relevant traits, classical liberals seem logically committed to endorsing institutions that treat ordinary people as equals under the law, as well as having equal moral standing, provided they meet a basic threshold of certain capacities. But they should not expect all groups to exhibit identical outcomes when people are free to pursue their goals and exploit their talents. Not all indication of difference is evidence of oppression.
By committing the egalitarian fallacy, many self-described liberals (or "progressives") have tacitly assumed that if individual or group differences are rooted in biology, the principle of moral equality cannot be saved. We think this is false. But we may be wrong. If we are, the current century will be a turbulent one.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.