1. Introduction

Despite much effort and multiple demands formulated over decades, women’s underrepresentation in STEM (science, technology, engineering, and mathematics) fields continues to be a problem. In the United States, women represent only 19% of graduates in physics, 29% in mathematics and statistics, 24% in engineering, and 20% in computer science (NSB 2018). Moreover, women with doctoral degrees in STEM continue to be employed at a lower rate: 19% of women between ages 30 and 65 are unemployed compared to 8% of their male counterparts. And, women are less likely to be employed in a science and engineering occupation: 18% are employed in their fields of study compared to 33% of men (NSF 2019).

When it comes to understanding what causes women’s underrepresentation in STEM, there is a storied debate that began, as Rolin (Reference Rolin2006) shows, with Jonathan Cole’s Fair Science (Reference Cole1987). Cole argues that there are two possible explanations for the underrepresentation of women in science: overt discrimination and self-selection. In response to this simplification, feminists introduced the idea of subtle forms of discrimination such as a chilly climate and gendered practices, thus calling into question the idea of women choosing to opt out of academia themselves, be it more or less voluntarily. The concept of subtle forms of discrimination provides a more complex and adequate picture of the distribution of agency and responsibility because it does not reduce the problem to a bifurcation in which there are “malicious agents” on the one hand and “helpless victims of discrimination” on the other. Instead, it envisages a richer concept of discrimination, one that includes the possibility that discrimination is sometimes an “unintended outcome of certain actions and social arrangements” (Rolin Reference Rolin2006, 132).

Gender bias, in its various forms, plays a central role in such subtle discrimination. Currently, evidence in a number of quantitative and qualitative, empirical and experimental studies is mounting, indicating that gender bias is not only present in academia (Steinpreis, Anders, and Ritzke Reference Steinpreis, Anders and Ritzke1999; Bug Reference Bug2010; Knobloch-Westerwick, Glynn, and Huge Reference Knobloch-Westerwick, Glynn and Huge2013; Handley et al. Reference Handley, Brown, Moss-Racusin and Smith2015) but also an important factor contributing to the gender gap (e.g., Nosek et al. Reference Nosek2009; Moss-Racusin et al. Reference Moss-Racusin, Dovidio, Brescoll, Graham and Handelsman2012; Reuben, Sapienza, and Zingalesc Reference Reuben, Sapienza and Zingalesc2014; Sheltzer and Smith Reference Sheltzer and Smith2014; Williams, Hall, and Williams Philips Reference Williams, Hall and Philips2014; Greenwald, Banaji, and Nosek Reference Greenwald, Banaji and Nosek2015). Consequently, in recent times the role of gender bias in the academy has been subjected to heightened scrutiny by philosophers approaching the issue from such different angles as social philosophy, epistemology, or philosophy of science (e.g., Dotson Reference Dotson2011a, Reference Dotson2011b; Lee and Schunn Reference Lee and Schunn2011; Crouch and Schwartzman Reference Crouch and Schwartzman2012; Hutchison and Jenkins Reference Hutchison and Fiona Jenkins2013; Leslie et al. Reference Leslie, Cimpian, Meyer and Freeland2015; Brownstein and Saul Reference Brownstein and Jennifer Saul2016; Bright Reference Bright2017; Leuschner Reference Leuschner2019). In light of this robust body of research—combined with the accumulation of anecdotal evidence provided by men and women from virtually all academic fields—it is reasonable to conclude that gender bias plays a significant role in causing and reproducing the underrepresentation of women in many academic fields.

Such findings notwithstanding, research on gender bias in academia is still continually subject to attacks alleging its lack of reliability or trustworthiness, as was most recently the case in new studies claiming to present evidence to the contrary. In their work published in 2011, social psychologists Stephen Ceci and Wendy Williams conduct a review of 20 years of correlation studies on gender discrimination in science, concluding that “the evidence shows women fare as well as men in hiring, funding, and publishing (given comparable resources).” Further, they find that “continuing to advocate strategies successful in the past to combat shortages of women in math-based fields today mistakes the current causes of women’s underrepresentation” (Ceci and Williams Reference Ceci and Williams2011, 3161).

In a 2014 monograph, they conduct an extensive life-course comparison of women in both math-intensive and non-math-intensive fields, claiming that women have come close to parity with men. They find that women doctorates in math-intensive fields “have equivalent access to tenure-track academic jobs in science, and they persist and are remunerated at comparable rates [to men]” (Ceci et al. Reference Ceci, Ginther, Kahn and Williams2014, 75). Furthermore, they conclude by explicitly rejecting the existence of gender bias in the academy: “Our analyses and research synthesis led us to … the conclusion that the overall state of the academy is largely one of gender neutrality” (125).

More recently, Williams and Ceci (Reference Williams and Ceci2015b) also published an experimental study on faculties from 363 universities and colleges in the United States, in which they evaluate hypothetical female and male candidates for a tenure-track position. According to the study results, there is “a 2:1 preference for women by faculty of both genders across both math-intensive and non-math-intensive fields, with the single exception of male economists, who showed no gender preference” (5360). With this experimental study, the authors aim to fill the lacuna in the literature on gender bias in hiring where, according to them, “no current experimental study demonstrating sexist hiring is based on actual faculty members who were asked to rate identically qualified tenure-track assistant professor applicants” (5361). Having obtained positive results for women candidates, Williams and Ceci go on to conclude that “after decades of overt and covert discrimination against women in academic hiring, our results indicate a surprisingly welcoming atmosphere today for female job candidates in STEM disciplines, by faculty of both genders, across natural and social sciences in both math-intensive and non-math-intensive fields. … It is an auspicious time to be a talented woman launching a STEM tenure-track academic career, contrary to findings from earlier investigations alleging bias” (5364–65).

Ceci and Williams’s results are surprising and unexpected as they challenge and, in many cases, are diametrically opposed to mainstream research on gender bias. On the basis of the wide-ranging conclusions that the authors reach, they present a neutral or even positive picture for women seeking academic careers, which seems to undermine the case for the existence of gender bias in academia. They imply a strong critical stance against not only the testimonies of multiple women who have suffered gender discrimination in their academic careers but also a swathe of scientific research confirming the impact of gender bias in the academy.

In this article, we will argue that research denying gender bias in academia is socially and epistemically detrimental. We will focus on the example of Ceci and Williams’s research and draw a parallel between their studies and studies denying anthropogenic climate change. Building on the Inductive Risk Account of Epistemically Detrimental Dissent (IndRA; Biddle and Leuschner Reference Biddle and Leuschner2015), we will argue that, operating similarly to climate change denial, denial of gender bias in academia is epistemically problematic given current societal conditions (sec. 2). Moreover, drawing on Kitcher’s Millian argument against the freedom of inquiry, we will argue that Ceci and Williams’s research is also socially problematic as it co-occurs with political and epistemic societal asymmetries (sec. 3). We will conclude that research denying the existence of gender bias has serious social and epistemic effects: it not only strengthens the already pervasive social biases of contemporary societies but also stifles discussion as intimidated researchers tend to become reluctant to address certain questions or to bring certain hypotheses to the table in the first place.

However, before we begin it should be noted preemptively that targeting specific dissenting views may strike some as dangerous, for example, as a potential pathway to censorship, silencing, and other unscientific attitudes. After all, society praises science for not being dogmatic and for thriving on debate and criticism. Indeed, de Melo-Martín and Intemann (Reference de Melo-Martín and Intemann2014, Reference de Melo-Martín and Intemann2018, chap. 5) stress this point in particular. However, manufactured dissent does not actually contribute to scientific progress; rather it, in fact, actively obstructs scientific practice, misinforming both policy makers and the public (Leuschner Reference Leuschner2018; Nash Reference Nash2018). It is, therefore, appropriate to disqualify such inappropriate dissent because the consequences of not doing so may be even more dangerous. A good example of this risk is illustrated by the tobacco industry whose dissenting campaign led to the deaths of millions of people worldwide (Proctor Reference Proctor2012). Given the growing number of documented cases of manufactured dissent (McGarity and Wagner Reference McGarity and Wagner2008; Michaels Reference Michaels2008; Oreskes and Conway Reference Oreskes and Conway2010; Elliott Reference Elliott2011; Proctor Reference Proctor2012; Markowitz and Rosner Reference Markowitz and Rosner2013; Gillam Reference Gillam2017), the distinction between epistemically fruitful and epistemically detrimental dissent seems more pressing now than it ever was. The case of dissent on gender bias in academia is, as we will now argue, another case of epistemically detrimental dissent.

2. The Inductive Risk Account of Epistemically Detrimental Dissent

Through an analysis of climate change denial, Justin Biddle and Anna Leuschner—also a coauthor of this article—have provided, an IndRA (Biddle and Leuschner Reference Biddle and Leuschner2015). Without ignoring the importance of dissent for scientific practice, they argue that, in some cases, dissent can function to the detriment of the epistemic goals of science. The IndRA identifies four jointly sufficient (but not necessary) conditions for epistemically detrimental dissent: “Dissent from a hypothesis H is epistemically detrimental if each of the following obtains: (1) The non-epistemic consequences of wrongly rejecting H are likely to be severe. (2) The dissenting research that constitutes the objection violates established conventional standards. (3) The dissenting research involves intolerance for producer risks at the expense of public risks. (4) Producer risks and public risks fall largely upon different parties” (273).

Thus, dissent meeting these conditions is epistemically detrimental as it affects academic debate. In climate science, for example, the discourse has been seriously affected by dissent on climate change. In this context, “dissent on climate change” does not mean “a skeptical view of the validity and utility of [climate] models” but, in accordance with Lahsen (Reference Lahsen2008) and Dunlap (Reference Dunlap2013), the case of “dissident” or “contrarian scientists” who “strongly criticize climate science” (Biddle and Leuschner Reference Biddle and Leuschner2015, 262). Such dissent has contributed substantially to a social atmosphere in which climate scientists are targeted by aggressive personal attacks from climate change deniers. Climate scientists are confronted with people touting manufactured pseudoevidential claims to disprove their findings, and they are subjected to continual doubt with regard to their competence and reliability, eventually creating an antiscientific atmosphere leading many of them, as Bradley puts it, to “keep a low profile and go with the flow” (Reference Bradley2011, 137; see also Brysse et al. Reference Brysse, Oreskes, O’Reilly and Oppenheimer2013; Lewandowsky et al. Reference Lewandowsky, Oreskes, Risbey, Newell and Smithson2015; Medimorec and Pennycook Reference Medimorec and Pennycook2015). As Oreskes and Conway write: “At a recent conference, a colleague told one of us that in IPCC discussions, some scientists have been reluctant to make strong claims about the scientific evidence, lest contrarians ‘attack us.’ Another said that she’d rather err on the side of conservatism in her estimates, because then she feels more ‘secure’” (Reference Oreskes and Conway2010, 264–65). Consequently, climate change deniers have and continue to obstruct scientific progress because “scientists fear to address certain topics and/or to defend hypotheses as forcefully as they believe is appropriate” (Biddle and Leuschner Reference Biddle and Leuschner2015, 262).Footnote ¹

The case of climate change denial shows that dissent meeting the four IndRA conditions is conducive to polemic criticism and concomitant personal attacks on researchers. This, in turn, has epistemically detrimental effects on those involved in the criticized approach given that the dissent gains powerful support in the society in which it is embedded. Researchers tend to feel intimidated and thus endeavor to avoid any failures by means of exceptional industriousness; they become reluctant to bring any aspect to the table that may not have been thoroughly thought through, or they even become averse to addressing specific questions at all (Brysse et al. Reference Brysse, Oreskes, O’Reilly and Oppenheimer2013; Lewandowsky et al. Reference Lewandowsky, Oreskes, Risbey, Newell and Smithson2015; Biddle, Kidd, and Leuschner Reference Biddle, Kidd and Leuschner2017; Leuschner Reference Leuschner2018).

In what follows, we argue that the recent studies denying the existence of gender bias in the academy meet the four conditions of the IndRA and, therefore, are prone to producing similarly problematic consequences. Of course, there are no producer and consumer risks involved in this case. However, this is only a matter of wording: if we speak more generally of risks of type I and II errors instead of producer and consumer risks, then, we argue, the IndRA can be easily applied to the case of dissent on gender bias in academia.

Condition 1 is clearly met because of the huge amount of empirical and anecdotal evidence of gender discrimination in academia. Claiming that these problems belong to the past, as Ceci and Williams do, and that things are just fine now sends the wrong message; that is, that it is the women’s fault if and, indeed when, they do not succeed. This attitude could in fact discourage women instead of giving them the opportunity to understand the problems they face and make informed decisions accordingly. Moreover, as Lee (Reference Lee2016, 266) convincingly emphasizes, the dissent can be detrimental to affirmative action and other programs seeking to support women academics.

Condition 3 is met here as well since Ceci and Williams, despite numerous objections (Zevallos Reference Zevallos2014; Williams and Smith Reference Williams and Smith2015), persist in rejecting the well-supported hypothesis that there is gender bias in academia. We will show this in more detail in the following section.

And, condition 4 is also met: the risks of type I and type II errors fall largely on different parties. In particular, wrongly accepting the hypothesis that there is gender bias in academia would lead to measures supporting women academics even though they were not required. This would be both costly to universities and unfair to men, while wrongly rejecting the hypothesis would, however, be especially harmful to women.

When it comes to condition 2, we need to go into more detail in order to show how Ceci and Williams’s studies flout conventional scientific standards. We will examine some of the most worrisome methodological concerns regarding Ceci and Williams’s studies without trying to be exhaustive. We begin with the reviews of their correlational studies (Ceci and Williams Reference Ceci and Williams2011; Ceci et al. Reference Ceci, Ginther, Kahn and Williams2014) and then examine their latest experimental results (Williams and Ceci Reference Williams and Ceci2015b).

3. Methodological Concerns Regarding Ceci and Williams’s Research

As mentioned above, Ceci and Williams’s (Reference Ceci and Williams2011) study is based on a review of correlational studies (i.e., studies investigating correlations between sex and disadvantages in academia), concluding that gender discrimination in academia no longer exists. Lee identifies substantial methodological problems in these correlational studies: “despite their laudable efforts to control for quality indirectly through proxy, the correlational studies cited by Ceci and Williams cannot rule out the possibility that a quality confound is responsible for cancelling out the impact of ongoing implicit gender bias in journal review, grant funding, and hiring. Nor can the correlational studies rule out the possibility of a quality-related sample bias in the sample of women and men” (Reference Lee2016, 266).

The correlational studies draw their conclusions exclusively on the basis of submission and acceptance rates. They do not take the possibility into account that there could be gender-specific differences in the quality of manuscripts, grant applications, or job dossiers. However, such differences are likely to exist: there is empirical literature suggesting that women—on the whole—tend to submit higher-quality papers than men in order to be maximally prepared for biased scrutiny (Sonnert and Holton Reference Sonnert and Holton1995, 152–54; Valian Reference Valian1998, 264–65, 275; Williams et al. Reference Williams, Hall and Philips2014; Lee Reference Lee2016, 266; Bright Reference Bright2017; Hengel Reference Hengel2017; Card et al. Reference Card, DellaVigna, Funk and Iriberri2020).

Moreover, there are issues with Ceci and Williams’s interpretation of their findings. In one of their more recent studies, Ceci and Williams once again claim for mathematically intensive fields that “manuscript reviewing and grant funding are gender neutral: Male and female authors and principal investigators are equally likely to have their manuscripts accepted by journal editors and their grants funded, with only very occasional exceptions” (Ceci et al. Reference Ceci, Ginther, Kahn and Williams2014, 76). From these results they reach the conclusion that “although in the past, gender discrimination was an important cause of women’s underrepresentation in scientific academic careers, this claim has continued to be invoked after it has ceased being a valid cause of women’s underrepresentation in math-intensive fields” (76). However, there is a significant issue with this interpretation. From the fact that there is no disparity in outcomes disadvantaging women in one specific academic context (e.g., manuscript or grant proposal assessment), it cannot be concluded that women do not experience discrimination in any academic context. For even if we allow for the possibility that gender bias is no longer present in those contexts (and that is a big “if”), that does not mean that there is no longer gender discrimination anywhere in the academy. However, this is exactly what the conclusions of their studies suggest: “that the overall state of the academy is largely one of gender neutrality” (125).

As a consequence, in order to explain the gender gap, the authors are forced into the position of claiming that the main cause of women’s underrepresentation in STEM “is not that women applicants are not being hired, but rather that they are choosing to opt out of academic science” (Ceci et al. Reference Ceci, Ginther, Kahn and Williams2014, 120). This explanation is highly problematic due to the fact that personal choices are constrained by institutional factors where gender bias is present. As Williams and Smith claim: “If science designs its professional ideals around a man married to a homemaker, that’s sex discrimination. … Requiring women to sacrifice having children as the price of an academic career in science guarantees a paucity of women” (Reference Williams and Smith2015, 5; see also Zevallos Reference Zevallos2014). In other words, personal choices are not made in an environment free of gender discrimination, contrary to what Ceci and Williams suggest. In this sense, their conclusions present faulty overgeneralizations insofar as they make unwarranted assumptions with regard to women’s freedom of choice.Footnote ²

With respect to their most recent experimental study, Williams and Ceci (Reference Williams and Ceci2015b) compare men and women job candidates only at the highest level, and, at least with respect to faculty hiring, one can remain skeptical as to how methodologically sound this is. There are probably important differences between career stages, and it seems likely that gender bias affects women academics mostly early in their careers, “before they reach the stage of being perceived as academic superstars” (Brownstein Reference Brownstein2015, referring to Bryce Huebner). This view is strongly supported by a survey asking 238 male and female academic psychologists to assess applications from men and women, finding significant differences between evaluations of the suitability for hiring and those for tenure with regard to male and female academics:

Consequently, Steinpreis and colleagues conclude that “a superb record may indeed function as a buffer for gender bias when making promotional decisions” (524). In other words, by using only extremely qualified candidates in their study, Williams and Ceci (Reference Williams and Ceci2015b) are disregarding the fact that graduate students and junior scholars can be (and in fact the evidence suggests that they are) targeted by gender bias—regardless of the fact that the “academic superstars” have reached a status granting them (more) equal treatment. Moreover, and in contrast to Ceci and Williams’s self-evaluation, it does not seem at all surprising that “academic superstar women” might be preferred over “academic superstar men,” as hiring committee members would likely rate a “big shot” woman as a “good catch” for their institute: not only do they gain a highly qualified colleague, but as a positive side effect they also improve their gender statistics.

Thus, examining gender bias in hiring processes in academia requires the comparison of different levels of qualification, not only top-qualified women vis-à-vis top-qualified men but also lower-level candidates of both genders. In addition, it is important to note that Ceci and Williams have responded several times to their critics via different media outlets such as American Scientist (Ceci and Williams Reference Ceci and Williams2015a), the Chronicle of Higher Education (Ceci and Williams Reference Ceci and Williams2015b), and even a five-piece response in the Huffington Post (Williams Reference Williams2015). The authors provide methodological explanations and address some of the objections that emerged after their 2015 experimental study came out. But their responses do not address what we consider to be the most fundamental limitations of their whole undertaking. First, they do not respond to the criticism of the frequent overgeneralization they make in their study conclusions, that is, from evidence against gender bias in a particular context to the conclusion that there is no gender discrimination in science writ large. And second, they do not acknowledge the fact that gender bias is likely to be tacitly involved in women’s choices within a sexist society. So, even if the results show that women are “opting out” instead of being “pushed out,” gender discrimination may still play a role in an individual woman’s decision, and, thus, it is incorrect to assume that there is no discrimination at work simply because it is a personal decision. In this sense, we find that Ceci and Williams have largely ignored the most pressing objections to their research program.

Claiming that gender discrimination is a problem of the past overlooks not only a huge swathe of anecdotal evidence from women academics (e.g., Monroe et al. Reference Monroe, Ozyurt, Wrigley and Alexander2008; Roos and Gatta Reference Roos and Gatta2009; Pololi et al. Reference Pololi, Civian, Brennan, Dottolo and Krupat2013; VanDerwarker et al. Reference VanDerwarker, Brown, Gonzalez and Radde2018), but it also bypasses the selectively rotten rungs on the career ladder. In contrast, Brennan provides a reasonable and cogent explanation for the situation, namely, that students who experience systematic disadvantages are unlikely to persevere in the field: “We might start out not seeing junior women as qualified, and then later we look and there are no senior women who are qualified. … It is often the case that the later judgment—‘There are no qualified senior women’—is correct even though our earlier judgment made it so” (Reference Brennan2013, 187).

This is not at odds with Ceci and Williams’s (allegedly contradictory) findings; furthermore, it provides an explanation for both the disadvantages while moving up the career ladder and the enormous amount of anecdotal evidence on the experience of gender bias provided by so many women academics—particularly by students and young researchers—from so many disciplines.

So far, we have listed a number of methodological issues with Ceci and Williams’s dissenting research. However, one could wonder whether these really constitute violations of established conventional standards, as required by the second IndRA condition, or whether Ceci and Williams merely make questionable methodological choices in their research.Footnote ⁴ For example, by using correlational studies based on proxies (Ceci and Williams Reference Ceci and Williams2011) or by selecting only top-qualified candidates to examine gender bias in hiring processes in academia (Williams and Ceci Reference Williams and Ceci2015b), they clearly make poor methodological choices insofar as there were better choices available for the intended research purpose. However, the argument could still be made that these issues themselves do not constitute violations of methodological standards in that the research adequately followed methodological rules and the researchers did not engage in any sort of scientific fraud.Footnote ⁵ Furthermore, it is also not a violation of a methodological rule for Ceci and Williams to ignore criticism. Indeed, while this lack of receptiveness to critical voices can clearly be seen as disregard of a social norm at the heart of the scientific enterprise and key to epistemic progress (Longino Reference Longino2002, 128–30), there is “no methodological rule that specifies the right balance of responsiveness or resistance to evidence and critics” (Carrier Reference Carrier, Christian, Hommen, Retzlaff and Schurz2018, 158).

Such considerations notwithstanding, Ceci and Williams do actually violate a methodological rule in that they make overgeneralizations in their conclusions going well beyond what their results could possibly warrant. This is not a mere poor interpretive choice. Limiting the scope of a scientific study accurately is a well-established scientific standard, which, if violated, leads others to believe something that has actually not been sufficiently proved in the study. Carrier (Reference Carrier, Christian, Hommen, Retzlaff and Schurz2018) calls this violation “false advertising”: the issues alleged to be addressed are in fact bypassed because of a discrepancy between the design of the study and its use, meaning that the setup of the study “makes it rather insensitive to the issue it purportedly addresses. This incongruity is concealed and glossed over in the interpretation of the results” (162).

Given this violation of a methodological standard, we conclude that Ceci and Williams’s dissenting research meets all four conditions of the IndRA and is, thus, conducive to a societal atmosphere in which scientists come under attack, and their credibility, competence, and reputation is publicly cast in a doubtful light. In the next section, we illustrate this point with examples of how researchers studying gender bias have frequently been targeted.

4. Attacks against Researchers in Gender Studies

So far, we have argued that Ceci and Williams’s dissent on gender bias in academia meets the four IndRA conditions. Just as the denial of anthropogenic climate change promotes a hostile social atmosphere for climate scientists, analogously, the denial of gender bias in academia is also conducive to a hostile social environment for researchers studying gender bias.

Despite the methodological problems of their research, Williams and Ceci have succeeded in publishing their work in prestigious journals. Given the deeply entrenched gender bias in society, it is not surprising that their publications have received much attention and recognition by readers who are only too happy to discover studies debunking “the gender myth” and to feel confirmed in their disapproval of feminism and political correctness. To illustrate this point, one only has to read through the comments following one of Ceci and Williams’s (Reference Ceci and Williams2015b) online publications: “Apparently, many critics of the Williams Ceci PNAS article have strong vested interests in finding sex discrimination anywhere and everywhere, thus ignoring the possibility that our society has evolved and that some attitudes have changed. Many of these critics have jobs that depend on their universities and employers believing sexism is rampant” (Astro Woman). “Actually, sexism is rampant in business and academia. But it’s sexism in favor of females, not the other way around” (RalphF). “The effect of this innately specious strategy [of changing and expanding aims], very quickly caused feminism to shift from a movement ostensibly pushing for; Fairness for Women, Gender Equality and Female Emancipation, to one blatantly pushing for; Privilege for Women, Gender Superiority and Male Emasculation” (). “As suspected there *is* a gender bias in academic STEM hiring—I guess men will just need to accept being discriminated against because … why?” (Hootyman). “Ah, yes. There is no creature so angry as a feminist who is told that life is getting better for women in some way” (karen straughan). “There is only one institutionalized negative bias. It is against white males. Forward” (megapotamus).

Although we acknowledge that authors have no control over the online comments they receive, we wish to point out that it is not unreasonable to expect researchers working on issues of sexism to be aware of the social problems related to their work and to seek to avoid garnering support from sexists.Footnote ⁶ Otherwise, they ratify an atmosphere of intimidation in science that is epistemically problematic. In the case of Ceci and Williams, their sometimes polemic tone conveys the impression that they are perhaps even interested in stoking such sexist attitudes. When asked in Nature about the methodological criticism of their studies presented by a number of distinguished colleagues, Williams deemed it appropriate, for example, to simply respond that “people find it hard to accept when there’s change, even for the better” (Deng Reference Deng2015, 276). In an interview in Science, again, she claimed that “it’s tempting to blame gender when you don’t get a job and you’re a woman. … It’s easier … than to admit that the entire premise of what you’ve done for the past 7 years of your life was flawed at the root” (quoted in Benderly Reference Benderly2015).

Such statements trigger polemic echoes and aggressive comments by people in online discussion forums and social networks. Researchers conducting studies confirming the existence of gender bias in the academy (and of gender bias in general) become discredited; their credibility, competence, and reputation are cast in doubt; and, in extreme cases, they are threatened and harassed.

Indeed, researchers working on gender bias in academia report that they have received hate e-mails and telephone calls and that they have become targets of right-wing social media sites; the attacks come from students and the public, who are in many cases incited by right-wing media such as Breitbart or Fox News (Andreassen and Myong Reference Andreassen and Myong2017; Ferber Reference Ferber2018). After the US 2016 election, the American Association of University Professors launched a campaign through which it sought to collect “experiences of targeted harassment and intimidation” of faculty members, and they received many stories from researchers of various fields including gender and race studies (AAUP 2017, 2018). “The attacks aimed at faculty take the form of harassment, abuse, and outright threats. Examples include ‘Traitor; Communist Jew pig; Fat cunt; N——; You deserve to be raped; I have seen your children; I know where you will be and I will be there.’ Under the guise of protecting free speech, these messages are clearly meant to silence” (Ferber Reference Ferber2018).

Such attacks are detrimental to scientific research, as they do not promote but rather stifle scientific discussion and, thus, epistemic progress. It seems likely, for instance, that the intimidation of researchers affects academic teaching, and if professors become too afraid to teach certain topics, this has implications on the discipline, the students who are attracted to the subject, the thoughts that are expressed, the ideas that are triggered, the debates that are raised, and the paradigms that are established.

Of course, Ceci and Williams do not participate in this intimidation directly. However, in the doubt they create regarding the reliability of gender studies, their research is conducive to this hostile atmosphere. Moreover, it is also socially problematic as Ceci and Williams operate in a social environment that is characterized by political and epistemic asymmetries, as we argue in the next section.

5. Political and Epistemic Asymmetries in Society

Unburdened by its methodological problems, Ceci and Williams’s research denying gender bias has been very successful in that it has caused a kerfuffle in the press and among fellow academics and has been widely disseminated and embraced in particular because of political and epistemic asymmetries in society. Certain media outlets and the internet play a pivotal role here. For example, Ceci and Williams’s studies were highlighted in the Washington Post (Kaplan Reference Kaplan2015) and the New York Times (Williams and Ceci Reference Williams and Ceci2014) as sensationalist news about women in academia who now have a “hiring advantage,” and academic reactions in social media, both for (e.g., Cesario Reference Cesario2017) and against (e.g., Zevallos Reference Zevallos2015), also emerged. Ceci and Williams themselves were also granted the opportunity to present their studies in the media in a way that garnered them attention: speaking with CNN, for instance, they claimed that “the only sexism [women] face in the hiring process is bias in their favor” (Williams and Ceci Reference Williams and Ceci2015a).

Such attention is, admittedly, typical for new and groundbreaking research results that are in conflict with a well-established mainstream. However, cases that meet the IndRA conditions are successful despite their methodological shortcomings, which has to do with the fact that there are substantial parties in society with a strong bias in favor of the research. This is expressed in condition 4 of the IndRA. However, the IndRA considers public risks and nonepistemic consequences as criteria for evaluating how epistemically beneficial or detrimental dissenting research is; thus, its focus is on the research realm and its epistemic goals, not on the effects that this dissenting research has on the societal realm. The argument in this section aims to close this gap. We will argue that the extreme success of such controversial dissent, despite its scientific shortcomings, is due to political and epistemic asymmetries in the society in which it is situated and examine how such dissenting research harms society.

The functioning and effects of such asymmetries are explained by Kitcher (Reference Kitcher2001, chap. 8) in his Millian argument against the freedom of inquiry. Following Kitcher’s line of argumentation, we suggest that a social environment characterized by political and epistemic asymmetries is one in which “there are sufficiently powerful inclinations to [support the critique], held by sufficiently powerful people” (102). Consequently, the criticism, even though it is methodologically flawed, is nevertheless embraced and supported by powerful parties, and this affects social improvement.

Kitcher shows that epistemic and political asymmetries exist with regard to certain research questions such as research concerning egalitarian hypotheses, meaning that social biases that are deeply entrenched in a society are much more easily exacerbated by confirming—even indecisive—evidence than ameliorated by contradictory evidence (Reference Kitcher2001, chap. 8). Kitcher discusses research cases in which it is highly unlikely that we will ever find decisive evidence. His concrete example is research regarding the question of whether there are biological differences in intelligence between certain social groups (e.g., between men and women or Blacks and whites). As intelligence research is highly unlikely to yield decisive results, Kitcher draws the normative conclusion that no research project on such questions ought to be pursued since, even in cases of indecisive results, people holding sexist or racist biases would feel confirmed in their misguided beliefs. Thus, such research would only lead to the aggravation of the situation—both epistemically and socially—because of existing biases against women and Blacks.

Kitcher’s argument can be applied to the case of climate change denial: it meets three asymmetries among the five asymmetries Kitcher identifies (Leuschner Reference Leuschner2018). With respect to the hypothesis of anthropogenic climate change (H_acc), two political asymmetries (PA) and one epistemic asymmetry (EA) hold:

As the results of climate science regarding the existence of anthropogenic climate change are decisive (with the probability of the existence of anthropogenic climate change being nearly 1), it does not meet Kitcher’s other two asymmetries (i.e., asymmetries that arise when results are indecisive), meaning that it is socially important to conduct research on anthropogenic climate change. However, because of the three asymmetries that are met by dissent on anthropogenic climate change, it is morally, socially, and politically detrimental to conduct research that denies the existence of anthropogenic climate change as it contributes to misinforming policy makers and the public and impedes climate change mitigation measures that would be important, particularly for people in developing countries who have been most affected by climate change. Thus, while research on anthropogenic climate change is desirable, dissenting research on this matter ought not to be pursued for moral, social, and political reasons.

The same holds, again, for the case of research on gender bias in academia: as we have strong evidence for the existence of gender bias in academia and scientific progress has already been made on this issue, and as there is reasonable hope that further progress will be made, research on gender bias ought to be pursued. In contrast, research that casts doubt on the existence of gender bias in academia ought not to be pursued because of the political and epistemic asymmetries: any evidence seemingly contradicting the existence of gender bias will have more influence on public knowledge than any evidence in fact confirming the existence of gender bias. To be more precise by way of example, an article simply claiming that there is no gender bias in academia will disproportionately strengthen the societal reservations against feminist concerns and demands, while an article that in fact presents evidence for feminist concerns and demands will not make a significant dent in the status quo (Handley et al. Reference Handley, Brown, Moss-Racusin and Smith2015).

However, one could rightly ask what this demand—that such research ought not to be pursued—means in practice.Footnote ⁷ This question concerns two intricate points. First, in general, one never knows where a specific scientific investigation of climate change or gender bias will lead, and, of course, we do not wish to make the sweeping claim that dissent in these fields is always problematic. There is normal scientific dissent on many details, in climate science as well as in gender studies. But just like in climate science, where the manufactured dissent ultimately always pertains to the factual status of anthropogenic climate change, in gender studies there are also cases of dissent that dispute “facts,” that is, hypotheses that are so clearly confirmed that what they postulate can be deemed beyond reasonable doubt. We take the existence of gender bias in academia to be such a “fact,” and, hence, we suggest that research explicitly seeking to undermine this “fact” (i.e., that strives to provide contrarian evidence) ought not to be pursued.

However, this line of thinking leads to a second point of contention one could be troubled by: What does it mean to say that such research “ought not to be pursued”? We do not wish to argue in favor of banning such research. Again, we follow Kitcher here, who points out that an official ban of such problematic research would allow deniers “to portray themselves as victims of an illegitimate public policy of stifling the truth” (Reference Kitcher2001, 106). Our conclusion is, thus, merely a moral imperative against conducting, supporting, or funding such research. It is clear that there can be borderline cases; there may be studies on these issues that do not seek to contradict the fact that there is gender bias in academia but nevertheless produce potentially dissenting results, that is, findings on specific aspects of the issue that could be misused by certain stakeholders in order to attack gender studies or affirmative action programs. However, we take it as a consequence of this moral imperative that researchers conducting such critical research on this issue behave responsibly by clearly and distinctly distancing themselves and their research from denialist positions.

Thus, research denying the existence of gender bias in academia, such as the dissenting studies by Ceci and Williams, is not just epistemically but also socially problematic because it hinders not only scientific progress by promoting an antiscientific atmosphere but also societal progress by supporting and exacerbating existing gender bias in our current democratic societies.

6. Conclusion

A huge and constantly growing body of empirical and anecdotal evidence shows that gender bias is a pertinent and recalcitrant problem in academia and society. This is not to say that gender bias is already fully understood in its functioning or that it is independent of other social problems. On the contrary, much more research is needed on all aspects of gender bias such as its interrelation with socioeconomic structural injustices, discrimination, intersectionality, the role of implicit bias, and stereotype threat. Unfortunately, those who seek to achieve a better understanding of these problems by doing research on gender bias have become targets of harassment and attacks that are substantially supported by manufactured dissent: denialists reach conclusions unwarranted by the available evidence, and they ignore fundamental objections to their methodological decisions.

We focused on the dissenting research by Ceci and Williams as an example of such denial, arguing that it meets the four conditions of the IndRA (Biddle and Leuschner Reference Biddle and Leuschner2015). As we know from the case of climate science, which is confronted with manufactured dissent denying the existence of anthropogenic climate change, dissent that meets these conditions has substantially serious epistemic effects: it promotes an antiscientific social atmosphere in which discussions are stifled because intimidated researchers become reluctant to address certain questions or to bring certain hypotheses to the table. Moreover, we have argued with recourse to Kitcher’s Millian argument against the freedom of inquiry that such dissent disproportionately supports social biases within current democracies due to existing epistemic and political asymmetries, also making it socially problematic.

This means that Ceci and Williams’s dissent counts as a case of manufactured dissent in the same way as dissent on anthropogenic climate change does. At first glance, this might seem implausible as there are no industrial stakeholders supporting them financially. However, there are nonnegligible fractions of (not only but particularly US) society (including the government) that support them nonmaterially. They receive disproportionate coverage in the media, spotlighting and helping to disseminate their work, then echoing through social networks. Normally, such inadequate work would be largely ignored, but here, because of the power differentials, it receives such strong support that it can flourish and successfully infiltrate scientific debates (O’Connor and Weatherall Reference O’Connor and Weatherall2019).

We acknowledge that behind a critique of dissent the danger of justifying the censorship of free debate may lurk, a point also emphasized by Ceci and Williams. In one of their opinion articles, they stress how concerned they feel about a recent Cato poll finding considerable percentages of respondents to be in favor of banning socially inappropriate statements from university discussions, including statements such as “men on average are better than women at math.” Ceci and Williams claim that “college students [should] be exposed to arguments on both sides of these issues, as part of their journey of intellectual development” (Reference Ceci and Williams2018).

Of course, we concur that one must be careful to avoid any general “discouragement to conduct studies or publish findings that may be counter to dominant views” (de Melo-Martín and Intemann Reference de Melo-Martín and Intemann2014, 598). Hence, we close by emphasizing that the epistemic and social concerns presented here regarding dissent on gender bias in academia do not refer to the factual content of the dissent (which we have criticized for methodological reasons) but to the fact that it is conducive to a hostile atmosphere for researchers in gender studies, sometimes even explicitly by employing a polemic tone. It not only contributes to the endorsement and, thus, the exacerbation of existing gender bias, but it is also epistemically detrimental. Our critique is, therefore, not a call for restricting but, on the contrary, for protecting free scientific and public debate.