In my prior post, I made four points:
1. The news outlet for the American Association for the Advancement of Science (AAAS) reported that the attendees at its conference on peer review made numerous claims about the power and prevalence of race and gender bias in peer review.1
2. That report described no evidence of such biases, it simply reported proclamations of bias.
3. I contacted one of the attendees who confirmed that no such evidence was presented.
4. I then contacted Dr. Molly Carnes, a second attendee, because she was quoted as having made such proclamations. I asked for sources. Although Dr. Carnes graciously provided sources, none provided evidence of race or gender biases in peer review.
For brevity, I did not include a discussion in my prior post of one additional source that Dr. Carnes did provide. However, that one source is another example of how blind spots and cherrypicking that plausibly are interpreted as motivated by political or social justice goals leads to unjustified conclusions.
Dr. Carnes directed me to this article: Budden et al, 2008, which claimed to have found evidence of sex bias in acceptance of journal articles in Ecology between 1997 and 2005. And it made a splash large enough to have been cited in an editorial in Nature as evidence of sex bias in peer review. So, finally, we have clear, credible evidence of sex bias in peer review in a modern American journal, right?
Not so fast.
Not long after the Budden et al article appeared, another team re-analyzed their data showing that there was no evidence of sex bias there. The takedown was sufficiently persuasive that Nature published a retraction of its original editorial claim, declaring, “Nature has concluded that ref. 1 [ this refers to Budden et al] can no longer be said to offer compelling evidence of a role for gender bias in single-blind peer review. In addition, upon closer examination of the papers listed in PubMed on gender bias and peer review, we cannot find other strong studies that support this claim.”
A great guest post you can find here on the sexism in science controversies, by Claire Lehmann, founder of the terrific blog site, Quillette, gives some pretty good insight into why Nature would reach such a conclusion.
Nature, in contrast to the AAAS conference, at least admits there is little evidence to support this claim. It gives me some reason to believe that science is not hopelessly polluted by social justice action agendas unhinged from evidence.
THE FULL BREADTH OF THE EVIDENCE
Neither here, nor anywhere, have I claimed anything so ridiculous as “there is never any bias against women in any field of science.” Papers claiming gender bias in STEM are legion, and there are too many to report and review here. In my view, some do indeed provide persuasive evidence of the presence of gender bias in the specific context in which it was studied, such as the famous Moss-Racusin et al study finding gender bias in selection of a lab manager.
1. Many studies find “gaps,” and reach conclusions that gaps constitute evidence of bias, when, in fact, discrimination is only one of many likely explanations for gaps (see, e.g., my post on Simpson’s Paradox; or Ceci & Williams, 2011, review of sources of the gender gap in science). Some find correlates of gaps around which impressive narratives can be told, without even testing for the existence of discrimination.
If one wishes to reach a knee-jerk conclusion that “gap=discrimination” then, by far, the most extreme discrimination in the social sciences and humanities is not gender bias, and is not even race bias – it is political bias (because Republicans and conservatives are far more vastly under-represented in almost any social science field than are women or ethnic minorities). However, all gaps – race, gender, political, or any other – are simply correlations. And as any intro psych student knows, you cannot infer any particular causal mechanism (in this case, discrimination) produces any particular correlation (in these cases, gaps), without direct evidence showing such causal relationships.
2. Some studies that have claimed to have found evidence of bias have been either debunked (such as the Budden et al study described here), have found evidence of bias in beliefs or attitudes without providing evidence of discrimination, and/or have not successfully ruled out plausible non-discriminatory sources of the gap under study (e.g., any correlation may result from causal processes in the exact opposite direction than the researchers assume -- for example, showing that a stereotype correlates with a gap could occur because the stereotype causes the gap or it could occur because the gap causes the stereotype or because any of a myriad of third factors may cause both the gap and the stereotype).
3. Further, the only way claims of “pervasive bias against women in science” can be maintained is by cherry-picking studies that find evidence of bias against women and then systematically ignoring studies finding evidence of studies finding either no bias or even bias in favor of women.
The idea that there even exist studies showing bias in favor of women in science may be a shock to some readers, but, shocked or not, such studies clearly exist:
1. Across five experiments, Williams & Ceci (2015) found a 2:1 bias in favor of hiring women in STEM fields
2. Lloyd (1990), in an experiment holding manuscript content exactly identical and manipulating only the gender of the author, found that male faculty were unbiased in their evaluations of papers by male and female authors, whereas female faculty were biased in favor of female authors. This of course produces a net bias in favor of female authors.
3.Veldkamp et al (2017) surveyed scientists, and found that, whereas men viewed men and women as equally intelligent, rational and objective, women were biased in favor of women, viewing women as more intelligent, rational, and objective than men. Although this is not a study of peer review, it is disturbingly consistent with Lloyd’s (1990) findings, which was a study of peer review.
4. A study of German sociologists (Lutter & Schroder, 2014) found that women were 1.4 times more likely than men to get tenure, and needed 23-44% fewer publications to do so.
5. Like Lloyd (1990), found that male academics evaluated men’s and women’s productivity in academia as comparable, but women viewed women’s productivity more positively than they viewed men’s.
To be clear, I am not claiming that even the evidence compiled here “refutes” the idea that there is sometimes bias against women in the sciences. This blog entry is meant to be a corrective to the “all bias all the time” perspective that pervaded the AAAS conference and pervades much discourse on these issues; it is not meant to be a thorough and even-handed review of all the evidence. Instead, by highlighting so many studies providing evidence of pro-female bias, I am highlighting how cherrypicking and double standards is the only way that perspectives claiming pervasive bias against women can be advanced.
My point is not to deny evidence of bias; it is to get those proclaiming bias to grapple with all the evidence, even that inconsistent with their claims. It is to point out that the full weight of the evidence is far more complex and less definitive than simple knee-jerk claims of bias suggest.
Which is worse?
1. For there to be race and gender bias in peer review of scientific journal articles and grant applications,
2. For scientists to claim unconscious biases discriminate against African Americans and women in peer review, without clear and consistent evidence documenting such biases?
Fortunately, we do not have to make this choice: Both are bad. Both have no place in science. A finding’s importance, or a grant proposal’s potential, does not hinge on the race or gender of the person producing it. And if its ratings do hinge on author demographics, we should want to know it, and take action to eliminate such biases both to end discrimination against those groups, but also for the sake of advancing a valid and credible science.
On the other hand, when scientists act like politicians and just sorta make stuff up or cherrypick to advance an agenda, this damages scientific validity and risks undermining the credibility of and support for science among funders and the lay public. When the public cannot trust scientists to make statements that are actually true to the full scope of evidence, we have all started down the road to Scientific Hell.
Unfortunately, the quickest way to improve the pace of scientific self-correction would be to increase the mortality rate among scientists. Don’t believe me? OK, then would you believe physics Nobel laureate Max Planck, who said:
“A new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die, and a new generation grows up that is familiar with it.”
1 Psych Today is often visited by lots of people not completely familiar with scientific processes, so this might be worth explaining. Scientific articles usually get published, and grants funded, only after peer review: experts on the topic are asked to evaluate a paper, and only if they say it is good enough to deserve publication does it get published. In my field of social psychology, rejection rates at the best journals are usually 80% or higher, which is the same as saying you are fortunate if you have a 15-20% chance of getting reviewers to say your paper is good enough to publish there.
As usual, please read my guidelines for commenting before doing so. In short, no snark, sarcasm, or insults, and please stay on topic.
For updates on new blog entries here, and pieces by others (including their blogs, editorials, long form essays, and scientific articles) on this and related topics, follow us on Twitter.