The New York Times recently had a piece by their new and woke science reporter, Katherine J. Wu, which is basically an indictment of science journals for not keeping track of the “diversity” of authors and reviewers of the papers they publish or reject. The implicit message is that science journals are racist, discriminating against papers by minoritized authors.
Click on the screenshot to read the article:
Wu’s implicit assumption is twofold. First, that a paucity of diversity—which of course means ethnic diversity, but minus Asians since they are surely overrepresented among authors—reflects racism on the part of scientific journals and reviewers. There is no consideration of whether a lack of diversity may represent simply a paucity of minority authors and reviewers. That itself may reflect racism, past or present, that narrows the opportunities of would-be scientists, but the article implies that it’s racism acting on Ph.D. authors trying to submit papers.
The second assumption is that more ethnic diversity in journals means better science. Well, that’s true in the sense that the more people who get the opportunities to become scientists, the higher the average quality of the science that is published. But I’m not at all convinced that members of any group, be they groups involving genders, religions, incomes, or ethnicity, have a special “point of view” based on their group identity that makes them do science differently. Science is science, and I don’t feel that Hispanics, say, have a different “way of knowing.” (There may be one exception here, that I’ve mentioned before: I think women scientists are responsible for shifting the focus in sexual selection from male traits alone to female preferences as well. But many men were also involved in this shift). In the end, the best science comes from giving everybody equal opportunities, not practicing remediation based on race at the publication level.
But the question is whether journals should be publishing more papers by members of minority groups. That is, is there a bias against, say, black or Hispanic authors that needs to be rectified by that form of “affirmative action” on the publication level—taking steps to accept more papers by minority authors?
It’s my opinion that the answer is “no”. This presumes that a paucity of papers by such authors is prima facie evidence for bias, when it may reflect only a paucity of minority-group members in the field, or of minority scientists submitting papers, or submitting fewer papers,—rather than reviewers deliberately discriminating against papers by minority authors.
It may be worth investigating this issue, but I consider it hardly worthwhile for two reasons. First, figuring out whether a paucity of papers from minority group members is due to racism at the reviewing level is very hard to do, though not impossible (see below). More important, it’s certainly true that the disparity between the proportion of minority-group members in the population and the number of papers published by members of that group is due largely not to racism but to an underrepresentation of Blacks and Hispanics in science. Figuring out why that disparity exists is the best way to achieve more proportionality in science, if that is your goal. And that’s really where our efforts should be going.
Here are the data given in the NYT piece from two scientific organizations showing disparities between population proportions and publication/reviewer proportions. The article makes the point that most journals, though, do not keep records of the ethnicity of authors and reviewers. (To clarify for non-scientists, when scientists submit papers to a journal, those papers are sent to several reviewers—usually two or three—who are experts in the area of research. Based on the reviewers’ assessment of the paper, the editor then decides whether or not to publish it. If the decision is “yes,” there is often some revision of the paper required, either in the discussion or the scientific analysis.)
I’m going to discuss authors here, not reviewers, because it is the quality of authors‘ work that, by and large, constructs the quality of the journal. How do we know if a journal is discriminating against minority authors? You can’t simply use a difference between the proportion of people in the field, or the proportion of people submitting papers on the one hand, and the proportion of papers published on the other, as a criterion for bias. That’s because members of different groups may submit papers less often, or of lesser quality, and that this would lead to differential representation that would not reflect racism. Bias must be proven, not assumed.
There are two ways to solve this problem. The first is the equivalent of doing “blind” auditions for orchestras—auditions in which those seeking an orchestral chair perform behind a curtain. That “blind” system removes all bias against sex or race. To do this with a paper, you simply remove the names of the authors, their institution, and the acknowledgments from the manuscript, so the reviewers don’t know who wrote the paper. (There are, of course, ways to guess, like if an author cites herself repeatedly, but in many cases this will indeed lead to quality appraisal ignorant of the author’s race or gender.)
I hit upon that system in the late 1970s when I was a postdoc, full of piss and vinegar and concerned that papers were getting preferentially published not because of race, but because of reputation. My idea was that famous people had an easier time publishing their papers than small fish (like me!). I wrote letters—real letters—to the editors of about 30 journals in my field, proposing that manuscripts be reviewed blind this way. I got only one response, and that was from an editor who said that he preferred knowing the authors, because famous authors were more likely to submit better papers! That may be true on average, but it’s not the best way to ensure the quality of papers in a journal! In fact, famous authors may get by more easily with shoddier work because of their reputations.
At any rate, some journals have now wisely decided to adopt the blind-author technique, and more power to them! It seems to me a step in the right direction to eliminate animus not just against groups of people, but against your scientific “enemies” or in favor of your scientific “friends”. (Believe me, this kind of bias is rife in science.) While you can get around this system by guessing, I think it does help ensure objective reviewing and thus higher-quality papers. (I should add that the NYT music critic opposed blind auditions because he said that while it increased the proportion of women in orchestras, it didn’t eliminate racial inequities; his view was clearly that equity trumped orchestral quality.)
The other way would be to do an experiment submitting identical sets of manuscripts with fake names that give clues to the gender or ethnicity of the authors. If manuscripts with women or minority authors are rejected more often than the same manuscripts with “white” or “male” names, that surely indicates bias. This was what was done in a laborious study of grant reviewing, using made-up “black”, “white”, “male” and “female” names on identical proposals. The study showed no evidence of racial or gender bias in grant evaluation. Needless to say, you don’t hear much about this study, even though it was a good one, as the results went against people’s certainty that there must be sex and gender bias in reviewing.
That experiment could be done with paper reviewing too, and really must be done before you can start making implicit accusations of bias. But I favor the blind-reviewing technique. You don’t have to do any experiments to see if that one makes things more equitable because, by eliminating a source of bias from the outset, it almost has to. It is my feeling that a “fake name” study wouldn’t show evidence of bias in pubication, but that’s my feeling alone. Better just to practice blind reviewing rather than speculate or do experiments.
In the end, my feeling is that affirmative action should not be applied to reviewing papers by people who already have doctorates, and, while I believe in affirmative action, I think it has to stop at some point in the hierarchy. My point comes after faculty hired hiring. I think it’s okay and useful to take race and gender into account when hiring junior faculty, as well as in college and grad-school admissions, but that’s where it stops. Ethnicity and gender should not be a consideration in getting tenure, full professorships, or in getting papers published—areas where merit alone should be the only criterion. Again, this is my view, and others may disagree.
Some of those who disagree think the whole system of a scientific meritocracy is flawed—that there isn’t even a scientific meritocracy. The NYT article says that:
Publishing papers in top-tier journals is crucial scholastic currency. But the process is deeply insular, often hinging on personal connections between journal editors and the researchers from whom they solicit and receive manuscripts.
“Science is publicized as a meritocracy: a larger, data-driven enterprise in which the best work and the best people float to the top,” Dr. Extavour said. In truth, she added, universal, objective standards are lacking, and “the access that authors have to editors is variable.”
To democratize this process, editors and reviewers need to level the playing field, in part by reflecting the diversity that journals claim they seek, Dr. Kamath said. “People think this is a cosmetic or surface issue,” she said. “But in reality, the very nature of your scholarship would change if you took diversity, equity and inclusion seriously.”
This whole section is to imply that there is little correlation between the merit of a paper and the chance of its being published. I think that’s a foolish conclusion, with the caaveats that Wu gives meant to imply a weak correlation at best. This is not my experience in reviewing papers or assessing published papers. Yes, sometimes a terrible paper gets published in a good journal, and a great paper gets rejected by a good journal, but there is surely a correlation between the quality of a paper and the chance that a. it will get published, and b. that it will get published in a prestigious journal.
No, to democratize the process, just do blind reviewing. That will go a ways toward eliminating bias. But even in the absence of that procedure, journals would be hard pressed to construct a system that would give preferential publication to papers by ethnic minorities. Regardless of what Katherine Wu thinks, science largely is a meritocracy, at least when it comes to publication, and I don’t think it would be good for science as a whole to bump papers up or down based on the race of their authors.