New study shows gender bias in science against female students

September 22, 2012 • 8:01 am

In February of last year I reported on a study by Ceci and Williams  showing that, at least according to the authors’ methodology, there was little evidence of gender bias against women in academia for getting grants, being hired as a faculty member, or getting papers published. Although women still occupy faculty positions disproportionately less often compared to their acquisition of degrees, I found this result heartening. It was, however, a meta-analysis of many other studies, and these can be problematic.

I was therefore disheartened to see a new study, by Corinne Moss-Racusin et al. in the same journal—the Proceedings of the National Academy of Sciences (download free at link, I think; if you’re unable, email me for the pdf)—showing bias against women of a different kind: the hiring of students to be laboratory managers. This is not a meta-analysis, but a single sociological study, and—though I’m not an expert in the field—the results look sound to me.

The authors did a simple thing: they sent a group of application materials from a fictitious undergraduate looking a lab-manager job to 127 American biology, chemistry, and physics professors of both sexes. The applications were not for jobs in the professors’ own labs, but simply applications that the faculty were asked to evaluate for the student’s competence and hireability, as well as to decide what salary and how much mentoring the applicant could expect to get from them.

The applications were identical except for one thing: they had the name of either a male ( “John”, n = 63 applications) or a female (“Jennifer”, n = 64 applicants).  As the study states, “Faculty participants believed that their feedback would be shared with the student they had rated . . ”

The applications were designed to be good but not perfect: that is, the applicant had a few flaws. This was done to ensure that there would be discernible variation in how the applications were judged. If the applicant was perfect in every respect, it would be harder to judge any bias on the part of the raters.

The results are disappointing, for they show a substantial disparity between males and females in all categories, with women at the bottom. Surprisingly, female faculty were as biased as male faculty. All of the male-female differences in perceived quality were statistically significant.

This graph tells the tale for perceived competence, hireability, and mentoring; women are lower on all counts:

Fig. 1. Competence, hireability, and mentoring by student gender condition (collapsed across faculty gender). All student gender differences are significant (P < 0.001). Scales range from 1 to 7, with higher numbers reflecting a greater extent of each variable. Error bars represent SEs. n(male student condition) = 63, n(female student condition) = 64.

And here are the means, grouped by sex, for perceived competence, hireability, mentoring, and the salary that was deemed appropriate for the candidate.  Note that the gender of the faculty member evaluating the application is also given.

In the part of the table below, the categories—”competence” through “salary” are in the same order as above, but this time the applicant was Jennifer instead of John:

The caption from the paper: Scales for competence, hireability, and mentoring range from 1 to 7, with higher numbers reflecting a greater extent of each variable. The scale for salary conferral ranges from $15,000 to $50,000. Means with different subscripts within each row differ significantly (P < 0.05). Effect sizes (Cohen’s d) represent target student gender differences (no faculty gender differences were significant, all P > 0.14). Positive effect sizes favor male students. Conventional small, medium, and large effect sizes for d are 0.20, 0.50, and 0.80, respectively (51). n(male student condition) = 63, n(female student condition) = 64. ***P < 0.001.

Oy vey!  The ratings of the female applicant were substantially lower than those of the male in every respect.  Means with different subscripts between the tables (e.g., a vs. b) are significantly different, while those with identical subscripts don’t differ significantly.

The salient results:

  • For competence hireability, and willingness to mentor the applicant, women were ranked roughly 25% lower then men.
  • This ranking did not depend on whether the professor who did the rating was male or female, so whatever bias is reflected here is shown by faculty of both genders. To me that is surprising.
  • Male faculty offered female applicants only 88% of the salary offered to males. The disparity was even greater for female professors, who were willing to offer the female applicants only 85% the salary of male applicants.
  • When they did a path analysis, combining “competence” and “salary” into one “composite competence variable,” the authors found that the strongest cause for all the disparities was this: “the female student was less likely to be hired than the identical male because she was viewed as less competent overall.”
  • A separate analysis of the faculty members’ views using something called the Modern Sexism Scale showed that the assessments of female (but not male) competence reflected “preexisting subtle bias” against women.  This supported the authors’ a priori hypothesis that “subtle bias against women would be negatively related to evaluations of the female student, but unrelated to evaluations of the male student.”

The conclusions?

  1. There is gender bias against women—and it’s pretty substantial—at this level of hiring.  This is, of course, in conflict with the results of the Ceci and Williams study mentioned above. It’s possible that once women get past being hired as a faculty member, discrimination lessens substantially, but I am not sure the Ceci and Williams study, being a meta-analysis, is sound.  In addition, every woman I know who is a faculty member in biology, and has discussed the issue with me, says she perceives sexism in the community at some level. Granted, those are anecdotes, but I know a lot of female faculty.
  2. Because the bias is evinced at the student rather than postdoc/faculty stage (if you accept the results of Ceci and Williams), interventions promoting female advancement in science should take place early in the academic career, while one is still an undergraduate.  This could involve, among other things, education of undergraduate advisers about the problem. As the authors note, “Because most students depend on feedback from their environments to calibrate their own worth, faculty’s assessments of students’ competence likely contribute to students’ self-efficacy and goal setting as scientists. which may influence decisions much later in their careers.” This suggests that women may abandon careers in academic science not because of bias manifested after they’re hired, but bias they perceive early in their careers.
  3. The bias against women was manifested equally by both male and female faculty.  This surprised me, but I’ve also been told by women that women are often harder on women than on men (again, anecdotes).

The authors’ conclusion is clear:

The dearth of women within academic science reflects a significant wasted opportunity to benefit from the capabilities of our best potential scientists, whether male or female. Although women have begun to enter some science fields in greater  numbers, their mere increased presence is not evidence of the absence of bias. Rather, some women may persist in academic science despite the damaging effects of unintended gender bias on the part of faculty. Similarly, it is not yet possible to conclude that the preferences for other fields and lifestyle choices that lead many women to leave academic science (even after obtaining advanced degrees) are not themselves influenced by experiences of bias, at least to some degree. To the extent that faculty gender bias impedes women’s full participation in science, it may undercut not only academic meritocracy, but also the expansion of the scientific workforce needed for the next decade’s advancement of national competitiveness.

I have only one beef with this.  I don’t give a hoot whether the USA beats all other nations in the quality and output of its scientists.  That, to me, is a form of chauvinism, and science, being an international venture, should be promoted everywhere. A rising tide lifts all boats. We should try to eliminate gender bias not because it will make the U.S. more competitive, but simply because it’s the right thing to do.


Moss-Racusin, C. A., J. F. Dovidio, V. L. Brescoll, M. J. Graham, and J. Handelsman. 2012. Science faculty’s subtle gender biases favor male students. Proc. Nat. Acad. Sci. USA, published online before print September 17, 2012, doi: 10.1073/pnas.1211286109

57 thoughts on “New study shows gender bias in science against female students

  1. As a hobby-linguist, I wonder if their choices of names made a difference.

    “Jennifer” conjures up a very different mood than “John.”

    I’d be interested to see the results of a similar study that used different names.

    1. “Jennifer” conjures up a very different mood than “John.”

      I must admit, I have no idea what you’re suggesting. Perhaps you’re basing your “mood” upon actual people you know with those names?

      And, while you’re at it, could you suggest some pairs of names that you think would have been less problematic?



      1. ” … Perhaps you’re basing your “mood” upon actual people you know with those names? … ”

        Which is something everyone does. Perhaps a study with many different names.

    2. I’m sure choice of name may make a difference in some cases, IE John vs Tyrone but I’m not convinced names played that much of a factor in this experiment other than the assumed gender identity associated with the name.

    3. That was my first thought. There may have been a name bias in there.

      I wish I could remember the technical name for it. But the concept does exist and is measurable.

      I know it’s especially acute (up to 30-to-1) between ‘white’ and ‘black’ names in resume studies. I know ‘asian’ and ‘hispanic’ names have the same issue, but not as strongly.

      I don’t know about the range within different names within a gender or across genders.

  2. This is depressing. Unlike the recent ban in Iran on women entering certain university courses, this form of bias is so subtle that even the perpetrators are most likely unaware of it.

  3. In absolute agreement with your last para, Jerry.

    Actual evaluation of female employees in academic/scientific settings is problematic– are they really incompetent or are they just so friggin’ psychologically stressed because of perceiving stereotype threat?

    I would add my anecdotes to yours, some women tell me they drop science careers for which they are qualified because the crap is just not worth it. Of course, some women achieve success despite gender bias. Their success does not mean there is not a significant problem though.

    1. This is sad. For me herpetology is a hobby, not a profession, but it is a hobby I take very seriously. Some of the best herpetology papers I have read are written by women. Outside of academic papers in field herpetology discussion forums, they often bring in perspectives that show a fascinating ability to pay attention to detail that men sometimes miss. I don’t want to stereotype because it’s not a hard fast rule, some men (like Dr. Stebbins) were exceptional at paying attention to detail, but it is a general (and anecdotal) observation I have made.

      For the record, despite my handle here, I actually am male.

  4. The observation that female students are viewed as less mentor-able is new to mew to me. And especially depressing. I had no idea that was an issue.

    1. In a way it’s surprising, but it turns out that we all have unconscious biases. Virginia Valian, a psychologist at Hunter College, has written some great stuff on this; see her book “Why So Slow?” or her website about the gender equity project on the college site.
      When I give presentations on this kind of thing (and this paper is only the latest in a long, long string of similar work), I often say half-jokingly that the equivalent bias in men and women addresses my early idea about The Great Male Breakfast Conspiracy. I had thought it was unlikely that all men got up in the morning, had breakfast together, and thought about how to make life difficult for women. But the result was just as if they had.
      Studies like this one show that we all have biases, and it happens without anyone having to engage in a conspiracy, whatever their gender.

    1. I think the way it’s supposed to work in science is that you (provisionally) believe the studies that present the best evidence, whether or not they happen to confirm your prior expectations.

        1. Again, that’s why we have science: to systematize and make transparent the process of observation and analysis, in order to avoid subjective bias. Data trumps anecdote.

          1. Wait, are you insinuating that my real life observations of the sun circling up over the the (flat) earth each day are mistaken?

            1. Are you sure you are not hallucinating this? Or that your eyes are accurately recording the events? Perhaps they are of an unusual shape — would that skew your observations? Where are you located? Why would that matter?

  5. I am SO not surprised by this. I saw numerous examples of gender bias when getting my PhD in the social sciences. Men getting advantageous TA positions or better field positions despite the fact that their female counterparts had better grades or more complete studies. I heard one professor explain this was because he trusted the men to ‘focus’ more in the field, as if the women were too busy putting on their make-up or something. One of my professors admonished a female advisee of his for getting pregnant (she was married) while he never said anything but ‘congrats’ to male students in the same circumstances. When questioned, he brushed this off as his ‘concern’ that she wouldn’t be able to finish the program rather than offering any assistance. She did finish, but wisely switched to a new advisor.

    1. Not surprising, but it is disheartening to hear of this kind of behavior. Especially when it occurs in a context, academic science, where one hopes this kind of behavior would be at a minimum compared to any other contexts. It would be better for all parties, and science in general, if the professor you are talking about were prohibited from any activity that could impact students’ academic or career paths.

      1. Why do people think academia is free from prejudice, bias, violence….? This is the kind of bias that makes folks outside of academia not trust you lot.

        1. Pardon me, but where the hell did I say or imply that academia is free from prejudice, bias, violence, or anything else? And who in the hell is this “lot” that you are grouping me with? I guarantee you I don’t fit in.

        2. Who are these “people” to whom you are referring? One thing is certain: this study was carried out by academics and, assuming its methodology is sound, is likely to help a bit in eliminating bias in the future. That’s makes “our lot” more trustworthy in that we are more likely to change.

  6. That is really disappointing, but not unexpected. For some percentage of people I think a program to promote awareness of this problem will probably have a fairly immediate positive effect. But for many I don’t think it will make any difference what is done. We’ll just have to wait until they retire to be rid of their obstruction.

  7. Thanks for reporting this. I have to add I’m not at all surprised women showed bias as well. If our culture keeps saying women aren’t good enough, of course many women will start to believe it.

  8. I find none of this surprising as it comports perfectly with all my anecdotal “evidence.” Women are treated differently. I am disappointed by the findings re women PIs but should not be surprised as institutional biases do behave this way. See evidence of racial profiling by minority law enforcement. I do need to look more closely at the study as I wonder, given my own completely unscientific collection of data, if the bias is more pronounced in fields like Chemistry and Physics – where the women seem to disappear, as opposed to biology, say. (I was once told by a Cambridge Professor that the reason there were so few women in chemistry is that “girls find chemistry boring.” Or perhaps they find being shut out a tad dull.

  9. The numbers say that the female faculty not only ranked female candidates lower than the male faculty did, but also ranked male candidates higher than the male faculty did.

    The “Modern Sexism Scale” bit makes me a little leery. I did a search on that term and found a short survey with very much less than stellar questions which don’t measure sexism at all. Simply adding this kind of thing to the study can skew the results.

    So I have some reservations about the results (I’m getting a vibe about the researchers looking for a certain conclusion), but their implications are quite unwelcome.

    The important stat is competence rating. The other two are susceptible to a particular prejudice for which a reasonable argument can be made (concerning investment of resources into someone more likely to cut short a career to raise children). But given identical information, competence should have been rated the same.

    1. “The other two are susceptible to a particular prejudice for which a reasonable argument can be made (concerning investment of resources into someone more likely to cut short a career to raise children).”

      I don’t think that is a reasonable argument at all. I think that in most cases that argument is a rationalization used to support a position that is already held. I think the frequency of women cutting short their careers in order to have children, and the impact it would have on their university or other employer, are both greatly overestimated by those using this argument. By now it is like an urban myth that is just accepted by some people without thinking much about whether there is any good reason to suppose that it is true.

      1. That young women leave careers early to have children is not a myth. It’s an observed fact. It’s probably less common today than it was 20 or 30 years ago, but it’s a fact that people in a position to hire someone are not unreasonable to consider.

        As far as the actual impact on the hiring institution when it does happen, you sound very much like you have no idea what it would be. You “think” a lot of things for which you seem to have no grounding.

        That said, it’s also a fact that I personally don’t want to live in a society where women are treated as baby-making machines. So I think that hiring young women is a risk worth taking to ensure that more women enter the workforce. I just can’t justify demanding that everyone else be willing to take the same risk.

        1. Yes, it is unreasonable to consider that, because people leave careers for a variety of reasons, and many women don’t have babies. Also, “modern” men are much more likely to take leaves for family reason than old-fashioned men are.

          It’s also unreasonable and illegal not to hire someone who is 10 years from retirement vs. someone who is 30 years from retirement for those same reasons. You never know who will really stay and young people are much more likely to make a career change after a few years.

          1. Besides which, it takes a hell of a lot of nerve for an employer to not hire somebody who might not be there in ten years when the employer is quite likely to “right-size” the workforce in much less than ten years to pad investor quarterly profits.

            I could maybe start to perhaps have some sympathy for the “But she might get preggers and then what’ll we do?” wolf cries if employees could still count on retiring at 65 with a party, a pension, and a gold watch. But those days are so laughably long gone that it’s ludicrous that we’re even having this discussion.

            And, for some reason, I seem to have an irresistible urge to quote Monty Python to anti-woman employers: “Stupid git!”



        2. ” You “think” a lot of things for which you seem to have no grounding.”

          For much of my working life I have been responsible, in part or in whole, for determining the need for, desired characteristics of, and vetting of, prospective new employees. I would not claim to be an expert on this issue, but I do have a good bit of direct experience with it.

          Having said that, I share your sentiments regarding the type of society that would be preferable. Could I offer you a fine glass of wine?

  10. It’s not just in male-dominated professions, either. I’m a librarian and I have often spoken up in a meeting only to have my idea ignored or dismisseed then a man floats the exact same idea and it’s brilliant. And both men and women react the same way. I’ve seen other women get the same treatment. And a man can say the stupidest thing and people pay attention as if “well once he gets past the stupid part he’ll get to the brilliant part” but women get cut off if we run on too long without saying something briliant. It’s more at the management level, but it’s still shocking. You’d think that women wouldn’t do that to each other but apparently we do.

  11. There’s one thing that could be done immediately to help eliminate this bias.

    Applications for posts should be in two parts. One part has the identifying personal information, and the other with has the information relevant to the application, with only a number reference to link it with the other. The person evaluating the application should only see the relevant info.

    Of course, this only helps in eliminating bias in selection of applicants for interview, but is an obvious and easy first step.

  12. The data are consistent with bias against women. Is gender bias the only possible explanation, though? No.
    We don’t know how John and Jennifer stacked up against other men/women respectively, who applied for such positions in the past.
    Hypothesis: What if John was better qualified than the other guys, whereas Jennifer was less qualified than the other women? i.e., what if the pop. of women tends to have better qualifications than did Jenny? Then the choice of John over Jenny would be evidence of the profs viewing women as generally better qualified than the men, women profs being even more aware of that.
    (But why would the profs want to compare men to men and women to women? many possible reasons, some of which don’t quite qualify as sexist–for example, perhaps a reflexive tendency to compare a candidate to the pop. they are believed to come from, for the sake of fairness.)

    The above is not to suggest for a moment that sexism in science is not a problem. It is. I am just noting that this particular study, while suggestive of sexism, does not quite prove it. And it is important that women especially be able and willing to recognize this, else they will fit the pernicious stereotype of women as marginal scientists.

    1. No, that would still be sexism. You see, John and Jennifer have identical qualifications. Yet on your hypothesis, John is ONLY compared to other male applicants and Jennifer is ONLY compared to other female applicants.

      Now, how could this possibly be fair? Surely you’d want the best applicants, regardless of gender. This is still a case of gender segregation. It really doesn’t matter whether the selectors are conscious of their biases or not; biases are all the more pernicious for being invisible to the people holding them.

      So whichever way you cut it, it looks very much like sexism.

      1. Also, while there might be some cases where a professor had a particularly good experience with a John and therefore is partial to the name, really enough that there would be statistically significant bias towards John vs Jennifer given the sample size?

        Both are very common names so faculty have likely had several students good and bad of both names. I suspect that is one reason why such common names were used, but I haven’t actually read the paper.

  13. I find it odd they didnt break down the data further.

    What were the stats in bio vs physics? We get the mean age of the faculty, but why not break down the data by faculty age, since that info was collected? And they had faculty position (assistant prof, full prof, etc), but didnt separate the data out that way either.

    What was the list of Unis? Were they biased towards the coasts? Why only ‘large research institutions’?

    It just seems like they collected a lot of data, and we got a very simplistic presentation of that data, especially for PNAS.

Leave a Reply