A few days a go, the New York Times published an Op-Ed by two psychology professors who argue that “Academic Science Isn’t Sexist.” On STEM Women, I look at the various methodological problems with the Op-Ed which is based on a review study conducted by the Op-Ed authors and two economists. The biggest issue is that the way they measure gender inequality does not match the data they have available. The researchers fail to account for institutional factors that impact on women’s under-representation in Science, Technology, Engineering and Mathematics (STEM).
STEM workers, like all people, are impacted by socio-economic issues. The authors of the study have failed to account for evidence of how parents, teachers, the media and other social influences discourage girls from STEM. Studies show girls and boys perform equally well in STEM-related tests, but stereotypes, lack of role models, and discrimination make it harder for women to succeed.
The researchers have also failed to account for issues of race, sexuality, disability and other forms of discrimination that further disadvantage minority women. I have shown why the theory of intersectionality is important in any analysis of why minority women are less likely to succeed in STEM. In a study headed by a White male professor who already has tenure, and three White women professors, the omission of race from the analysis is especially problematic. I show that any study that argues sexism is a myth, and which serves both White privilege and male privilege, must be read with extreme caution.
A couple of other highlights:
- The main argument in Op-Ed is that women would fare well in maths-intensive subjects, “if they choose to enter these fields in the first place.” This ignores the fact that individual choice is constrained by the institutional barriers girls and women face in their STEM education and careers.
- The study argues women academics are not as “productive” as men. The researchers measure this in terms of publications. They ignore research showing women are more likely to do additional teaching, admin and other duties at work, as well as more childcare at home, while male colleagues are more likely to be able to focus on research uninterrupted by these extra duties.
- The researchers argue that there are less women in STEM because they are choosing to “opt out” as they have children. They explain pay discrepancies by saying women choose to pay a “child salary penalty.” The researchers ignore evidence that women actually leave more for institutional reasons, such as sexism in the workplace. Research also shows male academics are actually fare better in their careers if they are married with children. I show how sociological research on family and inequality need to be taken into consideration when assessing the career trajectories of scientists.
More on STEM Women.
I want to put this story in broader context below and include a more detailed look at the data the researchers present. The purpose is two-fold. First, to further document how social science research can be used to serve an agenda. Ceci and colleagues do not see sexism because they are using individual explanations for the disparity in men and women’s outcomes in academia. Using a sociological approach, which places individual experiences in social context, I show why institutional forces matter.
Second, I want to make a sociological argument about why the researchers draw on an individual narrative. They claim that women are “opting out” of academia, and that if they were willing to work harder, and if their interest in STEM were stronger, women would flourish. Ceci’s research team is made up of two psychologists and two economists. I want to tease out why they draw these conclusions given their disciplinary perspectives.
Herbert Gans noted in his 1988 Presidential Address to the American Sociological Association that very few sociology books gain the media’s attention. He was speaking about the American context of course, but with bemusement, he points out that sociologists don’t really concern themselves with writing for popular audiences. He makes the comparison to psychology and economics, which are two social science fields that have a longer tradition in pop science. He writes:
Good nonprofessional sociology is useful to us for the same reasons as good popularisation. We shave a special interest in reducing bad pop sociology, however, because its low quality can reflect on us directly and quickly since the general public may not distinguish between professional and nonprofessional sociology.
Sociologists in America, Australia and elsewhere do, of course, engage with popular media, but perhaps not as much as the other social sciences. The social sciences in general are less lauded in popular media than the life sciences, which contributes to poor social science literacy by the public. Sociology tends to make news when we talk about gender and sexuality. These are two central concepts that have dominated sociological enterprise, but they are obviously not the only social issues we examine. Gender is polarising: people either think that the way things are today as they know it is perfectly fine and natural, while others believe gender relations need to change. So perhaps it is unsurprising that when social science studies about gender hit the news, they strike a chord with readers and the media alike.
This year alone, there was a study about how people perceive the severity of hurricanes based on whether they are named after a man or a woman. This study was largely reported on in sexist terms, or was otherwise criticised for being overly simplistic. There was another study by a male sociologist who argued that women are at greater risk of domestic violence if they are unmarried, which was of course a sexist and incorrect reading of the literature. The problem in both cases was that social scientists failed to communicate the complexity of these issues, and the media also ran with an overly simplistic reporting of the findings, appealing to the most basic, and lazy, gender stereotypes. Writing for Slate, psychology PhD student Jane Hu called out poor reporting of social science research:
“By reinforcing gender stereotypes and roles, the media affects people’s perception of women. Study after study has shown that pre-teens and teens learn about gender and sex from the media, and adults, too, subconsciously take those messages to heart. Gender stereotypes disrupt girls’ behaviour on social media sites, limit women’s perceived possible career paths, and bias women’s chances of winning an election…Writers, we need to stay vigilant and look beyond the easy gender narratives. Readers deserve better.”
Gender issues in life sciences publishing are similarly problematic. Richard Dawkin has launched a series of sexist tirades targeting women, downplaying the experience of rape and decrying the existence of “radical feminism” (a term that is about 30 years out of date). There have been a series of attacks on women who work in tech as well as women like Anita Sarkeesian, who uses social science to critique sexism in the tech industry.
My colleagues and I at STEM Women have covered a few other issues. This year has seen a scientific journal publish a “graphical abstract” that involved a virtually topless woman holding coconuts over her breasts, for a study about proteomics. ScienceAlert published a headless shot focused on a woman’s breasts for a story about caffeine shaker, which they presented as a science story. (It wasn’t.) Science Magazine ran a special on AIDS and used the headless photo of transgender sex workers which dehumanised both transgender women and sex workers, as well as evoking transphobic fear. All of these are examples of sexism in science, demonstrating that the prevailing culture is still one which is exclusionary and hurtful to women, and doubly reinforces the otherness of transgender women. These are also examples of how gender issues in general are still reported on poorly by some scientists. The media is more likely to pick up these sexist examples of gender and social science, while other excellent social science studies, on gender and other subjects, barely rate a mention.
The New York Times Op-Ed by Williams and Ceci sparked outrage amongst scientists. Some scientists on social media were outraged that this study was specifically produced by social scientists, and I was bemused to see at least one tweet by a life scientist about how the social sciences are biased. I have shown that the public’s willingness to accept scientific evidence is influenced by beliefs and values, and that scientists can also fall prey to the same short-comings. It should come as no surprise that social scientists can also be guided by their political position.
In the video below, Ceci and Williams explain that as long-time collaborators, they decided it was time to tackle the literature on sexism in academia, perceiving that the prevailing argument on sexism was incorrect. They describe wanting to form a multidisciplinary team, and elected to work with economists. Both disciplines are broadly concerned with individual patterns. (There are sub-fields within each discipline that look at other levels of analysis.) Sociology is set up to do the opposite: we connect individual biography to history and culture. Sociology, studies of intersectionality, and other theories that question the centrality of Whiteness in academia, and seeks to challenge other issues of social privilege established by Western scholarship traditions. These critiques have a bearing on the study headed by Ceci, a White, male tenured academic. In my article for STEM Women, I have noted that the study’s methodology and analysis serve both White privilege and male privilege. In the first instance, the data illustrates gender inequality but the authors reject this as evidence of sexism in academia. In the second instance, a team of tenured professors led by a White male who reject sexism exist are actually perpetuating sexism. Elsewhere, I’ve shown that sexism describes not only overt abuse and discrimination, but also the everyday interactions and unexamined processes that disadvantage women.
Bias in interpretation of data
The paper begins by drawing on 2010 data by the National Science Foundation’s, or NSF’s 2010 Survey of Doctorate Recipients. The authors’ calculations of this dataset show women are under-represented in maths-intensive fields, which the authors define as Geoscience, Engineering, Economics, Mathematics/Computer Science, and the Physical Sciences (GEEMP). Conversely, the researchers find that there has been greater growth amongst women academics in the fields classified as being less reliant on maths – Life Science, Psychology, and Social Science (LPS). It’s crucial to note that across all disciplines, with the exception of junior Life and Social Sciences faculty, women are still a minority.
The authors do not adequately address this discrepancy between the GEEMP and LPS growth. Their entire argument that sexism does not exist in science is predicated on the simple broad trend that women’s absolute numbers have been growing in STEM since the early 1970s. To put it another way, the authors argue that because there are more women in STEM today than ever before, this means that equality is growing, and so they say that inequality is a thing of the past. This position is not faithful to the literature reviewed, nor does it reflect the data presented.
The authors’ analysis of the NSF data shows that women make up around half of all high school graduates in the USA. The proportion of women Bachelor graduates, PhD recipients and Assistant Professors has grown all fields, but growth has been more pronounced in the Life and Social Sciences. For example, while around 58% of Life and Social Scinece graduates with PhDs are women, only 26% of their maths-intense counterparts are women. Between 25% to 44% of GEEMP Assistant Professors are women and 7% to 16% are full professors. In LPS, women fare better, with 66% of tenure-track Assistant Professors in psychology being women and 35% full professors; 45% of assistant professors in social science (excluding economics) are women and 23% of full professors; and 38% of assistant professors in life science are women and 24% full professors. Despite these numbers clearly showing that women are the minority in all fields with the exception of junior academics in Social Science, the authors begin their analysis by claiming that data on gender inequality is “contradictory.”
There is no contradiction. Women are clearly disadvantaged.
The paper states that women were less likely to be in senior management as deans, directors, or department chairs in top tier research universities (“R1”), but that they were “equally likely to be presidents, provosts, and chancellors.” Data on non-R1 leadership was not discussed. Instead, from these findings, the authors conclude that inequality is about women’s choices, not due to systemic inequality.
“Another way to think of this is that far fewer women are interested in (or perhaps capable in, as we discuss below) GEEMP fields to begin with, but once women are within GEEMP fields, their progress resembles that of male GEEMP majors.”
The researchers consider several explanations for the discrepancy in academic retention of women. I summarise their analysis under sections roughly corresponding to original study before summarising the overarching gaps in the study:
- Biology and childhood
- Work “Productivity”
- Children and Marraige
- Funding and Pay
- Job Satisfaction
Let’s look at each of these in turn.
Biology and childhood
The authors spend many pages at the beginning of their article discussing biological research which they say cover “in utero through high school.” This consists of studies about twin development and spatial ability. Their analysis notes that mathematical abilities cannot be conclusively tied to foetal development or biologically-driven differences between boys and girls.
If their data do not support biology, why bring it up? The logic is hard to follow, but certainly they present no evidence to back up the claim in their Op Ed:
“As children, girls tend to show more interest in living things (such as people and animals), while boys tend to prefer playing with machines and building things. As adolescents, girls express less interest in careers like engineering and computer science.”
Later, they show that this expression of interest is not the mere outcome of biological or innate curiosity. The data show it’s culture.
The study details numerous studies showing there are no gender differences between girls and boys’ mathematical abilities, regardless of the level of difficulty of the test.
The authors show there is some gender variation in the “extreme right tail” amongst the top performers in maths tests especially in elite maths competitions. The researchers argue that the girls who are in the “extreme right tail” are not as gifted as the boys. They claim the scores that get a girl into the top 5% of maths achievers would correspond to the top 10% of the male distribution, and similarly for the top 10% girls whom they say is the equivalent of the top 20% of boys.
My STEM Women colleagues and I have dispelled the myth that girls can’t do math. The scientific evidence shows that boys receive more coaching and encouragement by parents and teachers, while girls are discouraged from developing an interest in math. This coupled with Cecil and colleagues’ earlier data showing that girls do indeed perform as well as boys in maths tests actually suggest that the reason there are more boys in the “right tail” is because they receive more support to get them there.
Ceci and colleagues show that, over time, gender gaps between men and women pursuing higher education degrees in STEM have been closing, but there are variations amongst different disciplines.
Ceci’s team data show that more male students in their early to mid-30s were in the paid workforce (96%) compared to women (84%), and over two-thirds of working men (37%) already had the job title of “scientist” or “engineer” (37% of men) while only 18% for women did the same. This tells us that male students are already finding work in STEM and more perhaps more importantly that there’s a gendered dimension as to why this is the case. Dr Mewburn shows that male postgraduate students, for example, get much more hands-on help from their supervisors, including on finding work.
The researchers present studies from the mid-1990s to the early 200s to show that the proportion of women applicants to some STEM jobs is disproportionate of what “would be predicted by their fraction of the applicant pool.” Cecil and colleagues think this must reflect how equal things are in academia. Bear in mind though, that few of these women will actually be hired, and even less will progress though to senior ranks.
Ceci and colleagues systematically reject various studies showing that women undergraduate students are less favoured for lab manager posts and summer jobs, and that they’re downgraded or kept out of certain activities, such as computational tasks. They say this is not evidence of hiring bias, despite the fact that internships, lab posts, and other experience are exactly the types of things scientists need to do to try to get a full-time job
Ceci’s team dismiss the significance of a recent study showing elite male in Biology faculty are less willing to take on women students saying this is not necessarily evidence of gender bias in hiring. We are meant to accept that the senior researchers who are less willing to mentor junior women in summer jobs, the faculty staff who downrate women’s teaching and research skills, are not part of the hiring pool.
The application process in science obscures how each step of selection is riddled with institutional bias. I have made this point with respect to fellowship programs, but similar principles apply to job seeking. If hiring committees do not control for the institutional factors that prohibit women from progressing successfully at various stages of their career, we will be waiting a long time to see greater gender equality.
Using NSF data from 2008, Ceci and colleagues find that male Assistant Professors on average published 2.1 articles more than their women counterparts in the preceding five years. The publication gap is wider amongst full professors, with men publishing 2.8 more papers than women. Despite this evidence of disparity, the authors argue that because more women are publishing in STEM since the 1970s, this is evidence that gender bias has been eliminated in STEM, even though they note the increase is not statistically significant.
Ceci and colleagues do not make a distinction between single and multi-author papers, even though other research shows women are less likely to be first author on a science papers, which puts women at a disadvantage matters to promotions.
Women in some fields publish relatively less academic papers than women in other fields. Ceci and colleagues muse that this is because women are “less likely to specialise in their research topics.” Elsewhere they argue that women may prefer to teach more, even though other studies they cite show women academics are overburdened with teaching, pastoral care of students, admin duties, and additional work with which male academics are not lumbered.
The researchers’ analysis of NSF data from 2010 show that women and men work similar hours in academia, and in maths and computer science, childfree women will actually work more hours than men. Due to the variation in publication and time amongst different disciplines, the researchers conclude that time is not a factor leading to reduced research output for women.
Ceci and colleagues argue that the publications system is “gender neutral.” Ceci’s team’s conclusion does not align with other evidence that women are disadvantaged, as the academic system tends to assign less prestige to the journals and research fields in which women are more concentrated.
Childcare and marriage
Ceci and colleagues draw on highly outdated data from 1978 to 1995 to argue that having children does not impact on the productivity of women. Later they go against this idea. Going back to NSF data from 2008, Ceci and colleagues argue that there are no “significant” gender differences in publication output between early career researchers who are childless. They also find that men with children publish more than than childless men. Together, these findings are used to argue the fact that mothers in academia choose to be less “productive.”
Ceci’s team draw on studies published in 2007 which suggest that women may be unable to keep up with the “fast paced” nature of work in maths-intensive fields, which they say demands at least 40 hours per week. The researchers cite evidence that women experience difficulties managing work and family demands on top of having to do additional pastoral care with students and other services and activities that their male colleagues were not expected to manage.
The studies Ceci and colleagues draw on show that while men are more likely to leave STEM careers because of pay, remuneration or location issues (95% of men versus 85% of women), women are three times more likely due to family concerns (15% of women versus 5% of men). Still, family reasons constituted a relatively small reason for all STEM workers who leave academia. I will show later on that it is not family alone that may lead women to swap their STEM career, but rather institutional sexism.
Funding and pay
Ceci’s team report that various studies from the 2000s show that men and women have similar rates of funding success. They see this is evidence of egalitarianism. Yet they note that men are awarded more money for projects on average. They explain this by saying men are more likely to be principal investigators overall, as well as more likely to head big projects. This is yet another example of gender disparity, but Ceci and colleagues refuse to engage with this as evidence of academic inequality.
Similarly, the authors conclude that women are not disadvantaged when it comes to promotions. This is despite the fact that another NSF study (from 2004) finds that controlling for various socio-economic factors, women are disadvantaged when it comes to finding a tenure track position, as well as getting tenured, and also in being promoted to full professor.
There are variations in pay and promotions amongst different disciplines. The biggest downward trend has been in Economics: in 1995, women Professors earned 95% of what men did, but in 2010, women Economists earn 75% of their male counterparts. Ceci and colleagues explain this not through gender inequality, but again through individual choice of women; that is, their “preferences, productivity, job matching, and negotiation.” The authors argue that mothers who are Economists choose to pay the “child salary penalty.” In the author’s words, these academic women are “electing” to stay home with children instead of continuing their career. This is not an individual choice, of course, it is an outcome of institutional conditions that make it difficult for working mothers to return to work.
The researchers apply this logic more broadly. While they have no data to back up their conclusions, Ceci and colleagues argue that women are “electing” to drop out of academia.
The researchers further argue that women are choosing to “opt out” of academia because they are unhappy with STEM work. The researcher’s analysis of NSF data show that in 1997, women academics were generally less satisfied than men with their jobs, whereas in 2010, men and women’s job satisfaction was at similar levels. Ceci and colleagues cite various other studies showing that there are uneven levels of happiness in different disciplines, especially in the Life and Social Sciences. A 2004 study argues:
“Specifically, females were significantly less satisfied than males with the commitment of their department chair and senior faculty to their success, and they were dissatisfied with their professional interactions with senior colleagues, leading to dissatisfaction with how well they fit in in their department.”
This would suggest a sexist culture within academia contributes to whatever dissatisfaction women feel about STEM, not their STEM work itself.
Ceci and colleagues draw on a 2008 study using NSF data, which finds women in STEM felt “isolated and had fewer role models and lower job satisfaction.” Ceci and colleagues cite various studies from the 1990s through to the mid-2000s showing that “dissatisfaction with research support and advancement” leads to women to change their careers. While it is obvious, I draw attention to the fact that here we have data contradicting the authors’ NYT Op Ed premise that there is no sexism in STEM.
As you see, there are many extrapoloations and inaccuraries in the way the researchers have interpreted their data. Sociology and related fields are set up to critically evaluate social phenomena. We also address social problems in terms of systemic disadvantages. It pays to be better informed about how to read social science in popular media, so that political agendas that aim to uphold the status quo are not accepted as neutral science.
An earlier vesion of this post was published on Google+ and my Tumblr.