Have you ever been in a meeting where a colleague says “I’m a great supporter of gender equality, but I’m totally opposed to quotas!” Or, “I believe in diversity, but I won’t stand for positive discrimination.” Maybe you felt a bit troubled by such statements, thinking: that sounds fair, but somehow I don’t think it is… how do I rebut this?
Bias is omnipresent in our society, and some of us are keenly aware of rampant bias in sectors like technology, engineering and politics. Efforts to thwart the effects of bias in communities and institutions prompt a spectrum of diversity initiatives. Many times these lead to backlash. It’s been just a year since the memo “Google’s Ideological Echo Chamber” spread through the company’s internal channels, then became public. Yet, another wrangle is already blasting online with the article “Why Women Don’t Code,” by a university lecturer. What do we do when privileged individuals continue to turn a blind eye on the injustices around them? They insist on points like “women are less likely to choose computer science,” and that it’s just due to natural differences.
Community and institution leaders have much of the power to catalyze change, but in the meantime, I want us to build, share and refine a portfolio of rebuttals for the recurring themes of the anti-diversity manifestos. Let’s get started.
Facts on bias and discrimination
On the wake of a university lecturer’s op-ed on “Why Women Don’t Code,” it’s pertinent to consider the study that shows how “Science faculty’s subtle gender biases favor male students” (Ref. 1). In this randomized, double-blind study, faculty were presented with the application materials of a student with male or female name. Evaluating the applicants for a lab position, the faculty members rated male applicants as more competent, assigned them higher starting salaries, and offered them more mentoring — with identical application materials. Male and female faculty were equally biased against the female student applicant. The study involved 127 professors in biology, chemistry and physics, across the country. They rated student applications using a validated scale for competence and employability, and selected a starting salary. The mean starting salary offered to the female student was on average $3,730 lower than that offered to the male student (12%). The female student was also less likely to be hired, being viewed as less competent. Faculty gender, age, field, or tenure status were not factors, suggesting that their bias was unintentional and influenced by cultural stereotypes, rather than conscious sexism.
FACT — Empirical evidence shows that science faculty discriminate against female students.
Students, on the other hand, are biased against female professors. Universities, for the most part, use one tool to assess teaching effectiveness for faculty tenure and promotion decisions: student course evaluations. These surveys not only fail to reflect teaching effectiveness, but are significantly biased against female instructors. A recent study (Ref. 2) analyzed datasets from a natural experiment in a French university, and a randomized, controlled experiment in a US university. The first dataset included surveys of six first-year courses by 4,423 students, over five years, and the second set included 43 surveys for 4 sections of an online course. Using advanced statistical analyses (nonparametric permutation tests, and statistical tests of significance), the authors found a weak and statistically insignificant association between the student rankings of instructors and their performance in an anonymously graded exam. In other words, the student ratings don’t reflect teaching effectiveness. They do, however, show a statistically significant association with instructor gender: both the French and US students favor male instructors. In the US experiment, the students were randomly assigned to four discussion groups led by two teaching assistants: one male and one female. The teaching assistants switched names in one pair of discussion groups. The data shows that students rate instructors significantly higher when they perceive them to be male — even in “objective” attributes, such as promptness.
FACT — Dozens of studies show that student teaching evaluations are biased against female instructors.
(I’ve shared a list of more than two dozen references on gender bias in teaching evaluations.)
Students also underestimate the academic performance of their female classmates. A study in a college-level biology course (a subject with equal enrollments by gender) showed that male students are more likely to be named by their peers as being knowledgeable on the course topic (Ref. 3). The study used longitudinal social network analyses, and found that the peer-perception effect persists after controlling for class performance and students’ outspoken personality. The bias specifically reflected male students over-estimating other male students, by an amount corresponding to a 0.57 grade difference in a 4-point scale. The authors concluded that uneven peer perceptions by gender are a social factor that could negatively affect women’s professional confidence, contributing to attrition from STEM careers.
Biased peer assessments continue to plague women as they progress in STEM careers. Peer review of publications and conference submissions, grant-proposal reviews, and even code review and acceptance of pull requests in open-source software, are all rife with implicit bias against women.
- Gender bias found in how scholars review scientific studies, Science Daily (2013)
- The great citation hoax, blog post by Prof. Curt Rice (2013)
- Women considered better coders — but only if they hide their gender, The Guardian (2016)
- Pull request acceptance of women versus men, PeerJ PrePrints (2016)
- Facebook’s Female Engineers Claim Gender Bias, The Wall Street Journal (2017) — female engineers received 35% more rejections of their code than men.
Women are underrepresented in the #peerreview process & all editors apply same-gender preference https://t.co/sXaKYUKfnE #genderbias
men have statistically significant greater odds of receiving grants than women by about 7%" https://t.co/Sokhx1Wafv #everydaysexism
Advice for women entering academe: become better informed about the intellectual and social isolation you are likely to encounter.-J Winkler
FACT — Peer groups at all levels of science and technology show implicit bias against women.
Occupational gender segregation
Fewer than 1 in 5 women worked for wages at the turn of the 20th Century. By 1940, 27% of American women older than 16 were working for wages, rising to 55% by 1986. (Census data cited in p.9 of Ref. 4.) The figure was 57% in 2017. Women and men have always been segregated by occupation, at levels that remained stable between 1900 and 1970. Aggregate level of occupational sex segregation decreased in the 1970s, as women integrated in jobs previously dominated by men (p. 16, Ref. 4). The percentage of women in the labor force as a whole increased from 38 to 42.6% between 1970 and 1980. Nearly three dozen occupations saw disproportionate increase in the representation of women, in a variety of sectors: bartenders and bus drivers, in the service sector; buyers and accountants, in the administrative sector; public-relations professionals; advertising specialists… and notably, computer operators, which were categorized as “clerical occupations” (p.17, Ref. 4).
Most employed women in the 1970s were in clerical, service or semi-professional jobs. Computing occupations were a rare opportunity for technical and professional work. The sector was vertically segregated, however, with men better represented in higher-paid and more prestigious roles. Although 59% of “computer operators” were women in 1980, their representation was high in data entry (92%), and low in computer programming (31%) and systems analysis (23%) (p.170, Ref. 4; also, Ref. 5). Besides sex segregation by rank, salaries of computer operators rose more slowly than inflation in the 1970s, leading to more men leaving the field. This, coupled with strong expansion of the sector leading to personnel shortages, explains the growth in women’s representation in computing jobs during the ‘70s.
As the field professionalized, the participation of women in Computer Science increased steadily, until peaking at about 37% in 1984… then falling (18% in 2014). Some have tried to explain this as an effect of families purchasing early personal computers for boys, but not for girls, resulting in a gender skills gap widening during secondary education, and a hardening of stereotypes. A group at Stanford reporting on their efforts to increase the number of women in the major pointed to another chain of events (Ref. 6): universities took steps to reduce enrollments, due to a lack of faculty. Admission became more competitive, mathematics requirements increased, and introductory courses took on the dreaded “weed-out” hallmark. All these steps disproportionally discourage women and minorities.
Intrinsic sex differences in ability
The latest study we have available on the question of differences in ability was published this year (Ref. 7). It looked at young children under the age of 8, and found no statistical differences in mathematical ability between girls and boys. Lack of significant difference is insufficient to conclude that the two groups are statistically equivalent, but this study (unlike predecessors) also included tests of equivalence in performance, and also of statistical difference in variability. The latter is important in light of previous findings of greater variability in boys’ performance, leading to more high-performing and low-performing males than females (at the tails of the distribution). This new longitudinal study, involving more than 500 infants and children, found no differences in early mathematical ability, in any measure.
Adult differences in perceived ability or developed skill in science and mathematics are a complex interplay of sociocultural effects, and cannot be ascribed to intrinsic or biological differences. Cultural stereotypes, parental guidance, teacher encouragement, societal expectations, later-life discrimination, all strongly shape abilities and performance in STEM.
Diversity & Inclusion
“I am completely opposed to any positive discrimination.” — email from a conference organizer, when I pointed out their plenary speakers were all men (2015).
“I’m firmly opposed to quotas” — a colleague responding face-to-face to my talk at a faculty retreat, titled “Engineering Gender Balance” (2016).
Diversity and inclusion are unequivocal concerns across academia and industry. At the very least, institutions today are expected to conduct implicit-bias training of key groups (like hiring committees at academic departments, or industry management), and compile diversity reports (on the basis of internal surveys of climate and demographics). Beyond that, any initiative approaching affirmative action is contentious. And, sadly, while leaders at many institutions are not really committed and just want to “check the box,” numerous individuals (often privileged) hold strong beliefs about diversity and inclusion, without a basic understanding of the social-justice concepts underpinning it all.
Implicit-bias has to be understood as a social-justice issue first, before any training exercise can effect change. Unfortunately, though these training events do raise awareness of implicit bias, they may not be effective at changing behavior. In fact, implicit-bias trainings can backfire, and sometimes may lack a thoughtful and tested curriculum or use dubious methods. The infamous author of the Google anti-diversity manifesto did, after all, attend a diversity program at the company, which exasperated him enough to write the memo as a response.
Most trainings have not been systematically evaluated (a few exceptions); the available evidence suggests that they can change awareness of bias & motivation to address, but not the actual behavior. See, e.g, @JenniferJoyGaba's dissertation from my lab. https://t.co/srhtVg55aU
Now a word about quotas… such a misunderstood intervention! Almost always, the response to the mere mention of gender quotas is an argument about meritocracy: “We can’t lower the bar!” Let’s clear this up: merit is a fallacy. Education, hiring, promotion, pay, elected positions — none of these are meritocratic. Prof. Rainbow Murray said it best in her article on gender quotas in politics, for the blog of the London School of Economics: judgements on “quality” and “merit” tend to be based on the status quo, and the opponents of quotas are implicitly arguing for preserving this status quo. Objective measures of merit simply don’t exist. What does exist is research evidence that quotas raise quality, contrary to common wisdom. Think about it: if implicit bias prevents us from making objective assessments of the capabilities and work of others, and equally qualified men and women are perceived differently, it means that correcting gender bias raises the bar.
The evidence: one study published in Science (Ref. 8) found that competition enhanced the performance of boys, while it reduced the performance of girls. The hypothesis follows that competition may lead to reduced participation by women — a hypothesis tested in another study (Ref. 9), where affirmative action changed the applicant pool by increased participation of high-performing women. Curt Rice explained these two studies and their implications: affirmative action in a competition scenario (think of hiring!) increases the talent in a participant group as more highly-qualified women enter the competition. Another study in Sweden (Ref. 10) looked at zipper quotas in municipal politics (where parties alternated between men and women on the ballot), and found that this measure raised the quality of the men who were elected, and the quality of the group overall. Curt Rice’s TL;DR: mediocre men were replaced by more highly qualified women.
The anti-quota argument is that they are unfair, but what they are is an intervention to fix unfair situations, and they raise overall quality.
“A Girls’ Night Out Dance Party” — sounds like fun, no? In August 2017, this event in Oceanside, California, sold out weeks in advance. A man showed up at the door, however, and after he was denied entry, he sued. The man has filed more than two dozen lawsuits for discrimination (NBC7 news). In fact, a series of lawsuits in California are weaponizing the Unruh Civil Rights Act to silence women’s empowerment initiatives. It’s a modus operandi of the members of the National Coalition for Men (NCFM), for example. Last year, the company Ladies Get Paid, which organizes conferences, workshops and coaching events, had to settle a sex discrimination lawsuit to avoid crippling legal fees, after a man turned up at an event, got turned away, then sued.
LadiesGetPaid was sued by a group of men for gender-discrimination. Contribute to our crowdfunding campaign to keep us in business. https://t.co/68kOeuVLNa #equalpay
That’s the extreme end of the backlash against gender-equality interventions and initives. On the subtle side is, simply, denial. Particularly within my sphere, it’s most concerning that male faculty just don’t really believe bias research. Empirical studies with groups from the general public and university faculty (Ref. 11) found that men evaluate the quality of research on bias as lower than do women. Despite their scientific training, on the face of empirical evidence about gender bias, they refuse to be convinced by it. The worse part: men faculty in STEM showed greater difference with women in their assessment of this research.
Men faculty in STEM remain skeptical of implicit-bias research, even when faced with evidence.
“Women generally … have a stronger interest in people rather than things, relative to men” —Damore.
“Women simply chose to pursue other interests” —Reges.
The suggestion that occupational gender segregation reflects the “choices” of women often comes with the premise that fundamental gender differences lead to those choices. If that were the case, we would not see examples of rapid desegregation of certain occupations, like experienced in medicine. The fraction of medical-school graduates who were women rose rapidly, from 16% in 1975 to 40% in 1995, and today medicine has gender parity. Women’s interest in this profession changed. Perhaps the medical system itself changed to welcome more women. People’s interests and choices are not innate; they have their origins in culture and society. Andy Ko’s response to Reges points to “scientific evidence showing that the culture of a learning community is what determines their interest joining it” (citing Ref. 13). The science of human development has established that people don’t choose their interests and occupation freely. Experience, economic opportunities, social expectations, parental guidance, and cultural pressures all govern the choices people make.
“Women are underrepresented among roofers (0.6%) by two orders of magnitude more than CS […] Why so much attention to CS and tech?”—Reddit commenter.
“Men are underrepresented in many fields, like social work, education, psychology, healthcare, and veterinary medicine. Where’s all the hand-wringing about how men are being excluded?”—Quillette commenter.
Occupational gender segregation is one of the leading factors in the wage gap, and therefore desegregating high-paying, high-demand occupations is a social-justice concern. Additionally, researchers now recognize a form of sex discrimination that is hard to decipher: jobs filled predominantly by men have higher wages than jobs filled predominantly by women, even when the skill, effort and education level are comparable. The principle of comparable worth deals with this dilemma (Ref. 12). Examples of comparable-worth disputes in Ref. 12: in 1975, nurses sued the city of Denver because their wages were lower than tree trimmers and sign painters, and California school employees complained in 1985 about librarians and teaching assistants (mostly women) being paid less than men custodians and groundskeepers. Opposite examples, where male occupations are paid less for comparable skill and effort, are rare. A separate, compounded factor is hiring discrimination against women seeking to enter predominantly male occupations. Evidence of discrimination in hiring and promotion has accumulated from various empirical studies (e.g., using randomly assigned resumes that vary only in names). Anticipating more discrimination in some occupations, women alter their “choices.”
Frankly, I don’t know how to finish this narrative. And I ran out of time. This write-up is the outcome of preparing to give the Diversity Luncheon keynote at the SciPy Conference (12 July 2018). I consider it only the beginning: an invitation to educate each other to empower our activism with as much evidence as we can muster.
Note: I have focused on gender bias, out of personal interest; some of these points also apply to other dimensions of diversity and inclusion.
- Moss-Racusin, Corinne A., John F. Dovidio, Victoria L. Brescoll, Mark J. Graham, and Jo Handelsman, “Science faculty’s subtle gender biases favor male students.” Proceedings of the National Academy of Sciences109.41 (2012): 16474–16479, doi:10.1073/pnas.1211286109 (2012).
- Boring, Anne, Kellie Ottoboni and Philip Stark. “Student evaluations of teaching (mostly) do not measure teaching effectiveness.” ScienceOpen Research Section SOR-EDU (2016), doi: 10.14293/S2199–1006.1.SOR-EDU.AETBZC.v1. —See also media coverage: “Bias Against Female Instructors”in Inside Higher Education (Jan. 2016).
- Grunspan, Daniel Z., Sarah L. Eddy, Sara E. Brownell, Benjamin L. Wiggins, Alison J. Crowe, and Steven M. Goodreau. “Males under-estimate academic performance of their female peers in undergraduate biology classrooms.” PLoS ONE 11(2): e0148405, doi:10.1371/journal.pone.0148405 (2016).
- Reskin, Barbara F. and Patricia A. Roos, Job Queues, Gender Queues: Explaining Women’s Inroads Into Male Occupations, Temple University Press, 1990.
- Strober, Myra H. and Carolyn L. Arnold, Integrated Circuits / Segregated Labor: Women in three computer-related occupations, Stanford University Technical Report, 1984 [PDF]
- Roberts, Eric S., Marina Kassianidou, and Lilly Irani. “Encouraging women in computer science.” ACM SIGCSE Bulletin34(2): 84–88, doi:10.1145/543812.543837 (2002). [PDF]
- Kersey, Alyssa J., Emily J. Braham, Kelsey D. Csumitta, Melissa E. Libertus & Jessica F. Cantlon, “No intrinsic gender differences in children’s earliest numerical abilities,” npj Science of Learning, 3: 12, doi:10.1038/s41539–018–0028–7 (2018).
- Villeval, Marie Claire. “Ready, steady, compete.” Science335, no. 6068: 544–545, doi:10.1126/science.1218000(2012).
- Balafoutas, Loukas, and Matthias Sutter. “Affirmative action policies promote women and do not harm efficiency in the laboratory.” Science335, no. 6068: 579–582, doi:10.1126/science.1211180 (2012).
- Besley, Timothy, Olle Folke, Torsten Persson, and Johanna Rickne. “Gender quotas and the crisis of the mediocre man: Theory and evidence from Sweden,” American Economic Review107, no. 8: 2204–42, doi:10.1257/aer.20160080 (2017).
- Handley, Ian M., Elizabeth R. Brown, Corinne A. Moss-Racusin, and Jessi L. Smith. “Quality of evidence revealing subtle gender biases in science is in the eye of the beholder.” Proceedings of the National Academy of Sciences112(43): 13201–13206, http://dx.doi.org/10.1073/pnas.1510649112 (2015)
- Hidi and Renniger, The Four-Phase Model of Interest Development, Educational Psychologist, v41 n2 p111–127, 2006.
- England, P. Comparable worth: Theories and evidence. Routledge, 2017.