Invisible Women: Exposing Data Bias in a World Designed for Men

by Caroline Criado Perez

  The implicit bias is clear: expense codes are based on the assumption that the employee has a wife at home taking care of the home and the kids. This work doesn’t need paying for, because it’s women’s work, and women don’t get paid for it. Bovasso sums it up: ‘You can get $30 for takeout if you work late (because your wife isn’t there to cook you dinner) or $30 for Scotch if you want to drink your face off, but you can’t get $30 for a sitter (because your wife is at home with the kids).’ In the event, Bovasso was able to get her company to cover the cost of her childcare – but as she points out, ‘these have been exceptions I’ve had to ask for’. Which is women all over: always the exception, never the default.

  And in any case, not all employers will grant these exceptions. The Fawcett Society’s 2017 report on local government in England and Wales found that despite regulations dating from 2003 that call for ‘all councils to offer an allowance to cover the caring costs that councillors incur when fulfilling their role’, in reality, provision is patchy.133 Some councils don’t reimburse caring expenses at all, and most that do only pay a ‘contribution’. Rochdale Borough Council’s scheme ‘pays just £5.06 per hour, and specifically states that it is “a contribution rather than full reimbursement of carers’ expenses” – although this important caveat is notably not made for travel expenses’. Adding to the sense that this is a matter of priorities rather than resources, most local-government meetings take place in the evening (when childcare is most likely to be needed), and although it is standard practice in many countries from the US to Sweden for councillors to remotely attend or vote at meetings, current law does not allow for this cheaper alternative.

  It is abundantly clear that the culture of paid work as a whole needs a radical overhaul. It needs to take into account that women are not the unencumbered workers the traditional workplace has been designed to suit, and that while men are more likely to fit into this automaton ideal, increasing numbers of them no longer want to. After all, it is simply a fact that none of us, including businesses, could do without the invisible, unpaid work carers do. So it is time to stop penalising them for doing it. Instead, we must start recognising it, valuing it, and designing the paid workplace to account for it.

  CHAPTER 4

  The Myth of Meritocracy

  For most of the twentieth century there were no female musicians in the New York Philharmonic Orchestra. There were a couple of blips in the 1950s and 60s, when a woman or two was hired, but those aside, the proportion of women sat stubbornly at zero. But then all of a sudden, something changed: from the 1970s onwards, the numbers of female players started to go up. And up.

  Turnover in orchestras is extremely low. The composition of an orchestra is fairly static (at around one hundred players), and when you’re hired, it’s often for life; it’s rare that a musician is fired. So there was something remarkable going on when the proportion of women in this orchestra grew from a statistical 0% to 10% in a decade.

  That something was blind auditions.1 Instituted in the early 1970s following a lawsuit, blind auditions are what they sound like: the hiring committee can’t see who is playing in the audition, because there is a screen between them and the player.2 The screens had an immediate impact. By the early 1980s, women began to make up 50% of the share of new hires. Today, the proportion of female musicians in the New York Philharmonic stands at over 45%.3

  The simple step of installing a screen turned the audition process for the New York Philharmonic into a meritocracy. But in this, it is an outlier: for the vast majority of hiring decisions around the world, meritocracy is an insidious myth. It is a myth that provides cover to institutional white male bias. And, dishearteningly, it is a myth that proves remarkably resistant to all the evidence, going back decades, that shows it up as the fantasy it most certainly is. If we want to kill this myth off, we’re clearly going to have to do more than just collect data.

  The fact that meritocracy is a myth is not a popular one. Around the industrialised world, people believe that not only is meritocracy the way things should work, it’s the way things do work.4 Despite evidence suggesting that, if anything, the US is less meritocratic than other industrialised countries,5Americans in particular hold on to meritocracy as an article of faith, and employment and promotion strategies over the past few decades have increasingly been designed as if meritocracy is a reality. A survey of US firms found that 95% used performance evaluations in 2002 (compared to 45% in 1971) and 90% had a merit-based pay plan in place.6

  The problem is, there is little evidence that these approaches actually work. In fact, there is strong evidence that they don’t. An analysis of 248 performance reviews collected from a variety of US-based tech companies found that women receive negative personality criticism that men simply don’t.7 Women are told to watch their tone, to step back. They are called bossy, abrasive, strident, aggressive, emotional and irrational. Out of all these words, only aggressive appeared in men’s reviews at all – ‘twice with an exhortation to be more of it’. More damningly, several studies of performance-related bonuses or salary increases have found that white men are rewarded at a higher rate than equally performing women and ethnic minorities, with one study of a financial corporation uncovering a 25% difference in performance-based bonuses between women and men in the same job.8

  The myth of meritocracy achieves its apotheosis in America’s tech industry. According to a 2016 survey, the number one concern of tech start-up founders was ‘hiring good people’, while having a diverse workforce ranked seventh on the list of ten business priori-ties.9 One in four founders said they weren’t interested in diversity or work-life balance at all. Which, taken together, points to a belief that if you want to find ‘the best people’, addressing structural bias is unnecessary. A belief in meritocracy is all you need.

  Actually, a belief in meritocracy may be all you need – to introduce bias, that is. Studies have shown that a belief in your own personal objectivity, or a belief that you are not sexist, makes you less objective and more likely to behave in a sexist way.10 Men (women were not found to exhibit this bias) who believe that they are objective in hiring decisions are more likely to hire a male applicant than an identically described female applicant. And in organisations which are explicitly presented as meritocratic, managers favour male employees over equally qualified female employees.

  Tech’s love affair with the myth of meritocracy is ironic for an industry so in thrall to the potential of Big Data, because this is a rare case where the data actually exists. But if in Silicon Valley meritocracy is a religion, its God is a white male Harvard dropout. And so are most of his disciples: women make up only a quarter of the tech industry’s employees and 11% of its executives.11 This is despite women earning more than half of all undergraduate degrees in the US, half of all undergraduate degrees in chemistry, and almost half in maths.12

  More than 40% of women leave tech companies after ten years compared to 17% of men.13 A report by the Center for Talent Innovation found that women didn’t leave for family reasons or because they didn’t enjoy the work.14 They left because of ‘workplace conditions’, ‘undermining behaviour from managers’, and ‘a sense of feeling stalled in one’s career’. A feature for the Los Angeles Times similarly found that women left because they were repeatedly passed up for promotion and had their projects dismissed.15 Does this sound like a meritocracy? Or does it look more like institutionalised bias?

  That the myth of meritocracy survives in the face of such statistics is testament to the power of the male default: in the same way that men picture a man 80% of the time they think of a ‘person’, it’s possible that many men in the tech industry simply don’t notice how male-dominated it is. But it’s also testament to the attractiveness of a myth that tells the people who benefit from it that all their achievements are down to their own personal merit. It is no accident that those who are most likely to believe in the myth of meritocracy are young, upper-class, white Americans.16

&n
bsp; If white upper-class Americans are most likely to believe in the myth of meritocracy, it should come as no surprise that academia is, like tech, a strong follower of the religion. The upper ranks of academia – particularly those of science, technology, engineering and maths (STEM) – are dominated by white, middle- and upper-class men. It is a perfect Petri dish for the myth of meritocracy to flourish in. Accordingly, a recent study found that male academics – particularly those in STEM – rated fake research claiming that academia had no gender bias higher than real research which showed it did.17 Also accordingly, gender bias is in fact plentiful – and well documented.

  Numerous studies from around the world have found that female students and academics are significantly less likely than comparable male candidates to receive funding, be granted meetings with professors, be offered mentoring, or even to get the job.18 Where mothers are seen as less competent and often paid less, being a father can work in a man’s favour (a gendered bias that is by no means restricted to academia).19 But despite the abundance of data showing that academia is in fact far from meritocratic, universities continue to proceed as if male and female students, and male and female academics, are operating on a level playing field.

  Career progression in academia depends largely on how much you get published in peer-reviewed journals, but getting published is not the same feat for men as it is for women. A number of studies have found that female-authored papers are accepted more often or rated higher under double-blind review (when neither author nor reviewer are identifiable).20,21 And although the evidence varies on this point, given the abundant male bias that has been identified in academia, there seems little reason not to institute this form of blind academic audition. Nevertheless, most journals and conferences carry on without adopting this practice.

  Of course, female academics do get published, but that’s only half the battle. Citation is often a key metric in determining research impact, which in turn determines career progression, and several studies have found that women are systematically cited less than men.22 Over the past twenty years, men have self-cited 70% more than women23 – and women tend to cite other women more than men do,24 meaning that the publication gap is something of a vicious circle: fewer women getting published leads to a citations gap, which in turn means fewer women progress as they should in their careers, and around again we go. The citations gap is further compounded by male-default thinking: as a result of the widespread academic practice of using initials rather than full names, the gender of academics is often not immediately obvious, leading female academics to be assumed to be male. One analysis found that female scholars are cited as if they are male (by colleagues who have assumed the P stands for Paul rather than Pauline) more than ten times more often than vice versa.25

  Writing for the New York Times, economist Justin Wolfers noted a related male-default habit in journalists routinely referring to the male contributor as the lead author when in fact the lead author was a woman.26 This lazy product of male-default thinking is inexcusable in a media report, but it’s even more unacceptable in academia, and yet here too it proliferates. In economics, joint papers are the norm – and joint papers contain a hidden male bias. Men receive the same level of credit for both solo and joint papers, but, unless they are writing with other female economists, women receive less than half as much credit for co-authored papers as men do. This, a US study contends, explains why, although female economists publish as much as male economists, male economists are twice as likely to receive tenure.27 Male-default thinking may also be behind the finding that research perceived to have been done by men is associated with ‘greater scientific quality’:28 this could be a product of pure sexism, but it could also be a result of the mode of thinking that sees male as universal and female as niche. It would certainly go some way to explaining why women are less likely to appear on course syllabuses.29

  Of course before a woman gets to face all these hidden hurdles, she must have found the time to do the research in the first place, and that is by no means a given. We’ve already discussed how women’s unpaid workload outside of paid employment impacts on their ability to do research. But their unpaid workload inside the workplace doesn’t help either. When students have an emotional problem, it is their female professors, not their male professors they turn to.39 Students are also more likely to request extensions, grade boosts, and rule-bending of female academics.31 In isolation, a request of this kind isn’t likely to take up much time or mental energy – but they add up, and they constitute a cost on female academics’ time that male academics mostly aren’t even aware of, and that universities don’t account for.

  Women are also asked to do more undervalued admin work than their male colleagues32 – and they say yes, because they are penalised for being ‘unlikeable’ if they say no. (This is a problem across a range of workplaces: women, and in particular ethnic minority women, do the ‘housekeeping’ – taking notes, getting the coffee, cleaning up after everyone – in the office as well as at home.33) Women’s ability to publish is also impacted by their being more likely than their male colleagues to get loaded with extra teaching hours,34 and, like ‘honorary’ admin posts, teaching is viewed as less important, less serious, less valuable, than research. And we run into another vicious circle here: women’s teaching load prevents them from publishing enough, which results in more teaching hours, and so on.

  The inequity of women being loaded with less valued work is compounded by the system for evaluating this work, because it is itself systematically biased against women. Teaching evaluation forms are widely used in higher education and they represent another example of a situation where we have the data, but are simply ignoring it. Decades of research35 in numerous countries show that teaching evaluation forms are worse than useless at actually evaluating teaching and are in fact ‘biased against female instructors by an amount that is large and statistically significant’.36 They are, however, pretty good at evaluating gender bias. One of these biases is our old friend ‘men are the default human’, which shows up in objections to female lecturers straying away from a focus on white men. ‘I didn’t come out of this course with any more information except gender and race struggles, than I came in with,’ complained one student who apparently felt that gender and race were not relevant to the topic at hand: US confederation.37

  Falling into the trap we encountered in the introduction, of not realising that ‘people’ is as likely to mean ‘women’ as it is to mean ‘men’, another student complained that, ‘Although Andrea stated on the first day she would teach a peoples [sic] perspective it was not illustrated how much was going to be focused on first nation and women’s history.’ Incidentally, it’s worth taking the implication that this lecturer focused almost exclusively on ‘first nations and women’s history’ with a pinch of salt: a friend of mine got a similarly unhappy review from a male student for focusing ‘too much’ on feminism in her political philosophy lectures. She had spoken about feminism once in ten classes.

  Less effective male professors routinely receive higher student evaluations than more effective female teachers. Students believe that male professors hand marking back more quickly – even when that is impossible because it’s an online course delivered by a single lecturer, but where half the students are led to believe that the professor is male and half female. Female professors are penalised if they aren’t deemed sufficiently warm and accessible. But if they are warm and accessible they can be penalised for not appearing authoritative or professional. On the other hand, appearing authoritative and knowledgeable as a woman can result in student disapproval, because this violates gendered expectations.38 Meanwhile men are rewarded if they are accessible at a level that is simply expected in women and therefore only noticed if it’s absent.

  An analysis39 of 14 million reviews on the website RateMyProfessors.com found that female professors are more likely to be ‘mean’, ‘harsh’, ‘unfair’, ‘strict’ and ‘annoying’. And it’s get
ting worse: female instructors have stopped reading their evaluations in droves, ‘as student comments have become increasingly aggressive and at times violent’. A female political history lecturer at a Canadian university received the following useful fredback from her student: ‘I like how your nipples show through your bra. Thanks.’40 The lecturer in question now wears ‘lightly padded bras’ exclusively.

  The teaching evaluation study that revealed women are more likely to be ‘mean’ also found that male professors are more likely to be described as ‘brilliant’, ‘intelligent’, ‘smart’ and a ‘genius’. But were these men actually more in possession of raw talent than their female counterparts? Or is it just that these words are not as gender neutral as they appear? Think of a genius. Chances are, you pictured a man. It’s OK – we all have these unconscious biases. I pictured Einstein – that famous one of him sticking his tongue out, his hair all over the place. And the reality is that this bias (that I like to call ‘brilliance bias’) means that male professors are routinely considered more knowledgeable, more objective, more innately talented. And career progression that rests on teaching evaluations completely fails to account for it.

  Brilliance bias is in no small part a result of a data gap: we have written so many female geniuses out of history, they just don’t come to mind as easily. The result is that when ‘brilliance’ is considered a requirement for a job, what is really meant is ‘a penis’. Several studies have found that the more a field is culturally understood to require ‘brilliance’ or ‘raw talent’ to succeed – think philosophy, maths, physics, music composition, computer science – the fewer women there will be studying and working in it.41 We just don’t see women as naturally brilliant. In fact, we seem to see femininity as inversely associated with brilliance: a recent study where participants were shown photos of male and female science faculty at elite US universities also found that appearance had no impact on how likely it was that a man would be judged to be a scientist.42 When it came to women, however, the more stereotypically feminine they looked, the less likely it was that people would think they were a scientist.