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

by Caroline Criado Perez

  Fairly simply, to be honest. Henry Jackson and Jonathan Schenker are clearly well-meaning men who don’t intend to shut women out. But it’s Sergey Brin and the pregnancy parking all over again: even the best of men can’t know what it’s like to go through the world as a person with a body which some other people treat as an access-all-areas amusement arcade. This just isn’t something that Jackson and Schenker have to face on a regular basis, and therefore it really isn’t all that surprising that they missed ‘something so obvious’.

  Male violence is far from the only issue keeping women out of VR. From the oversized headsets, to research showing that VR causes motion sickness in women to a far larger degree than it does in men,43 to the fact that narrow computer displays favour men in tasks that require spatial awareness,44 you’re left with another platform that just doesn’t work well for women – and that is therefore likely to have fewer women on it.

  We don’t know exactly why women are more likely to experience motion sickness while using VR, but Microsoft researcher danah boyd conducted a study that suggests a possible explanation.45 Human eyes use two basic cues to determine depth: ‘motion parallax’ and ‘shape-from-shading’. Motion parallax refers to how an object seems bigger or smaller depending on how close you are to it, while shape-from-shading refers to the way the shading of a point changes as you move. And while 3D VR is pretty good at rendering motion parallax, it still does ‘a terrible job’ of emulating shape-from-shading.

  This discrepancy creates sex differences in how well VR works, because, as boyd discovered, men are ‘significantly more likely’ to rely on motion parallax for depth perception, while women rely on shape-from-shading. 3D environments are literally sending out information signals that benefit male over female depth perception. The question is: would we be so behind on recreating shape-from-shading if we had been testing 3D VR on equal numbers of men and women from the start?

  Tom Stoffregen, professor of kinesiology at the University of Minnesota, has an entirely different theory for why women experience more motion sickness than men. The classic theories, he says, are ‘focused almost entirely on sensory stimulation’. The idea is that what you feel in your inner ear doesn’t match up with what you see with your eyes. And ‘that’s true,’ says Stoffregen, ‘but it’s not the only thing that changes. The big thing that traditional theories have not talked about,’ he says, ‘is changes in what you need to do to control your body.’

  In the normal course of your day, your body is constantly making micro-adjustments in order to keep you stable. When you stand, when you sit, when you walk. But when you’re in a moving environment – say a car, or a ship – what you have to do to remain stable changes because your body is being destabilised. So, says Stoffregen, ‘You are being physically required to move in a different way and you haven’t learnt how to do it yet.’ And like cars and ships, he says, VR destabilises the body. Hence the motion sickness.

  The VR industry has so far shown little interest in Stoffregen’s research. ‘They understand that it’s a serious problem,’ but they’re going the wrong way about fixing it, he says. ‘The people who design VR think it’s just an object that you put in front of your eyes, and the idea that it would have anything to do with anything other than your eyes is incoherent to them.’ But, he says, VR developers have to understand that they’re doing more than ‘simply placing screens in front of people’s eyes. Whether they like it or and whether they know it or not.’

  VR developers also have to start collecting data systematically – and separating it by sex. ‘Most of the data on motion sickness in VR are anecdotal,’ explains Stoffregen, ‘and they come from the people who are working in these companies just using the systems themselves or trying them out at computer technology conferences or whatever. So they’re completely unsystematic – and most of these people are men.’

  One of the most convincing aspects of Stoffregen’s theory is how it finally explains why I get car sick in every seat other than the driving seat: it’s all about control. When you’re walking, you are in control of your movements. You know what’s coming. On a ship, or in a car, someone else is in control – unless you’re the driver. ‘The driver knows what the motion of the car is going to be and so the driver is able to stabilise his or herself in what we call an anticipatory fashion,’ explains Stoffregen, ‘whereas the passenger cannot know in quantitative detail what the car is going to be doing. And so their control of their own body must be compensatory. And anticipatory control is just better than compensatory control. You know, that ain’t no rocket science.’

  But where does the sex difference come in? ‘Everybody who studies motion sickness has known that women are more susceptible than men basically forever,’ says Stoffregen. ‘It’s just an utterly uncontroversial fact. It’s just there. But, and he includes himself in this group, ‘very few people did any research on it or made any effort to try and figure that out.’ Plus ça change.

  But in 2010, Stoffregen made a discovery. ‘I was sort of fooling around in the literature and I came across some results that I didn’t know about,’ and which showed that there are sex differences in body sway. ‘These are small subtle differences. You can’t just watch somebody and see them, but in terms of the subtle quantitative details of how the body moves back and forth, there are in fact reliable sex differences. And as soon as I saw this, I mean, I’m talking the minute that I saw this I knew that oh, OK, now I have something to say about the sex difference in motion sickness. Because my whole story about motion sickness is that it’s related to the control of the body.’ Since then, Stoffregen has also discovered evidence that ‘women’s postural sway changes across the menstrual cycle’. And this is significant because ‘a woman’s susceptibility to motion sickness changes across the menstrual cycle. And those two things link up, believe it or not.’

  A considerable gender data gap remains. We don’t yet know exactly how and when women’s body-sway changes. But as a woman who suffers from extreme car sickness, I am excited and enraged by Stoffregen’s findings, particularly because of how it ties into another gender data gap I’ve been looking into: car design.

  When you’re sitting down, you’re still swaying. ‘If you’re sitting on a stool then you’re swaying around your hips,’ explains Stofcegen. ‘If your chair has a back, then your head is swaying on your neck. The only way to really get rid of that is to have a headrest and to use it,’ he adds. And I feel like one of those cartoon light bulbs has just gone off in my head. What if the headrest is at the wrong height, at the wrong angle, and the wrong shape to accommodate your body? Could women’s increased propensity to motion sickness in cars be exacerbated by cars being designed around the male body, I ask. ‘I think that’s quite possible, sure,’ Stoffregen replies. ‘The quality of the stabilisation, if it’s the wrong height or whatever . . . what you’re telling me is new to me, but it sounds perfectly plausible.’

  But here I run into yet another data gap: the available research on whether car headrests have been designed to account for the female body is seemingly non-existent. This gap is hardly unexpected though: car design has a long and ignominious history of ignoring women.

  Men are more likely than women to be involved in a car crash, which means they dominate the numbers of those seriously injured in car accidents. But when a woman is involved in a car crash, she is 47% more likely to be seriously injured than a man, and 71% more likely to be moderately injured,46 even when researchers control for factors such as height, weight, seat-belt usage, and crash intensity.47 She is also 17% more likely to die.48 And it’s all to do with how the car is designed – and for whom.

  Women tend to sit further forward than men when driving. This is because we are on average shorter. Our legs need to be closer to reach the pedals, and we need to sit more upright to see clearly over the dashboard.49 This is not, however, the ‘standard seating position’. Women are ‘out of position’ drivers.50 And our wilful deviation from the norm means that we
are at greater risk of internal injury on frontal collisions.51 The angle of our knees and hips as our shorter legs reach for the pedals also makes our legs more vulnerable.52 Essentially, we’re doing it all wrong.

  Women are also at higher risk in rear-end collisions. Women have less muscle on our necks and upper torso than men, which make us more vulnerable to whiplash (by up to three times53), and car design has amplified this vulnerability. Swedish research has shown that modern seats are too firm to protect women against whiplash injuries: the seats throw women forward faster than men because the back of the seat doesn’t give way for women’s on average lighter bodies.54 The reason this has been allowed to happen is very simple: cars have been designed using car-crash test dummies based on the ‘average’ male.

  Crash-test dummies were first introduced in the 1950s, and for decades they were based around the fiftieth percentile male. The most commonly used dummy is 1.77 cm tall and weighs 76 kg (significantly taller and heavier than an average woman), and the dummy also has male muscle-mass proportions and a male spinal column. In the early 1980s, researchers argued for the inclusion of a fiftieth percentile female in regulatory tests, but this advice was ignored.55 It wasn’t until 2011 that the US started using a female crash-test dummy,56 although, as we’ll see, just how ‘female’ these dummies are is questionable.

  In 2018, Astrid Linder, research director of traffic safety at the Swedish National Road and Transport Research Institute, presented a paper at the Road Safety on Five Continents Conference in South Korea in which she ran through EU regulatory crash-test requirements.57 In the EU, there are five tests a car must pass before being allowed on the market: one safety-belt test, two frontal-collision tests, and two lateral-collision tests. In no test is an anthropometrically correct female crash-test dummy required. The seat-belt test, one of the frontal-collision tests, and both lateral-collision tests all specify that a fiftieth-percentile male dummy should be used. When Linder looked at regulatory tests worldwide, she found that while there are ‘several local differences’, regulatory tests are still using the fiftieth-percentile male ‘to represent the whole adult population’.

  There is one EU regulatory test that requires what is called a fifth-percentile female dummy, which is meant to represent the female population. Only 5% of women will be shorter than this dummy. But there are a number of data gaps. For a start, this dummy is only tested in the passenger seat, so we have no data at all for how a female driver would be affected – something of an issue you would think, given women’s ‘out of position’ driving position. And secondly, this female dummy is not really female. It is just a scaled-down male dummy.

  Consumer tests can be slightly more stringent. When I spoke to the EuroNCAP (a European organisation that provides car safety ratings for consumers) they informed me that since 2015 they have used male and female dummies in both front-crash tests and that they do base their female dummies on female anthropometric data – with the caveat that this is ‘where data is available’. And that, says Linder, is quite the caveat: ‘To my knowledge, little or even no such data is available,’ she tells me, adding that there is ‘currently no way EuroNCAP can identify the protective systems that protect both males and females the best way.’ In any case, EuroNCAP acknowledged that ‘sometimes’ they do just use scaled-down male dummies. But, as we’ll cover extensively in the next chapter, women are not scaled-down men. We have different muscle-mass distribution. We have lower bone density. There are sex differences in vertebrae spacing. As Stoffregen has noted, even our body sway is different. And these differences are all crucial when it comes to injury rates in car crashes.

  The situation is even worse for pregnant women. Although a pregnant crash-test dummy was created back in 1996, testing with it is still not government-mandated either in the US or in the EU.58 In fact, even though car crashes are the number-one cause of foetal death related to maternal trauma,59 we haven’t even yet developed a seat belt that works for pregnant women. Research from 2004 suggests that pregnant women should use the standard seat belt,60 but 62% of third-trimester pregnant women don’t fit the standard seat-belt design.61 A three-point seat belt can also ride up on women who carry low, which a 1996 study found can treble or quadruple force transmission to the abdomen compared to when the belt is worn below the uterus, ‘with a corresponding increased risk of fetal injury’.62 Standard seat belts aren’t great for nonpregnant women either: apparently, in an effort to accommodate our breasts many of us are wearing seat belts ‘improperly’ which again, increases our risk of injury (another reason we should be designing explicitly female dummies rather than just smaller male dummies).63 And it’s not just a woman’s belly that changes in pregnancy: breast-size changes can also diminish seat-belt efficacy by affecting positioning. Here again, we find an example of a situation where we have the data on women, but are just ignoring it. Clearly what is needed is a wholesale redesign of cars using complete data, and this should be fairly simple since it’s not exactly hard to find women to model a test dummy on.

  Even with all these gaps, the 2011 introduction of the female crash-test dummy in the US still sent cars’ star ratings plummeting. The Washington Post reported on the experience of Beth Milito and her husband, who bought a 2011 Toyota Sienna, based primarily on its four-star safety rating.64 But all was not as it seemed. The passenger seat, which Milito says she is likely to be sitting in when they are ‘out and about as a family’, had a two-star rating. In the previous year’s model, the front passenger seat (tested on a male dummy) had earned a top five-star rating. But the shift to female dummies revealed that in a front collision at 35 mph a female passenger had a 20-40% risk of being killed or seriously injured. The average risk of death for that class of vehicle, explains the Washington Post, is 15%.

  A 2015 report by the Insurance Institute for Highway Safety is excitingly headlined ‘Improved vehicle designs bring down death rates’ – which sounds great. Perhaps this is the result of the new legislation? Unlikely. Buried in the report is the following telltale line: ‘The rates include only driver deaths because the presence of passengers is unknown.’ This is a huge gender data gap. When men and women are in a car together, the man is most likely to be driving.65 So not collecting data on passengers more or less translates as not collecting data on women.

  The infuriating irony of all this is that the gendered passenger/driver norm is so prevalent that, as we’ve seen, the passenger seat is the only seat that is commonly tested with a female crash-test dummy anyway, with the male crash-test dummy still being the standard dummy for the driver’s seat. So stats that include only driver fatalities tell us precisely zero about the impact of introducing the female crash-test dummy. In conclusion, a more accurate headline for the report would be ‘Improved vehicle design brings down death rates in the seat most likely to be occupied by men, but who knows about death rates in the seat most likely to be occupied by women even though we already know women are 17% more likely to die in a car crash.’ Admittedly, this is less snappy.

  When I speak to Dr David Lawrence, director of safety-literature database the SafetyLit Foundation, he tells me that ‘in most US states the quality of police crash reports is at best poor for use as a research tool’. Little data is gathered about anyone other than the driver. Written police reports have often been handed to ‘contract companies for data entry’, most of whom use prison labour for data entry. ‘Data-quality checks were rare and when quality was evaluated it was found wanting. For example, in Louisiana for most crashes in the 1980s most of the occupants were males who were born on January 1st, 1950. Almost all of the vehicles involved in crashes were the 1960 model year.’ Except they weren’t. These were just the default settings.

  Lawrence tells me that even though this problem has been found in ‘many other states’, the data hasn’t improved ‘because no changes in the data-entry practices were made. The federal government required that states provide police crash-report data to NHTSA (The National Highway T
raffic Safety Administration) but set no standards for data quality nor penalty for sending junk data.’

  Astrid Linder has been working on what she says will be the first crash-test dummy to accurately represent female bodies. Currently, it’s just a prototype, but she is calling on the EU to make testing on anthropometrically correct female crash-test dummies a legal requirement. In fact, Linder argues that this technically already is a legal requirement. Article 8 of the legally binding Treaty of the Functioning of the European Union reads, ‘In all its activities, the Union shall aim to eliminate inequalities, and to promote equality, between men and women.’66 Clearly, women being 47% more likely to be seriously injured in a car crash is one hell of an inequality to be overlooking.

  In some ways it’s hard to understand why a proper female crash-test dummy hasn’t been developed and made a legal requirement in car tests years ago. But on the other hand, and given all we know about how women and their bodies are routinely ignored in design and planning, it’s not surprising at all. From development initiatives to smartphones, from medical tech to stoves, tools (whether physical or financial) are developed without reference to women’s needs, and, as a result these tools are failing them on a grand scale. And this failure affects women’s lives on a similarly grand scale: it makes them poorer, it makes them sicker, and, when it comes to cars, it is killing them. Designers may believe they are making products for everyone, but in reality they are mainly making them for men. It’s time to start designing women in.

  PART IV

  Going to the Doctor

  CHAPTER 10

  The Drugs Don’t Work

  It took twelve years for Michelle to receive a diagnosis. ‘I was about fourteen when I first started having symptoms,’ she tells me. ‘I was too ashamed to go to a doctor for it.’ She kept her urgent, painful, frequent, sometimes bloody bowel movements a secret for two years, until one night, it hurt too much to hide anymore. ‘I couldn’t move from the foetal position on my bathroom floor. I was afraid I was dying.’ She was sixteen.