There are lots of conversations about the lack of diversity in science and tech these days. In response, people constantly ask, "So what? Why does it matter?" There are many ways to answer that question, but perhaps the easiest is this: because a homogenous team produces homogenous products for a very heterogeneous world.
This is Design Bias, a Motherboard column in which writer Rose Eveleth explores the products, research programs, and conclusions made not necessarily because any designer or scientist or engineer sets out to discriminate, but because to them the "normal" user always looks exactly the same. The result is a world that's biased by design. -the Editor
Before it was a television show, Grey's Anatomy was a textbook. Published in 1858, Gray's Anatomy (spelled with an "a") quickly became the gold standard in medical illustration, featuring detailed diagrams of everything from the tiny bones in the hand to the internal structure of the eye.
Gray's Anatomy is still in print today, in its 41st edition, but if you open the book up and flip through the pages you might notice something. Or really, the dearth of something: women. And it turns out it's not just Gray's Anatomy that has this problem—almost all medical textbooks are heavily biased towards depicting male bodies.
In 2014, Rhiannon Parker, a researcher at the University of Wollogong in Australia, set out to quantify just how bad that bias actually is. By analyzing more than 6,000 images from 17 anatomy textbooks published between 2008 and 2013, Parker and her colleagues found that only 36 percent of the anatomical images with an identifiable sex were female.
Even more discouraging is the results weren't all that different from a study done in 1994, in which 32 percent of images represented female bodies. "I expected there to be a much bigger improvement on representation," Parker told me.
Not all books were equally biased. General Anatomy, 2nd ed., had the highest proportion of male bodies at 5:1, while Human Anatomy and Physiology, 9th ed. was the only textbook to have the same proportion of male and female bodies. Gray's Anatomy for Students—a condensed version of Gray's Anatomy—has nearly three times more men as women.
"The textbook with the worst ratio was edited solely by men"
Interestingly, the editors of Human Anatomy, the textbook with parity, are Elaine N. Marieb and Katja Hoehn, both women, while the textbook with the worst ratio was edited solely by men.
Madelene Hyde, the vice president for content and education at Elsevier, which publishes several top anatomy books including Gray's Anatomy, said they are "trying more and more to be as balanced as possible with our textbook images."
"We take gender and racial diversity seriously and are working to include more diverse images in our textbooks," she told Motherboard in an email.
Read More: The Design Bias of Heart Failure
In some cases, showing a female body makes sense, if the content is specifically about female health. But Parker found that even in cases where there is no reason to show one sex over another, men are more likely to be depicted as the "normal body." This lines up with previous research from 1992 that found that even when it comes to medical imagery around reproduction, men outnumbered women in textbooks 2.5 to 1.
"For the anatomy titles that do not solely focus on surgical anatomy (interior rather than exterior), we do our best to provide images of diverse subjects," Hyde said in her email. "When possible, we try to replace older images of rare clinical conditions. As a provider of a significant volume of global health content, we try more and more to be as balanced as possible with our textbook images."
Parker also saw other trends in her dataset. Even though she was looking for images where the body being depicted might be identified by readers as male or female (whether by genitalia or by traditional gender markers), she also found that bodies were also overwhelmingly white, slim, and young. Of the images of women she did find, 86 percent of them were white (compared to 76 percent of identifiably male bodies).
Male bodies are almost always muscular, while female bodies are drawn thin. Only 2.7 percent of the images the team analyzed depicted visibly disabled bodies. And only 2.2 percent of the textbook images depicted elderly patients. "You don't have any elderly in these textbooks even though the elderly need more healthcare," Parker said. "I think that was quite surprising to me, that those types of representations weren't really in there."
Parker has also researched the impact the images might have on medical students. In a survey of 456 anatomy students studying at the School of Medicine at the University of Wollongong in 2018, she found that the disparity in bodies didn't affect their explicit bias, but it increased their scores on implicit bias tests that measure more subtle attitudes.
"People know what they should be saying, what they should be thinking" when it comes to gendered assumptions, Parker said. "But that doesn't stop the insidious impact of biased images. The way they normalize certain images and body types, you can't necessarily stop that from impacting your biases."
There is plenty of research to suggest that biases held by doctors have real, negative impacts on patients. "Coronary heart disease is seen as a male disease," Parker said. Another study found that many textbooks don't address the differences in the way coronary heart disease presents in women, opting to show men only, which could contribute to women being misdiagnosed at higher rates than men.
It's not just coronary heart disease, either. Parker notes that fatphobia among doctors leads to people not seeking medical attention when they need it, and that Black women in America are three times more likely to die in childbirth than white women, which many experts say is largely due to doctor bias against Black women.
Even beyond the bias it could engender, depicting the same body over and over again as white, male, and athletic isn't the best way to teach future doctors. "I think we all fall into [the idea] we want to draw pretty people; everybody is attracted to pretty people," said Jill Gregory, a medical illustrator and associate director of instructional technology at the Icahn School of Medicine at Mount Sinai. "But especially when you're illustrating a disease of older people, you shouldn't have a 20-year-old with coronary artery disease."
Stock image sites and online medical resources fall into the same trap. Gregory was recently asked to do a series of medical illustrations for an anatomy class at Mount Sinai. To depict breast health, "the professor had gotten random images off of Google," she said, which mostly showed a very specific kind of breast—perky, white, mostly with implants. "They're like these little balls on your chest. Most people don't have breasts like that," Gregory said.
So Gregory created an illustration of an elderly breast instead, collapsed with the nipple facing down. She said that students in the course were much more likely to encounter breasts that looked like her illustrations than the ones commonly depicted in books or stock images.
"It takes time for people to say, 'I should use diverse skin tones, maybe someone in a wheelchair'"
And sometimes illustrators are explicitly told to not include certain kinds of bodies. Parker said that in her research, she heard stories from illustrators who reported being asked not to show female nipples unless the illustration had something to do with breast health, for example.
By using past books as examples, illustrators are also perpetuating the bias. "Frank Netter, the father of medical illustration—his work is 100 percent white people," said Gregory. "It's a matter of not being lazy thinkers and starting to be, like, 'I shouldn't just default to the same skin tone.'"
This is exacerbated by the fact that the majority of medical illustrators are white (85 percent of people who responded to an Association of Medical Illustrators survey in 2018 identified as such), a problem that the association started tackling head on a few years ago with a special diversity committee (which Gregory is a member of).
The solution lies in the hands of publishers and illustrators. Gregory says that it's crucial to explain to people working in the field why this bias is harmful. "It's really about awareness. It takes time for people to say, 'Oh right, I should use diverse skin tones, maybe someone in a wheelchair, different kinds of people.'"
But are things getting better? Parker isn't so optimistic. She points out that doctors have a really rigid idea of what a healthy body looks like, and that's a hard thing to change. "No, I don't think it's changing," she told me. "I do think that maybe the gender disparity is slowly changing. But the intersections with ethnicity and body type, I don't think they're being paid attention to, and are getting worse."
Gregory, on the other hand, sees progress, even if it's been slow. She says that even she has changed her methods around this issue over the years. "Ten or 15 years ago I was maybe throwing something in for the heck of it, but now every time I approach an illustration I think about how I can include diversity in my images." Between that and recruiting a more diverse set of illustrators in the field, Gregory has hopes that things will get better.
Correction: In the original version of the story, Frank Netter's name was spelled incorrectly.