A community of optimists hosted by Melinda Gates

The dangers of gender bias in design

Section of the book cover for Invisible Women by Caroline Criado-Perez
Caroline Criado Perez is a writer and feminist who uses social media to fight sexism in her native United Kingdom. Her successes include the fight to install a statue of a suffragist in London’s Parlaiment Square (where it sits among eleven statues of men) and the campaign to keep a woman on English bank notes.

We recently spoke with Criado Perez to discuss her latest book, Invisible Women. The book explores how everyday objects, technologies, and experiences—from seat belts, to voice recognition software, to public restrooms—are designed for, and by, men, and how this bias impacts not just the comfort, but also the safety, of women worldwide. We’ve included illustrations throughout that highlight the bias in these designs.

What inspired you to write Invisible Women?

It was when I discovered that the classic heart attack symptoms that I'd always been taught—pain in the chest and down the left arm—were actually the heart attack symptoms for men. Women are more likely to experience breathlessness, nausea, fatigue, and what feels like indigestion. But because public health information focuses on male symptoms, women don’t realize they’re having heart attacks. Worse, doctors don’t realize. The result is that women are more likely to die following a heart attack than men.

I was so shocked that this was happening in medicine. Our society positions science as neutral; as objective and free of bias. Science deals in facts. In truth. Only, now it turned out that our cultural positioning of men as the default humans was corrupting science. And as a result, women were dying.
On the face of it, allocating equal space to male and female restrooms may seem fair. But for a variety of reasons—from the time it takes to change a tampon or menstrual pad, to the greater likelihood that a woman will bring a child or elderly person to the bathroom with her—women take up to 2.3 times as long as men to use the toilet. 1
Illustration by Andrew Jernberg

In the book, which cites hundreds of studies from around the world, you talk about how women are systematically excluded from data gathering. How, exactly, did you notice this showing up in your research? Can you give us an example that really surprised you?

When local officials in the town of Karlskoga in Sweden looked at their snow-clearing schedules, they realized that they had designed them to meet the needs of men. Men tend to have much simpler travel patterns than women: a twice daily commute in a car. But because women have to combine their paid work with their unpaid care work (women still do 75 percent of the world’s unpaid care work), their travel patterns are more complicated. They make lots of short interconnected trips, and are more likely to use public transport. As a result, the order in which the snow was being cleared (major roads first; local roads and sidewalks second) benefitted men.

So they decided to switch the order around—and found to their surprise that the number of admissions to the emergency room fell dramatically. Because it wasn’t men in their cars who were falling over and fracturing their bones: it was women pushing buggies through the snow. If they had designed their schedule based on sex-disaggregated travel and hospital admission data in the first place, they could have saved a lot of money over the decades.
Because public health information focuses on male symptoms, women don’t realize they’re having heart attacks. Worse, doctors don’t realize.
I think that in the end, what shocked me the most over the course of my research were the excuses. The excuse I came across most often in the course of writing the book was that women are just too complicated. This excuse appeared in fields ranging from the economy, to travel infrastructure, to medicine. Women’s working lives are too complicated, our travel patterns are too complicated, our bodies are too complicated. And instead of engaging with that complexity, researchers prefer to just exclude half the world. They choose to save money rather than to save women’s lives.
Attempts to introduce alternatives to the traditional three-stone stove have failed for decades. A major reason: The stove’s designs repeatedly fail to take into account the needs of the women who actually use them. 2
Illustration by Andrew Jernberg

You argue in the book that biased data limits tech innovation—say more about that?

The data gaps in tech manifest in two ways. First, because the datasets on which we train algorithms are hopelessly male biased, voice recognition software doesn’t recognize female voices, translation software translates female doctors into male doctors, and image-labeling software labels men as women if they are standing next to an oven. And these are the least harmful examples.

It gets much less amusing when you start thinking about women being diagnosed by algorithms trained on current medical data. Because of the way machine-learning works, when you feed it biased data, it gets better and better—at being biased. We could be literally writing code that makes healthcare for women worse.
We could be literally writing code that makes healthcare for women worse.
The other issue is how male-dominated tech is. This is a data gap in its own way: white middle-class men from America simply cannot be aware of the needs of all of humanity. And so the tech that they develop will inevitably be biased towards white middle-class men from America.

Tech is littered with examples of how this plays out, from Apple’s “comprehensive” health app that you could use to track your copper intake but not your period, to step tracker apps that forget women usually don’t have pockets big enough to carry their phones on them at all times.

Then there’s the issue of funding. Ninety-three percent of venture capitalists are men, and these teams suffer from the same problem as male-dominated developer teams: they simply aren't aware of certain female needs. As a result, entrepreneurs developing new tech for women need good data because they will not be able to rely on a VC already having personal experience with, for example, how terrible all the current breast pump options are.

But because we lack data on female bodies, such entrepreneurs are less likely to have the information they need to make the case for their ideas. They are therefore less likely to get funding. And so tech for women remains, for the most part, of the shrink it, pink it and price it up variety, rather than genuinely catering to women’s needs.
Voice recognition software "hears" male voices more easily than female ones. This can be dangerous, such as when a car’s voice command system fails to understand what a female driver is saying, or when the notes that a female health care provider dictates about a patient’s course of care are riddled with errors. This is because AI assistants are trained on male biased data. 3
Illustration by Andrew Jernberg

You mention the “default male” in the book as being the body type for which so many things are designed, from seat belts to cell phones. Men (and women, and people of all gender identities) obviously come in different shapes and sizes. What does the default male body look like? And how are you able to determine if something was designed for a “default male” body?

In the 1930s, the influential Swiss architect Le Corbusier came up with “the human scale” for architecture, and by "human" he meant a six-foot man with his arm raised. This was the “modular man.”

In the 1950s, when car crash test dummies were first introduced, the dummies were based around the 50th-percentile male, as if he were the 50th-percentile human. In the last few years, car regulators seem to have belatedly realized that women exist, and have since introduced what they call a “female” car crash test dummy. In reality, this dummy is simply a scaled-down male dummy—and women are not scaled-down men. What's more, in the EU, this dummy is required in just one of five regulatory tests, and only in the passenger seat. Male remains the default.

When typical female heart attack symptoms are called “atypical,” this is default male. When “gender-neutral" toxicity levels for chemicals are determined using data on men, this is default male. When researchers complain that female bodies are too complicated to test on, this is default male—how else could you justify excluding half the world for reasons of simplicity?

Unfortunately, my research has taught me to be more surprised when women have been accounted for than when they haven’t. You can more or less bet on most things you encounter in the day-to-day as having been designed for the average man.
Seat belts were designed for the average male body. Driver’s seats, too, were designed with the male body in mind, and fail to take into account the fact that women tend to sit closer to the steering wheel 4.
Illustration by Andrew Jernberg

What needs to happen to ensure more products and everyday objects are designed with a range of body types in mind?

There’s a very simple answer to this: hire more women and promote them to leadership positions. Make sure your product-design teams are truly diverse. Recognize that diversity is not simply a tick box to get past: It is the key to designing products that really work for everyone. And collect sex-disaggregated data, from the very beginning of every process!
Clip-on mic packs are designed for waistbands and pockets, both of which many dresses lack—making it harder to mic a woman for a public speaking engagement or television appearance. 5
Illustration by Andrew Jernberg

As someone who seems so well-informed about injustice, how do you remain optimistic, in the face of that? What do you do to maintain your optimism?

I am not always optimistic. Just like everyone, I have moments of despair. If I look around and see how much there is to fix, it can all feel overwhelming and pointless. So I try to focus on the things I can fix. Fixing things—and working to fix things—makes me feel less angry and more hopeful.

Feminism is a movement. It relies on millions of people around the world doing their part. The more of us who do our little bit, the faster we will achieve equality.