NASA scheduled the first all-female space walk for the end of last month. But a mere four days before the historic event was meant to happen, they scrapped the plan and subbed in a male astronaut, claiming it was because they did not have enough space suits in the proper size to fit all the women astronauts.

Unfortunately, women all too often must make do with equipment designed for men, an oversight that can be more than a PR embarrassment. Many police stab vests fail to accommodate women’s breasts, causing the protective gear to ride up and leave the wearer’s torso exposed. Although the U.S. military designed uniforms to fit female bodies, they failed to develop boots that match women’s narrower feet and higher arches. And equipment design is not the only arena where this happens.

In her new book Invisible Women: Exposing Data Bias in a World Designed for Men, journalist Caroline Criado Perez explains how researchers in fields from medicine to transportation fail to collect data on women. This affects aspects of daily life in the home, the workplace and everywhere in between, with results that range from inconvenient to deadly. For example, vehicle safety systems designed and tested based on the default male will not necessarily protect female bodies. Indeed, in a car crash, women are 17 percent more likely to die and 47 percent more likely to experience serious injury than men are.

Scientific American spoke with Criado Perez about why this gender data gap exists and how we can start fixing it.

[An edited transcript of the interview follows.]

What made you decide to write Invisible Women?

It was coming across the data on heart attacks: I discovered that what I had always thought of as just the standard heart attack symptoms—pain in the chest and down the left arm—are in fact male, and women tend to experience different heart attack symptoms. [Ed. note: Only one in eight women will feel chest pain during a heart attack, but they may experience pains in different areas, including the jaw and back, as well as shortness of breath and nausea.] As a result of this, women are being misdiagnosed, so it’s not just a case of public health information not being good enough; it’s also a case of doctors not being trained to recognize potentially fatal symptoms in women.

I was incredibly shocked that I hadn’t known that. We’re taught to think of science and medicine as objective and that facts are just facts. Medical science, it turns out, suffers from the same lack of female representation as everywhere else.

In the book, you include several other examples of how the gender data gap affects daily life, often in surprising ways. Can you talk about the case of snow removal?

The snow clearing is a really useful sum-up of the entire problem. In this town called Karlskoga in Sweden, they were doing a gender audit of all their policies. As it turns out, snow clearing has a lot to do with gender. The way that they had traditionally done it was the way a lot of cities and towns do it: they clear the major road arteries first and then move on to local roads and pavements. And that prioritizes typically male travels over typically female travels because men are more likely to do a commute on major roads and they’re more likely to drive. That’s partly because women tend to have less money, but also because if a household has a car, men tend to dominate access to it. Women are more likely to be using public transport and therefore more likely to be pedestrians as well.

[In Karlskoga] they decided to change it round and do the local roads and pavements first because they figured that it was easier to drive through three inches of snow than to walk or push a buggy through three inches of snow. But what they didn’t expect was that it would also end up saving them money. Suddenly hospital admissions went down: women [had been] slipping and falling in icy conditions beforehand, and [now] they weren’t. If they had been collecting sex-disaggregated data on who was falling in icy conditions and ending up in the hospital, they would have realized that it was mainly women. Then they would have asked, “Why is that?” and they would have got to the snow clearing that way.

We know that women are more likely to die from heart attacks than men are. What are some other ways the lack of female data affects health care?

The problem is that women have different bodies to men and the sex differences that we’ve found operate all the way down to our cells. The evidence that we have so far shows that women’s bodies do react differently to drugs, that women exhibit different symptoms for various diseases, that diseases progress in different ways in women. If we’re basing all our knowledge on male bodies, we’re going to end up not being able to spot and treat disease in women. And that is in fact the case.

Look at adverse drug reactions, which women suffer in much greater numbers than men. The second most common adverse drug reaction for a woman is that the drug simply doesn’t work. And these are drugs that have been tested and work in men.

There’s another study that is highly suggestive in terms of how many treatments we may have missed out on that would work for women. They took male and female cells, and they exposed them to estrogen and then exposed them to a virus. The female cells were able to use the estrogen to fight off a virus, and the male cells weren’t. If they’d just been testing in male cells—which 90 percent of studies are—they would have concluded that estrogen isn’t relevant.

Why are researchers failing to collect data on women?

I don’t think that there’s a vast conspiracy to seriously injure and kill women. The issue is that we’re so used to seeing the male body and lifestyle as just the standard human body and the standard human way of doing things. Medical researchers say things like, “Women are too complicated to measure because women have a menstrual cycle.” Yes, women do have a menstrual cycle—but I don’t see that you would be making that excuse if you were really logically thinking about how that is the body that 50 percent of the population have. You would only make that excuse if you think that women are a variation on men. That is how we end up consistently forgetting to count women.

Some people suggest that algorithms and AI could help eliminate gender bias, but you disagree.

The issue there is that the data sets that algorithms are being trained on are incredibly male biased. We know that algorithms amplify biases: if you feed them biased data, they will become more and more biased. And this is already having an impact. For example, Google’s voice recognition software, which is meant to be the best on the market, is 70 percent less likely to recognize a female voice than a male voice because it’s trained on a voice database that is heavily skewed towards male.

But I am more concerned about the future of these algorithms. They are becoming increasingly important to our lives. They are already, for example, deciding what CVs get through to human eyes in all sorts of jobs. The tech world just seems so blithely unaware of the male bias problem they have, and they keep making these errors that show us they really haven’t got a grip on the situation. Ridiculous things like Apple forgetting to include a period tracker in its comprehensive health app, or Siri being able to find you Viagra but not an abortion provider. They point to a sector that has not got a handle on how to design [algorithms] for women as well as men. Almost the worst part about it is that these are private companies, and so these algorithms are protected as proprietary software—we can’t even examine the biases that are being built into them. I find that very concerning.

How should we solve this problem instead?

I can’t give a very exciting answer—it’s literally just “collect data on women and separate it out from the male data.” On the one hand, that makes me feel really hopeful that this can change, because it is so simple; we just need to have the will to do it. On the other hand, I feel a bit hopeless because it’s not like I am the only person to have ever noticed this. As I said, there are all these incredible researchers already working on this, but there is this reluctance to do anything about it.

The evidence is that unless you regulate for this, it doesn’t happen. If you look at medical research, for example, things are by no means perfect, but for National Institutes of Health (NIH) funding you’re meant to include women in the human stage of trials, and they’ve just introduced that for the animal stage as well. That has massively increased the number of women who get included in studies. But if you’re a private pharma company, which a huge number of clinical drug trials are funded by, or if you’re doing generic drugs, there is no such requirement, and so the representation of women in the trials is much worse. Governments need to recognize that this is a serious issue and that it’s not one that they can just leave up to people to do on their own, because they just won’t do it.

What about nonlawmakers? Can we do anything to help solve the gender data gap?

I think everyone can and should challenge male defaults, to not allow the male to occupy the default space when you’re talking about anything. Like when you’re talking about sports: we say “football” instead of “male football,” but we always say “women’s football.” Part of the problem is that we just don’t realize that we’re doing it—9 times out of 10, when we think we’re talking gender-neutrally, we’re actually talking about men. It allows us to not realize that this is what’s happening. I think that changing that would be quite a significant help towards shifting the problem.