Reading Invisible Women: Data Bias in a World Designed for Men by Caroline Criado Perez was hard. The book is about the patterns of the gender data gap. There are so many that at times reading about them made me feel small and inconsequential. Realizing that large parts of the world are not designed for you, or in some cases are even actively working against you is painful and difficult.
The gender data gap is not malicious. It is the result of a way of looking at the world and documenting it from one particular point of view: that of men. The male perspective has come to be seen as the norm, while the female perspective is seen as a niche. Even though we make up about half of the world population.
The book is filled with examples of where the gender data gap is hurting women. I urge both men and women to read this book. It will change your perspective. At least it did mine. It will also make you more vigilant, which I think is what we all need to be to get to a state where women and our bodies are no longer seen as odd or complicated. After all, we are about 50% of the world’s population.
If I were to mention all the examples in the book, I would be rewriting it. I picked some examples that seemed particularly painful to me.
Crash test dummies, used for testing how car safety and airbags, are 1,77m tall and weigh 76kg. That’s based on the average height and weight of a male body. Women are smaller and lighter on average and are much more likely to sustain serious injuries in a car crash. That’s not because we’re more likely to be involved in serious car accidents, but because cars aren’t as safe for us as they are for men.
In recent years there are some crash tests for which alternate dummies are used, but only in the passenger seat. And the alternate dummy is simply a smaller and lighter version of the male body, so it doesn’t account for having breasts for instance.
This gender data gap is costing lives.
New medicines are usually only tested on men. The reason for this is that women’s bodies might respond differently to medicine depending on their hormone levels. Levels that differ during the month. Because of this testing medicine on women is considered “too complicated”. This is ignoring the fact that once the medicine has been approved for use based on tests on male test subjects women will start using the same medicine. Women’s bodies might respond completely different to the medicine, but these reactions have never been tested.
This gender data gap is costing lives.
The average temperature in an office is based on male bodies and metabolism. It has been scientifically proven that for women to be comfortable the temperature needs to be about 5 degrees warmer. Nobody working in an office will be surprised about this. We’ve all seen men walking around in shirts and women shivering while wearing vests or while hiding under blankets. What’s remarkable is that this is mostly just made fun of and filed under “things women nag about”.
One more, then I’ll stop. Voice recognition systems are trained using large databases of voice recordings. Unfortunately, thanks to the gender data gap, these databases mostly contain male voices. This means that voice recognition systems are trained to recognize male voices. And they do!
These systems are used in more and more places, but a common use is in cars. Using voice recognition instead of using your hands and having to look at buttons or a screen allows for safer driving. Unless it doesn’t work, in which case it might be more dangerous. I’ve experienced this myself (and I have a pretty low voice for a woman). When I had picked up my new car from the garage and was driving home in it I figured I would use voice recognition to call a friend. The system didn’t understand me at all the first few attempts and eventually tried to call someone I really didn’t want to talk to. I nearly crashed trying to abort the call…
Fortunately, the leading voice technology supplier, ATX, has a solution for fixing the issues with women’s voices. According to their vice-president, what women need is “lengthy training” on how to use the voice recognition software. The fact that women aren’t willing to submit to it is their own fault.
Writing this down makes me furious all over again. Research and our surroundings all assume that the default human being is a man. Many of the tools that we use and the places that we visit have forgotten to take half of humanity into account. This is tiring and inconvenient at best but deadly at worst. These are not exceptions. As I said in my introduction, the book is filled with similar examples touching on government policies, pensions, toilets, video games, and language.
It will take a very long time to fix the gender data gap and the resulting biases because they are everywhere. As the use of algorithms increases the problem is likely to get worse before it gets better. This is not an attack on men. We are wired to think that our own experiences mirror those of human beings in general. This is called projection bias. We naturally see ourselves as the center of the world, as we experience the world through our own eyes. There is no way around this. Seeing things differently is hard work. This is true for everybody. And I can imagine that this experience is magnified for white straight men, who constantly see their own experience reflected back to them by the culture in which we live.
The way to fix it is to get more women involved in the designing stage of everything. From software to buildings to policies. After all, women are less likely to forget about women.