Invisible Women: Data Bias in a World Designed for Men

Originally published in 2019, Caroline Criado Perez’s Invisible Women: Data Bias in a World Designed for Men exposes the fact that despite differences in biology and societal roles, men are used as the standard default in virtually all matters from product design, to medical testing, to the tech industry, which creates algorithms that discriminate against women. Perez, a feminist activist and student of feminist economics, has the credibility to detail this problem in this work of non-fiction.

The lack of data about women’s needs, preferences, life experiences, and concerns causes everything from inconvenience to physical harm and death for women from everyday products, living areas, transportation systems, the workplace, and other situations. Women’s contribution to society is overlooked and undervalued: Unpaid work, 75% globally of which is performed by women, is not included in the calculation of gross domestic product. Male violence against women is not accounted for in systemic designs, making the world less safe for women. The excuses for not including half the population in data statistics and planning are unacceptable.

The book is an international bestseller. It won the 2019 Royal Society Science Book Prize and Financial Times Business Book of the Year, was a finalist for the Los Angeles Times Book Prize and Orwell Prize, and was longlisted for the Andrew Carnegie Medal for Excellence in Nonfiction.

This guide references the paperback edition published in 2021.

Please note: Perez describes the derogatory terms aimed at women in the political arena, and this guide provides one example of an offensive term to be true to the book’s content.

Summary

Women have long been absent from or grossly underrepresented in most fields and excluded from power: There is male bias in evolutionary theory and even language itself; historically, women’s accomplishments were hidden from public view or claimed by men. The contemporary obsession with data and data-driven decision-making makes the absence of women of special concern in the 21st century. The underrepresentation of women in data has the potential to bias decision-making more than ever.

First covering everyday life and the workplace, Perez demonstrates how reliance on the male standard inconveniences and harms women. Women’s interests are not considered in city planning and public transportation—the design is with a man’s typical commute from home to work in mind, with housing situated a great distance from work. This makes fulfilling women’s caretaking responsibilities and retaining paid employment difficult. The safety of women is not accommodated either: For example, transportation systems and toilet facilities can be dangerous places for women. Women perform the bulk of unpaid work, but this is not considered in the structure and schedule of the paid workplace. Women frequently have to reduce their paid work hours or leave the paid workforce because of their care responsibilities. They pay a heavy economic penalty for doing so, as part-time work is paid poorly and the economic loss is reflected in their pensions. When women seek paid work in high-status fields, such as technology, they face discrimination and a hostile work environment. There is a brilliance gap, with female geniuses forgotten. Not surprisingly, women do better with blind reviews.

The physical safety of women is at risk in the workplace too. Occupational research is much more likely to have been done in male-dominated fields, making those jobs safer. Protective equipment is designed for a typical male body and frequently does not work fit women. Conversely, in female-dominated fields, there is a lack of adequate occupational research—and undoubtedly, women are dying as a result. Moreover, poor women are especially vulnerable to exploitation and sexual harassment on the job.

Engineering, architecture, and medicine claim to use science to ensure objectivity and precision. Yet in these fields too, women are omitted or greatly underrepresented in planning. When products are made primarily or solely for women, male engineers fail to consult women, so the resultant products often fail to fulfill women’s needs. Other products are generally designed for the standard male and therefore do not work well for women: Voice recognition software, for example, is not as responsive to female voices, while the omission of women from automobile crash testing data can negatively impact female health and lives.

Artificial intelligence is subject to the same male bias. When female designers seek funding for products designed for women, they are met with skepticism from male investors. Given the lack of data on women, such designers are also at a disadvantage in making the case for their potential products.

With a well-documented history of discrimination against women, the medical system continues to favor the male standard. Textbooks overwhelmingly use male bodies. Women are underrepresented in clinical trials and comparatively little research has been done on maladies particular to women. When women exhibit symptoms unlike those of men, they are often dismissed. Physicians were once likely to banish women to asylums for behavior inconsistent with femininity; now they are more likely to prescribe sedatives to women and pain killers to men.

In the realms of politics and economics, women are at a significant resource and power disadvantage. Performing 75% of the world’s unpaid but essential labor, women get no credit for that work in the calculation of a nation’s gross domestic product (GDP). Deeming this omission the most significant data gap, Perez explains that the invisibility of women’s contribution allows legislators to cut social programs with ease. When that happens, women have to assume more of the unpaid workload and as a result, they reduce or eliminate their paid working hours.

Since tax systems do not properly measure the effects of incentives on women, they all too often enrich men at women’s expense or create incentives for married women to do less paid work. That is a loss to the nation’s productive output. When women are elected to political office, they are more likely to increase social spending and consider the female perspective. The road to election, however, is perilous for women. They are held to a double standard that requires them to behave like men, but to be penalized for so doing: Men are praised for their ambition, while women are criticized for it. Once in the public realm, women face a hostile environment, with sexual harassment and threats of violence increasingly common. As a result, many women exit the public sphere or refrain from entering it in the first place, compounding the problem of underrepresentation.

When political and economic systems break down, whether in conflicts or natural disasters, women pay a high price. Their interests are indefinitely put on hold. They are susceptible to rape, a common tactic of war, and gender restrictions often place women at greater risk of death. In some cultures, for example, they must await a male relative before fleeing a dangerous area. What is more, refugee camps and post-disaster situations do not consider women’s safety from male violence in their design and management.

Perez calls for the collection of sex-disaggregated data. Only armed with such information will systems serve women’s interests.