| 8 mins read
Bad data make bad science. Gender biased data distort research because it is predicated on the assumption that the human is male. Caroline Criado Perez describes the pervasiveness of this assumption and explains why it matters.
Perez’s book is a tour de force. She argues that male universality, defined as the treatment of men as the default human, causes and perpetuates critical gender gaps and biases in knowledge, in science, in medicine and in popular culture, resulting not only in continuing sexism, but also in costly and repeated policy mistakes.
Most data about humans are biased. All too often only men are counted, populations are not disaggregated by sex, or the ‘scientific’ tests that generated data were performed only on men. Characterised by silences, ignorance, disdain, disregard and misuse, these biases harm women. Perez’s contribution is to identify, describe, bring together and assess the consequences of biased data across most areas of social life.
The universal male
Silence disguises assumptions of maleness. Male experience tends to be seen as the universal experience whilst female experience is niche and/or marginal. A professor at Georgetown University made headlines naming her course ‘White Male Writers’ while numerous courses on female writers go unremarked.
Widespread is the practice of attributing women’s work to men, accompanied by a pattern of attributing men’s errors to women. Examples include the routine attribution to Thomas Hunt Morgan the discovery that sex is determined by chromosomes. It was Nettie Stevens who established this. Cecilia Payne‐Gaposchkin’s found that the sun is predominantly composed of hydrogen, not Henry Norris Russell.
Gender bias in political research
In research on politics, data on political attitudes and preferences have until recently been scarce, resulting in the widespread notion that political woman was a defective male. While sex was a background variable used to control sampling, it was rarely taken seriously in subsequent analysis, hence important differences between women and men were overlooked. Usually, women were subsumed into male data. Standard political discourses and analyses were and, to some extent, still are distorted, considering as they did, only half of the population.
Gender bias at work
The consequences of biased data and data misuse are particularly apparent at work. In modern workplaces, locations, hours, design, and regulation are built around men’s lives. The much‐anticipated new Apple headquarters did not include a day care centre. Companies routinely conflate working long hours in the office with job effectiveness. Glass stairs facilitate upskirting. The recommended office temperature is comfortable for the average man, but too cold for average women (by about five degrees Fahrenheit).
Occupational health research tends to be conducted on men. Protocols on tolerance levels for chemicals are set according to tests run on average sized men, but women are smaller and may be less tolerant for other reasons. The result is various illnesses, including several chemically induced cancers. Tools, machines, and protective clothing are designed for use by men. Space suits and PPE are designed for people without breasts and with penises.
Even when efforts are made to accommodate women, the assumption tends to be that they are mini men. Though women have been in the police and military for a long time, it is still the case that bullet‐proof vests do not accommodate breasts, hence ride up and expose stomach (terrific place to be stabbed or shot).
Gender bias in health research, technology, and town planning
The assumption of maleness also dominates health research. Although sex differences are found in every single organ and tissue in the body, and in the prevalence, course and severity of most common diseases, medical textbooks normally do not offer much sex specific information. Women are routinely excluded from trials for many common diseases.
Diseases or conditions that only affect women, such as pre‐menstrual syndrome (PMS), are chronically under studied. Heart disease research is conducted almost exclusively on men.
Technology and design are no exception. Apple watches were originally designed without tracking software for women’s menstrual cycles. Cars are designed for the average male body. Women are 17 per cent more likely to die and 47 per cent more likely to be seriously injured in car accidents than men. Almost all testing dummies for car safety are designed to copy the body of an average man.
Town planning offers more examples. Poorly lit public spaces are feared by women (about twice as much as by men) who avoid specific routes, adjust their travel patterns, times and modes of travel because of such fears. Yet, map apps do not provide safest route information.
Male dominated public spaces are not natural; they are symptomatic of ignoring women’s preferences. Gyms, parks, and sports grounds tend to be male dominated. So, women do not use them. Officials assume that this is because of a set of female preferences rather than fear. Such space therefore continues to be designed and managed as male.
Nordic research shows that only when sex differences are accounted for are public spaces equally used. Why? There are at least two problems: 1. Failure to consider or find out about women – mainly accidentally, but sometimes on purpose; 2. Not using what knowledge there is.
It seems perverse. You would think that markets might sort this out. So why does it not happen?
The male bias of data collection
The main reason is that women are excluded from planning and decision‐making processes, especially at the early stages. There is considerable resistance to correcting the data biases. Data collection is affected by path dependence according to which previous work determines present and future work. Habits of data collection that treat women as invisible or interesting only in terms of how they deviate from men tend to persist. Data are expensive, investment decisions favour long runs of statistics, and so forth, hence ensuring that they continue with little alteration.
Other problems include biased framing, biased survey design, biased sampling, and biased classification, all leading to flawed data. In short, the problem is systemic. Data collection is not a gender‐neutral process, but one saturated in male bias.
Perez argues that making women present at the problem identification and design stages of decision making is necessary to ensure that better questions are asked, and expensive mistakes avoided. Her book, which won the Financial Times and McKinsey and Company Business Book of the Year Award in 2019, could not be more convincing.
Unfortunately, as Perez notes, men do not read books about women. Hence the bad science and bad decisions she identifies and explains will continue.
Invisible Women: Exposing Data Bias in a World Designed by Men, by Caroline Criado Perez is published by Chatto and Windus. 318 pp. £16.99.