The Gender Data Gap: A World Built for Men?
By: Yewon Lee
No, not the gender wage gap, but rather the gender data gap. To put this into perspective, here are some prime examples of flaws in data which designates the male as the “typical” standard, especially in science and engineering, and how this may affect our future:
>>More men are in car crashes, but when women are in a crash, they are 17% more likely to be killed, 47% more likely to be seriously injured, and 71% more likely to have moderate injuries. Ever since crash dummies were instituted in the 50’s, they have been based on 5’10 and 168 pound “average male” proportions with male muscle mass and spinal column, which made most cars designed for men. As females tend to be shorter than 5’10, they sit closer to the front of the car, at risk of more injury from frontal collisions and leaving the lower body unprotected. In fact, “female” dummies were not used for research until 2003 and is simply a scaled-down male dummy representing less than 5 percent of the female population with a height of 5'0 and 110 pounds (proportions of an average 12 year old child) and female physical differences in the neck and torso not taken into account. Even then, much of the research with female crash dummies were used in the passenger seat, not the driver’s seat.
>>Women’s heart disease is misdiagnosed 50% more than men, and female symptoms are considered “atypical” relative to “typical” symptoms more commonly found in males. In the past, research was almost exclusive to male patients by male scientists which found that “typical”, well-known symptoms included chest pain, sweating, tightness, and dizziness, while women were more likely to have indigestion, nausea, and pain between the shoulders, all symptoms that can be dismissed as the flu, aging, or common cold despite heart disease being the leading cause death in women in the US.
>>In March 2019, four days before astronauts Anne McClain and Christina Koch were to make the first all-female spacewalk, plans were abruptly cancelled as NASA had built space suits of only three sizes based on the average male, and consequently did not have enough safe, well-fitted suits for all the women. A male astronaut replaced Anne McClain in March and the first all-female spacewalk was not completed until October.
>>Personal protective equipment (PPE) design and research are based on the male populations of mainly Europe and the U.S., leading to women and ethnic minority men to have difficulty finding fitting equipment from construction hats to eye protection to police vests. Although many assume that women can simply use a scaled-down, smaller version of the existing “male-based” PPE, they do not account for equipment such as harnesses and vests which do not take physical differences in the chests and thighs into consideration. For instance, a female police officer was stabbed to death after removing her vest since it was too difficult to use a hydraulic ram with the vest on; other female police officers have reported needing breast-reduction procedures because of health issues stemming from the improperly fitting PPE.
What’s so significant about the world built for men is the concurrent technological revolutions made in fields like artificial intelligence (AI). When we are utilizing more and more AI and algorithmic technologies, the gender data gap gets much more serious. Algorithms are only useful based on the information and data that we feed them, so consistently using exclusively male data and little to none female data furthers the magnitude of the problem. In the U.S., 72% of curriculum vitaes were not reviewed by humans, but by AI programs that could very much have algorithmic bias; on the same note, Amazon recently terminated an AI program that favored hiring men over women from the data that was used to develop the program, and we may not know of other cases in the future as these algorithms are protected under proprietary data for private companies. And with AI being used for the first time in medicine, specifically heart disease prediction, using male dominated data despite the differences in disease manifestation, we may not know how accurate it is to predict female cases. In the future, as AI technologies become more prominent, we need to ensure that the data is collected and utilized in a representative manner.
We like to think of science, medicine, and engineering as relatively static and unbiased because “data is data”, but this mindset can have significant impacts in how we operate now and in the future. As the next generation of researchers, physicians, AI developers, etc., we have to keep an open mind and be aware of the gaps and biases in our data. Ultimately, the problem may be on us to solve and for us to progress our future research and design methods for the generations to come.
Actions
Fully investigate the data gap and its effects. Researchers must actively look into collecting sex-disaggregated data and seek diversity within the research setting.
Research the effects of diversity. One recent study found that male doctors’ performance improved when working with female colleagues; there must be additional studies in other settings outside of medicine that demonstrate the positive effects of diversity and the various perspectives it can bring to the workplace.
Actually implement diversity. A significant portion of the data still being used today has been collected since the days of extreme racial and gender segregation which inevitably led to white male scientists and engineers studying male focused topics. Today, much of the data gap bias comes from the lack of various perspectives in science and engineering and inherent human bias. With women representing ~50 percent of the population, there must be work done to actively promote diversity within these settings which can translate into better products and can close the data gap.