(US and Canada) Ever since I can remember, I have been passionate about data. I am not sure if it is the challenge of solving one big puzzle, or the power and information data gives. But the more I learn, the more I get hooked.
When I flashback 25 years ago to my first role in data, I realize that finding discrepancies through analysis just came naturally to me. It allowed my overthinking mind an outlet while creating a massive dopamine hit every time a complex problem was solved. Throughout these years, I was fortunate enough to be mentored by some incredibly smart women in data. Their brilliant minds and knowledge sharing helped pave the way for me to be where I am today. Their contributions do not surprise me since women have been contributing to the fabric of data for quite some time.
One of the first women to make her mark on data was Florence Nightingale. While Nightingale is well known for her heroic efforts in the field of nursing, she is lesser known for her contributions in data visualization. Nightingale was exposed to unsanitary conditions in military hospitals during her time as a military nurse in Istanbul in 1854. She quickly came to realize that most of the deaths in these hospitals could have been avoided if conditions were at the standards hospitals are today.
Nightingale was convinced that improving the sanitation in the medical wards could save up to 10 times more lives, pushing for widespread healthcare reform. She knew she would face skeptics so she began to compile data and produce charts, her first of which was known as the “Rose Chart.” Through her representation of color schemes and patterns, Nightingale’s data collection resulted in thousands of lives being saved after the government deployed sanitation committees to improve hospital hygiene in 1855.
A century later, in 1953, Katherine Johnson became one of the first black women in NASA’s space program. Johnson was always considered a brilliant mind and started her career as a mathematician. She soon left NASA and settled for a teaching job due to the field being dominated by white men.
Initially, Johnson did the mathematical calculations in her head until NASA brought in computer technology later that decade. During that time, computer programming was deemed trivial work, so it was given to the women of NASA to do. These women came to be known and referred to as the “human computers” of NASA. Johnson’s calculations were so accurate that John Glenn asked her to personally calculate his flight pattern into space. Johnson not only broke the gender barriers of that time, but also the racial ones as well. In 2015, she received the Presidential Medal of Honor from President Barack Obama.
If women excel in data and have contributed to the roots of data science, then why is it that only 26.7% of tech-related jobs are held by women?
Not only do women deserve a seat at the data table, but their presence also puts much needed “checks and balances” in the collection of data. Like it or not, everything in the world is interpreted through the lens of the individual who is viewing it, and that includes data. An under-representation of women can cause a bias in policies, as explained in Carolina Criado-Perez’s book “Invisible Women: Exposing Data Bias in a World Designed for Men.” Examples of how biases can actually harm women’s interests are exhibited in the examples below:
Female police officers in Britain have physiotherapy due to bruising by kilt belts in uniforms designed with men in mind. To make it worse, their stab vests only partially cover their breast leaving them unprotected and vulnerable to being harmed or worse, fatally wounded.
A study done by Rachel Tatman revealed that a random sampling of transcripts from Google speech recognition showed a 70% greater chance the commands given by male voices would be more accurate than those given by female voices. In addition, words and phrases spoken by women were captured correctly approximately 47% of the time, compared to men's at a rate of 60%.
Cars have been designed around the male body, which is typically taller and heavier than a female’s. These designs do not factor in the differences in the male and female pelvis, seat belt for pregnancy and a woman’s reach to the pedals.
Smartphones have been designed for a man’s hand, not a woman’s hand which is approximately 1 to 2 inches smaller in circumference.
Once a data set becomes biased, it can be very difficult to correct. For example, Amazon had to shut down its new AI recruiting tool in 2017 for not rating candidates for technical jobs based on gender-neutrality. Amazon’s system penalized resumes with words and phrases containing “women,” and downgraded resumes including women’s colleges. The program, developed mostly by men, was trained to vet applicants by observing patterns submitted to the company over a 10-year period. Amazon’s example demonstrates how male dominance in the tech industry can create machine-based bias.
According to a study by Massachusetts Institute of Technology (MIT) across various facial-recognition software, a major U.S. technology company claimed to have a 97% accuracy rate, but was really only accurate when it came to the data set it was designed against. The data set used to assess information was more than 77% male and more than 83% white. In addition, MIT confirms that these algorithms work better on lighter-skinned faces, citing an error rate in the software of 34.7% for dark-skinned women compared to an error rate of 0.8 for light-skinned men. This is why it is critical that more women, especially minority women have a seat, or better yet, build their own table when it comes to data.
The good news is that, although women represent 26.7% of the tech workforce, the numbers are steadily increasing year after year. In order for those numbers to continue to rise, young women need to be given opportunities to build on foundations that develop into data-driven careers. Currently, universities such as Stanford are building programs that focus on women in data science. Several government programs and non-profits are spinning up initiatives as well in order to level the data playing field.
Having an inclusive lens will help to prevent unconscious bias and allow for more diversity in the collection and analysis of data. By recognizing the gender data divide and empowering more women we will not only reduce bias but also reduce the risk that these biases will enter the data value chain.
About the Author
Jennifer Mezzio is the Managing Director and HR Divisional Data Officer at Silicon Valley Bank (SVB), responsible for developing and governing Human Resources’ global data and information strategy across the enterprise. SVB is the financial partner of the innovation economy – helping individuals, investors and the world’s most innovative companies achieve extraordinary outcomes through the power of integrated financial services.
Mezzio joined SVB in 2017, leading the Business Requirements CoE for the enterprise until 2019 when she moved into Human Resources. She is currently responsible for building and executing the HR Data Office, jumpstarting HR’s data efforts and transforming HR into a data-driven organization.
Mezzio brings over 25 years of experience in IT strategy, business/systems/data analysis, data warehousing, data governance and management, and program/project management. Mezzio earned her master’s degree in Organizational Leadership and Computer Information Systems from Quinnipiac University in Hamden, Connecticut.