An actionable guide to supporting women in data science
Life experiences, career trajectories, and lessons from three leading female data scientists
The 2021 Global Gender Gap Report by the World Economic Forum found that women make up only 32.4% of professionals in Data and AI and 14.2% in Cloud Computing. Even though more women than men graduate college in the US, we still see a considerable gender gap in STEM fields. So why aren’t there more women in data science, and what can we do about it?
This article summarizes a conversation with Dr. Andrea Jones-Rooy, Dr. Angela Radulescu, and Marianne Aubin Le Quéré hosted by the Women in Science at New York University in April 2022. You can find their full bios and more resources in the event announcement link below.
Jump to the last section for tips on how to start making a difference!
Contents
- Meet Andrea, Angela, and Marianne
- The journey into data science and its challenges
- Personal experiences with gender inequality and what we can do about them
- How we can encourage women to pursue data science
- What you can start doing today to improve diversity in data science
- Resources
Meet Andrea, Angela, and Marianne
Dr. Andrea Jones-Rooy (they/them) is an NYU Data Science Professor and the Director of Undergraduate Studies at the Center for Data Science.
Andrea consults for companies that want to measure talent using data on employee performance. They often find themselves explaining to the companies that the performance data is actually internalized biases that the company has turned into numbers. Andrea helps companies develop better metrics that promote equality in the workspace.
We tend to put numbers on a pedestal without first interrogating where they came from. — Andrea
Dr. Angela Radulescu is an Assistant Professor at Mt. Sinai’s Center for Computational Psychiatry in New York and was an NYU Moore-Sloan Faculty Fellow at the Center for Data Science in 2022.
Angela came to data science from psychology. She’s fascinated with human behavior and works primarily in cognitive science studying human learning and decision-making. She uses reinforcement learning algorithms to peek inside people’s minds.
I like to pretend that my path was linear, that I knew what I was doing at every given point in time. But I was subject to the same stochastic nature of the algorithms I study and arrived at a topic after sampling different things. — Angela
Marianne Aubin Le Quéré is a Cornell Information Science Ph.D. student. Marianne describes her department’s origin story as a bunch of computer scientists and people across disciplines who wanted to integrate the study of people into their work more deeply.
Marianne does computational social science: she’s motivated by social questions primarily from communication and journalism, and she uses computer science approaches to answer them. She has also co-founded Graduate Students for Gender Inclusion in Computing at Cornell.
Sometimes you need to talk to people to find out the right questions to ask before using any data scientific approaches. — Marianne
The journey into data science and its challenges
When Andrea started a PhD in Political Science, they wanted to write “big, sweeping essays on the nature of war and the human condition.” Instead, they had to sit through linear algebra. It turned out their program was very computational science-heavy, but Andrea stuck with it out of stubbornness.
As a kid, I didn’t think I had it in me to do science. I didn’t consider myself particularly proficient in math, Bunsen burners, and lab coats — things that I thought of when I thought of science. — Andrea
Andrea says their biggest challenges were self-esteem and impostor syndrome, especially in graduate school.
I was lucky to stumble into a science program because otherwise, I would have ruled out my own life experience. — Andrea
In addition to teaching data science at NYU, Andrea is a business consultant, a comedian, and a circus performer. They self-deprecatingly confess that a lot of things fall by the wayside.
How do I do it all? I disappoint everyone and never see my friends and family. Everything I do is late. My reply rate to emails is one to eight, you know, in a good year. — Andrea
Unlike Andrea, Angela went to high school in Eastern Europe, where there was “a lot of math and engineering, but you were either under the line or above the line.” It created a love/hate relationship with math for her, where math performance was deeply tied to self-esteem.
So when Angela went to college in the US, she chose to avoid math and pursue subjects like journalism. She eventually fell in love with her psychology classes, which, to her surprise, had a lot of math.
The brain can be described in mathematical terms. — Angela
During her Psychology PhD at Princeton, she ended up in a very Machine Learning-heavy lab.
This is where my self-esteem was at its lowest. Many people around me had tons of machine learning skills, and I had no formal ML training in undergrad. But I stuck with it. — Angela
A big turning point for Angela was when she got over her fear of programming. From that point, she decided that she could learn anything that came her way. She taught herself machine learning and coding. She says it wouldn’t have been possible without mentorship.
I think back on it now, and I remember how hard it was to convince myself I could do those things. But with a lot of practice and help from a thousand great mentors and teachers, my mindset around the whole process changed. — Angela
As a high school student, Marianne could not decide what she wanted to do, taking subjects like English and history but also math and physics. She gravitated the most towards English but initially decided to make her college education as broad as possible. She took a computer science class on a whim as a sophomore at Brown and fell in love with the act of programming.
For some reason, for me, doing math with computers is much easier than doing math with pen and paper. — Marianne
Since then, she was always looking to combine English and computer science. Marianne spent several years working a product manager job that allowed her to do just that while completing a master’s degree. However, working in advertising, she soon started thinking about the ethics of her field.
She decided to pursue a PhD that would allow her to combine English and computer science in her research while exploring the ethical considerations and other questions that come up in communications.
Anxiety and imposter syndrome have been my biggest demons. — Marianne
Although too much self-doubt can be devastating, Andrea says that a healthy dose of doubt in your work can be good, especially in science.
Being a good scientist means being very skeptical and doubting most of what’s presented to us. — Andrea
Questions like “Do I know what I’m doing? Is this the test that I should be running? Is this the right way to do this?” can feel like a personal shortcoming. But they are reasonable to have while doing science.
Sometimes I try to remind myself that having doubts about whether the work is good — if it’s not too much — is helping me do better work. I’m trying to turn some of these doubts around as a strength. — Andrea
Personal experiences with gender inequality and what we can do about them
Andrea, Angela, and Marianne all reported having struggled with self-esteem and imposter syndrome at some point. But how much of it is internal, and how much of it is external, due to how a white male-dominated field can treat minorities?
Right out of college, the product teams I worked with were engineers in their forties who were almost all men. I had many experiences that I thought were microaggressions and microexclusion. I would suck it up for months. — Marianne
Marianne says that reflecting on how the situation made her feel and confronting the person has helped her deal with the personal experiences of gender inequality at her company.
Once, somebody said something that I thought was very infantilizing. I brought it up to him, very frightened, afterward. To me, that was something that brought personal closure and set us up on an even playing field for the future. — Marianne
Angela agreed with Marianne. She has also experienced microexclusions, especially in undergrad and graduate school. But back then, there was not as much language to describe what she was experiencing.
Throughout graduate school, I went from an internalizing state of mind to more externalizing. I started thinking that there might be something off in the environment. But it took me time to find the language to talk about it. I was in a very male-dominated, top-down environment, and, at the time, the language didn’t exist. — Angela
There was one situation that stood out in Angela’s mind. She was giving a talk together with a male collaborator with equal project ownership. After the talk, people, even women, were misattributing ideas and quotes that she said to the male collaborator.
These are very low-level subjective biases that people are not always aware of. — Angela
Since then, Angela started to amplify and repeat what other women say, especially when a group passed over something they said.
It can often feel like you’re transgressing. But you have to do it. Because if we don’t do it for each other, the bias will remain. — Angela
How we can encourage women to pursue data science
For Andrea, an essential part of combatting gender inequality is increasing the visibility of women in data science and showing girls what a scientist can look like and do.
Seeing women and minorities in lab coats holding Bunsen burners is fantastic. But it’s also probably helpful to see that you can do science in many different ways. Ways that aren’t a sterile lab or something from Breaking Bad. — Andrea
Another part is creating more outreach programs for middle and high school students. Growing up, Andrea felt science was just a collection of facts about whales and double helixes. It wasn’t until graduate school that they started figuring out how science works and getting pleasure out of it.
If young girls could get experience doing data science projects, being engaged in a problem and trying to reason how to solve it, they could see science as something more than baking soda volcanos. Luckily, there are many introductory tutorials available online. But we still need to encourage girls to try them out.
One of the great things about data science is that the data is online. The tools are online. You can just start doing it. — Andrea
Angela mentioned the concept of stereotype threat from psychology. Studies find that women perform worse on math exams if you remind them of their gender.
We are conditioned to have an association between women and bad at math. — Angela
It’s essential to break this conditioning. Getting out there and showing the girls what we do and how we look, allowing them to ask questions about the life of a scientist, can help them associate women with science, data, and math.
For Angela, mentorship is one of the most important ways we can do so.
We should take seriously our roles as mentors and allies. — Angela
Marianne suggests that academic institutions and companies should create formal incentive structures that reward people for doing things that promote diversity and inclusion. Right now, academia rewards papers and citations, not gender inclusion initiatives.
I put a lot of energy and effort into championing gender inclusion. But I do it on top of my work rather than as a part of it. — Marianne
But formal offices with funding, recognition, and the responsibility to champion diversity can help. They can sponsor and publicize events, bringing monetary and reputation benefits to the organizers. They can turn diversity efforts into conferences or papers and other things that are traditionally rewarded in the academic and corporate worlds.
Institutions must spend more resources and time establishing proper systems and incentive structures. Like the title of Laura Bates’ book reads, Fix the System, Not the Woman.
Another problem is when decision-makers don’t think there is a diversity problem. Sometimes it is hard to convince the administration that there are issues with small data sets. But there’s just not a lot of data on, for instance, Title IX, and the data is extremely sensitive.
Marianne explains that turning individual experiences into powerful narratives to affect change is hard. But we need to be creative and keep trying. For example, the gender inclusion group Marianne is a part of has developed projects to create art from stories of sexual harassment victims that express the emotions and frustrations of the situation.
Things you can start doing today to improve diversity in data science
- Amplify female and minority voices. Acknowledge, repeat and act on what minorities are saying.
- Stand up for yourself and others. Communicate your feelings when you think someone is talked down to, excluded, or made fun of based on gender or any other identity.
- Show that science is not just about lab coats and Bunsen burners. Science can look like coding, interviewing, writing, reading research papers, and much more.
- Learn to associate women with science. And teach others to do the same. Read about female scientists, watch documentaries and movies promoting inclusive narratives, follow women in STEM on Twitter.
- Become a mentor. We need to support and encourage each other. If you’re a college student, mentor high schoolers. If you’re a grad student, mentor undergraduates.
- Create structures that reward diversity and inclusion actions. Often there are very few professional rewards for working on diversity initiatives. We need structural change for meaningful impact.
- Turn individual stories into powerful narratives. Some effects that don’t reach statistical significance in your sample still exist. Help make them seen.
Resources
- Picture a Scientist is a documentary about women in science.
- Data Feminism by Catherine D’Ignazio and Lauren F. Klein is a book about feminism and data science.
- The Dissemination of Culture by Robert Axelrod and The evolution of cooperation by Robert Axelrod and William Hamilton are some of the papers that made Andrea appreciate the value of math.
- Stop asking women how we manage work-life balance. Most of us don’t
- Stop telling women they have imposter syndrome
