The Data Science Infographic: Is “Being a Man” a Qualification?
What is the Data Science Infographic?
Ask any data scientist if they’ve seen a “What it takes to be a data scientist” infographic, and they’ll answer in the affirmative (and probably tell you about their side project of collecting and classifying them with a neural net). Because the title data scientist has yet to settle on a universal definition, folks are continuously dreaming up snappy ways to communicate the skills of our trade (I’m just thankful we’ve expanded past the Venn diagrams). The recipe is simple: a drawing of a geek in glasses and a (sometimes comically elaborate) list of skills.
Last night I attended San Diego’s regional Women in Data Science (WiDS) Conference 2018 event in San Diego. A speaker shared one of these beloved infographics, and I immediately noticed something special about it: it depicted a woman. Why did this strike me? I’d never encountered a woman in a data scientist infographic before.
A Little Data Magic
I’m a scientist, so my experience isn’t enough to determine if this is a general trend or just a function of my unique life experience. Instead, I conducted a 100% non-scientific (i.e., magic, which is literally the most popular antonym of the word science) review of these infographics. I collected all of the images on the first page of Google and Bing image search results for “data scientist”. We can all agree that folks will rarely go past the first page. I searched under Incognito in Google, but found via another non-scientific poll that these results don’t seem to vary by user, unlike the personalized search results we’re used to with Google.
Utilizing the cutting edge Viola-Ocular-Cerebrum algorithm (i.e., my own eyes and brain), I developed a model to classify the images into those depicting men, women, or both (and in two cases, beautiful unicorns that defy such archaic classification). This yielded a sample of 222 infographics and photos that these search engines put forth to represent data scientists.
These images are communicating more than technical requirements about coding and statistics skills, something not explicitly stated: data scientists are men.
71% of the image results show men and only men. Men get the to see themselves depicted in 85% of these search results if we also include images with both men and women. Women on the other hand are depicted without men in only 15% of the images, and show up at all in only 29%. It seems that my experience of having never encountered myself in such a depiction isn’t an outlier — it’s representative.
What’s worse? There were approximately 10 of these images that depicted people of color — that’s less than 5%, folks. This is not at all representative of the actual workforce which is usually estimated to be less than 50% white.
Why it Matters
Repeated exposure to these images has a cumulative effect. I feel this in myself when I have moments like the one at WiDS — moments of excitement when a woman takes the stage at a technical conference or when I see a woman represent my skill set in an image. It has a noticeable impact on my confidence and my perception of myself. It matters for me to see myself represented in depictions of my community — makes me feel like part of the gang. It is small and subtle, but it’s also real.
I also see it in little embarrassing moments we all want to avoid. For example, when I attended a professional event with a colleague who is a man, but not a data scientist, and folks greeted him as the scientist and waited for me to identify my purpose there. Not a single person at that event looked at me and actively thought,
“If woman, then not data scientist.” — Nobody ever
Instead their brain took a shortcut without their permission, based on information it collected from infographics and workplaces. That’s alright, but we can do better with the smallest amount of extra thought.
One could argue that 25 to 35% of data scientists are women, so what’s wrong with the images being representative? I would argue that this number would be higher were it not for small and subtle things just like this. Low representation is both a cause and an outcome of low representation. We won’t have a more diverse workforce if we can’t take the first step and envision a more diverse workforce.
More problematic, data science is absolutely not 95% white. Can a designer please make me an infographic depicting my many existing colleagues from India, China, and the rest of Asia? That part isn’t even aspirational. Even a friend who tried this exercise in Indonesia pulled up a whole page of white people.
Next time you want to share one of these gems, let’s reward the folks that took the time to be intentional about how they represent our field and who we want to be. To aid you in that, I’ve compiled a quick list of winning infographics for your consideration. If you’ve got more, put in the comments! If I need to update credits, put it in the comments!
2. DataCamp — They are winning at so many things lately, and now infographics!
3. AnalyticsVidhya’s beautifully gender neutral unicorns (we can agree to disagree on if they are hipsters or gender neutral, but let’s all admit there is substantial overlap these days).
4. This random French advertising site. I wholeheartedly disagree that you need a degree from a prestigious university to do data science (also 15,000%?!), but I love everything else about her.