EXCERPTS
The Knowledge Discovery and Data Mining conference is one of the top gatherings for data scientists around the world. Entrepreneurs and academics attend; many Fortune 500 companies send teams, too. The Chinese government has been recruiting there for years. One sign of the conference quality: Only 12% of 2,000 papers were accepted this year.
But there was a big problem, Ankur Teredesai knew, as he took over as co-chairperson: The number of women attendees had remained stubbornly the same, about 22%, for years.
“Academic conferences are very poor at attracting women,” says Teredesai, the founder of a 75-employee health care data firm, KenSci. Teredesai, who emigrated from India, has always been a bit bewildered about why women are seen as less than equal in STEM the United States. At his undergraduate program in India, his class of 75 students had 45 women.
“When I arrived in the U.S., the ratio had completely flipped. It was abysmal,” he said. “There were 10 women in a class of 100.”
“That struck me as very odd and then I entered the workplace. I saw the workplaces and it was quite shocking to see there were very few women, and there were very few people of different orientations and races. I started reading about these issues.
“It really opened my eyes about how messed up it is.”
As he took on leadership of the ACM SIG KDD Conference on Knowledge Discovery and Data Mining.this past year, its 25th, he decided to try to change it. “I made a commitment to myself that no matter what we would get this done.”
Sometimes, an organization’s turning point is getting the right person into a position of influence.
What The Evidence Says
People working on increasing women’s presence in the workforce make the issue seem pretty complicated: It’s about confidence, or about patriarchy, about men not wanting to work with women in the wake of MeToo, or about women not having a good road map for how to advance. I think it’s about all those things.
But moving the needle a significant amount may be much easier then we think. One of the biggest issues is time. Women spend 30-50% more time on domestic and caregiving activities than men, according to the International Labour Organization. In the United States, women spend 187 minutes a day on domestic chores, 42 minutes a day on caregiving, and 35 on community caregiving activities; men spend 122, 19 and 27. If you find a way to take some of those minutes off women’s shoulders, they have more time to work.
That explains why there is such a strong correlation between unpaid work and women’s participation in the workforce – which is important. Raising women’s labor force participation rates worldwide to the same level as men could add USD 12 trillion, or 26%, to global GDP by 2025, according to McKinsey. (Delila Khaled, a senior advisor to consulting firm Laurel Strategies reminded me of these stats and facts).
Back to the story: Working with his co-chair, Teredesai decided that the conference would offer child care. He thought it would boost women’s participation, enable young community members to come to the conference and help attendees focus better.
“It’s quite stressful to find childcare for overnight conferences,” he said. “A lot of people who are very good researchers and stellar researchers, they don’t take the trouble after having a child to come to conferences.”
He and his wife, a research scientist at Microsoft Research, have a 9-year-old daughter.
“So we would take turns hanging out with her and missing out sessions,” he said. “The sessions often go on to 9 p.m. at night, good networking events.”
As he and his co-chair did the budget, they layered in about $20,000 to subsidize the program, which was offered at $10 a day by Vancouver-based On the Go Mobile Event Childcare, founded by Jacob Boursaw. They offered the child care on the registration form, so that people knew it was available as they were signing up.
The conference also installed breast feeding pods throughout the convention centers and had a strong women in data science track.
About 35 kids on average attended the program, which was on site at the hotel (this year in Alaska). Seventy kids used it in total.
Guess what happened? Women’s attendance at the conference bumped up for the first time, to 28% – a six-percentage-point jump. Nearly 75 more women attended this year, a total of 851 out of 3,079.
Does it matter?
Of course it does: There are an increasing number of studies showing how important women’s participation and diversity are in the workplace. 85% of CEOs whose organizations have a diversity and inclusiveness strategy say it’s helped performance, according to PwC’s 18th Annual Global CEO Survey.
Teredesai gave me a specific example of how it helped in his company. KenSci is a 4-year-old firm working on operation challenges for health care organizations. The firm was working on a machine learning model that will predict which patient will leave the ER. (In Chicago, almost 4-5% of patients who walk into ED leave it.)
“While we were building the prediction model, we had a diverse team, two women, one man, and there’s sexual orientation diversity. I didn’t really think about it at that time but someone suggested, let’s see the performance of this model among Asians.
“Then we looked at the performance of the model by age, race and gender.”
“We found out the model was overpredicting the risk scores for African-Americans leaving the ER. There was an inherent bias. The diverse team was able to ask the right questions.”
The KDD conference, meanwhile, will continue offering child care. Twenty-eight percent still isn’t a great number, but a pretty tiny change – one that benefitted everyone at the conference – seems to have a made a difference. A story like this makes you wonder: How often is making a dent in the problem a lot simpler than it seems?