What's New

Q&A with Heather Krause, founder and principal data scientist at Datassist

Heather Krause's Toronto-based data analysis firm works with nonprofits, transnational corporations, government agencies and non-government organizations to explore some of the most pressing questions facing societies today. In 2016 Krause founded We All Count, a project under the Datassist umbrella that works to promote equity and ethics in data science.

Interact for Health: What prompted you to launch We All Count?

Krause: I started We All Count five years ago instead of quitting data science. I was about to walk away from an industry and a discipline that I saw doing active harm to some of the most vulnerable people in the world. Often this harm occurred despite our mandate to help them.

Under a banner of false "objectivity," I was enabling the most privileged people in every system to embed their own worldview into each piece of data we collected. I took the analyses and interpretations we had shaped through our own perspectives and held them up as objective fact. I would have looked people in the eye and said, "'m sorry your experience doesn't match up with our results, but the numbers don't lie." It didn't make sense. It didn't feel good.

So I wrote my resignation letters. As part of my transition out of these projects, I was looking for other data scientists to replace me and that's where I got stuck. I didn't think anyone else out there would have a better time with the inequity issues, because they aren't personnel problems, they are systematic. There were fundamental flaws with how data science--a tool developed for use on things--had been applied to people. With the advent of big data and machine learning, those problems were about to be reproduced at a scale that might be impossible to fix.

I had also seen the good that honest, equitable data science could do. If used carefully, ethically and transparently, data science can reveal injustices and provide solutions. It really can be used to improve everyone's material well-being and personal dignity. I felt that quitting was a quitter's response.

So here we are, five years into the We All Count Project for Equity in Data Science and I get to be a happy warrior for fairness instead of a despairing onlooker. I found out that there were thousands of people in the data science community who cared deeply about the people their data came from. I meet more every day.

Interact for Health: Why is it important to take an equity approach to data, research and evaluation projects?

Krause: It's vital because many of us are taught to believe that data, research and evaluation, particularly quantitative aspects of these, are objective, unbiased, and free from world views. That "the data doesn't lie," and this is simply not true. Research and evaluation cannot be free from bias and evidence cannot be without a world view. It's critical to get honest and transparent about this.

Interact for Health: What is a practical first step to take if someone is interested in applying an equity lens to data?

Krause: One of my favorite first, practical steps is to look at the research question or learning agenda that is the core of your data work. Ask yourself three questions about that research question. One: Where is the onus to change being placed in this question? (Hint: It is often on the marginalized or oppressed people.) Two: What is the implicit definition of success in this question? (Hint: It is often what the most privileged people think of as success.) Three: How can I adjust this core research question to reflect a more equitable approach?

Interact for Health: What are some of the barriers to taking an equity approach to data projects?

Krause: The biggest barrier, in my experience, is lack of awareness that an equity approach is necessary. A close second is the idea that we need to choose between robust, reliable science and evidence or fair and equitable science and evidence. This is most definitely a false dichotomy. The two are the same thing.

Interact for Health: Can you describe ways you have seen people overcome those barriers?

Krause: There are lots. I'd like to point to the work of Abigail Echo-Hawk and Yeshimabeit Milner. These two have been fantastic living examples of how to overcome and directly address these barriers.

Interact for Health: How can we ensure that equity becomes an integral part of data science and not an afterthought? 

Krause: Probably the best ways to ensure that equity becomes a non-negotiable in data is to demand it as consumers and citizens and data creators. Without us, there is no data.

Interact for Health: What excites you most about the field?

Krause: Right now what is most exciting to me is that the sheen of objectivity is falling away from data and tech. The bar is really being raised on how we understand and use data, particularly in the social sector. This is an exciting time to have the opportunity for an entire sector to rise to the new, higher bar.

Return to What's New