Data equity for everyone everywhere all the time!
Rachel Whaley with Yohance Barrett workshopping Applied Data Equity with a UCLA Anderson School of Management Master of Science in Business Analytics class April 2023
An interview with Rachel Whaley, Data Equity Program Manager, by Jen Holmes, nonprofit data leader and data equity practitioner
Jen: I was a grateful participant in the second cohort of LA Tech4Good’s Data Equity workshop in February, 2021. I learned so much, and have incorporated many of the concepts back into my work overseeing a database and analytics team at a large arts nonprofit. I recently caught up with the program’s founder, Rachel Whaley, to find out how the workshops and the community have evolved in the past two years, and what’s coming up next.
What led you to launch the data equity workshop program?
Rachel: I was finding in my own data work that I kept running into a lot of questions and problems that I did not know how to answer and that no one around me really seemed to know how to answer. Even worse than that, no one really knew who should be answering them.
Who decides what the drop down menu options are in the gender field? Or what do we make as a drop down versus a free text field? How do we think about what data we're collecting? If we have customers that are asking for data? To what extent is it our responsibility to be providing transparency as we balance protecting privacy of people's data? To what extent do we take a stance? And as technical data people – database administrators and analysts and data scientists who are in the weeds of the data – what is our responsibility to those questions?
I had an experience as a data analyst where I asked about this, and the question got passed around to everybody without ever getting a clear response. The answer was inertia. We had a binary gender field, and I didn't feel like that was a good answer.
So I started researching what it means to have more ethical data practices and stumbled across data equity as a concept that had never been taught in any of my formal education. I realized that was missing from the curriculum. We launched the workshops to see if there was anyone out there having the same experience, and to find others interested in trying to work on it.
❝The data community doesn't have baseline level ethics❞
It’s a big area of practice, so let’s start small. If you were going to give somebody new to the idea a starting point, what would it be?
Data is not something to be taken at face value. Humans create or design all data: for example, whether an app tracks your location, or whether the company sells that location data to third parties. All of those decisions stack on top of each other.
And sometimes data is presented as sort of objective, or statistical, or facts. Not to say that data can't be true, but it's never fully objective.
Sometimes when people get into this topic, one common response is to add demographic data collection to everything. That impulse is sometimes good, but it’s much more important to think about the context of what it means to collect someone's personal information, what it means to ask people how they identify in terms of race, gender identity, or sexual orientation, or any of these things that are core to someone’s identity and intensely personal.
For those who are marginalized in any of those identities, collecting their data isn’t just a matter of predetermined boxes, nor is it a neutral thing. There's a lot of critical thinking that needs to be done about how we think about data, about our identities, and about people in a thoughtful and expansive way.
As a data practitioner, it can be easy to assume that it’s on somebody else to figure out all the tough questions. To take requests people put into our ticketing system, and just execute the tasks. Or to build an algorithm that spits out a result, without building the system to be explainable and accountable.
Those explanations are very rarely technically valid. I find that much more often, people are unwilling to get uncomfortable, or get in there and figure out why something is the way it is, or even think about whether something should be the way it is.
❝Humans create or design all data❞
So tell me a little bit about the people participating in workshops. Where are they coming from? What are their backgrounds?
We have people coming from all different backgrounds, and the number one thing that continues to surprise me about that is how similar all of the challenges we're facing are. Our participants have quite a spread of tenure and level of involvement with data: from students to junior data analysts to machine learning engineers to senior leaders.
But I would say, by far, the most surprising thing is that we're all sitting there talking about things like, okay, how do we currently collect data? If we get it from a third party, what are the limitations? How can we be critical of sources like census data, which many organizations use, and which, as we talk about in the course, has, many, many, data equity considerations to think about if you're going to use it.
Jessica Lynn Medrano, Susan Tweedy, Eva Pereira, and Clara Hernandez, with moderator Rachel Whaley at our March 2023 Women’s Journey in Data panel.
Are people able to take workshop learnings back and really use them right away?
Part of what we talk about is power dynamics, and who gets to make decisions about things like what data even gets collected or used or what projects get executed. In some cases, we have leaders in the room participating directly. In other cases, we have much more junior level folks who aren't in decision-making roles. Our theory of change for this is that there are three levels of change: individual, organizational, and systemic. There are definitely steps that any individual person who works with data can take to make the practices in their day-to-day work better.
And there's a limit to how far you can get with individual effort, just like with any other sort of social change or movement. At some point, you need buy-in from folks at higher and higher levels to make more systemic changes. So when folks go back to their organizations and try to get buy in they have varying levels of success. And for people who already work at equity or mission driven nonprofits, this fits right into what they're already doing. People are already talking about some of these ideas.
If someone's working at a for profit organization that doesn't have a set of company values or statements that directly address things like equity and inclusion, that's a steeper hill to climb. A great strategy we have seen is folks taking these ideas and tying them to the rest of what the company is doing. A really common argument is, are the data sources that we use reflective of our customers or our clients or our audience or whoever we're trying to serve? And even if it's a for-profit business that has no other focus on equity, it's really hard to argue that the data you're using shouldn't reflect whoever your customer or audience is.
What’s surprised you the most over the first couple years running the workshops?
It's been really interesting to see how many people consider their work as being data-oriented. Cases where if you just glance at their job title, you wouldn't necessarily think that. For example, we had a classroom teacher who deals with data about his students. We've had folks from a variety of healthcare backgrounds who deal with patient-level data, we've had folks who come from a design background thinking about how data is visualized, presented, and communicated, and to whom and for what purpose. I don't know very much about the design space, but a lot of the same equity considerations apply.
Of course, the considerations are a little different if you're a machine learning engineer, versus a middle school teacher. You're dealing with different data on a different scale. But in either case, you might be collecting data about a person’s gender, which means you want to consider, do I make this a free text box? What are my needs for using the data? What am I actually trying to get at here? What am I trying to measure?
How has the program evolved over the years? What's changed?
We've collected feedback from every participant that's gone through every cohort that we've offered. And we really pointedly ask, what of this is helpful for you? What's working? What would you like to see changed or what could be better? We read every single piece of that feedback, and we've made a lot of changes from that. Some of the earlier materials and readings were a bit more theoretical, about how data equity is good for the world. We've moved toward giving more examples and case studies and clear hands-on frameworks you can apply and put into practice right away.
Tell me a little bit more about the custom and specialized workshops.
Feedback from participants led us to offer a data equity and healthcare workshop. And even within that, we had a huge variety of participants who worked in healthcare administration, health care related startups, technical roles, and people working day to day as healthcare practitioners. Again, even within a specific field, there's a huge range of how much people work with data, how technically they're working with data, the volume of data they're working with, whether they're working with algorithms or AI or not.
A healthcare-specific workshop was useful because there's a lot of specific terminology around things like HIPAA, which is a healthcare-specific information privacy law. And there was still a huge variety of people working on everything from, maternal health care, and outreach in non-medical settings, all the way up to state of the art big hospital systems.
We also did a nonprofit-specific session, which was a really interesting space because the nonprofit world has this dynamic of funders and recipients of their services. There are common threads among the sort of data that's required by funders, and what does that mean? And what can you do as an organization, when those requirements don't line up with the data equity goals that you've thought about.
On the flip side of that, funders have resources and power so they can demand any data they want. Participants are often giving up a ton of highly sensitive personal information, and may never see any outcome from it. Or that data is never shared back with them, or they're not really given any context as to how it's being used. These are common threads, even among a wide variety of nonprofits. Thanks to the specialized workshops, we were able to dig in a little bit more into some of those specific conversations.
❝We start from a design justice viewpoint❞
How do you customize the content for the specialized workshops?
The key concepts we're talking about are the same. We talk about starting from a design justice viewpoint, about data equity, data ethics, and then about AI and algorithms. Where it gets a little bit more specific is which case studies we look at. We bring in more specific examples and look at the nonprofit world, or the healthcare world. And the sessions are mainly discussion oriented: over half the time in each session is split into smaller groups. You're talking to your peers, there's a facilitator helping the group chew on some of the tougher questions together, and start to think about how they could apply them to what they're working on. When everyone in the same room is working on similar problems, you can get a little bit deeper into that conversation when there’s already some shared language.
What's your grand vision? The ideal outcome of this work?
A commonly agreed-upon set of standards or a learning pathway for the data field. There's nothing right now that exists that encompasses that whole spectrum of people around what it means to be an ethical data practitioner. What does it mean to work in a way that promotes equity and inclusion and justice. All of these ideas that we talk about in a broader societal sense. For me, the broad vision would be to have this be normal among the data community. To get to a place where there's some sort of baseline professional code of ethics. And if you go to school to study data science, or statistics, or computer science, or engineering or any of these things, there's some baseline level ethics course that you take. Other professions have this, like law, medicine, and accounting. It doesn't mean that bad, unethical things never happen in those professions. But it does mean that everyone has a baseline understanding of where the line is drawn. And the data community doesn't have that. That feels like an absolute bare minimum.
❝One thing the community has facilitated is just knowing you're not alone❞
Let’s end on a discussion about the community, it sounds like you have an amazing community
We are thinking about how to better support folks once they go through the workshop, because our program’s only three weeks long. And if you're trying to create change at your organization, or even just on your team, that takes a lot longer, it can take many, many months, depending on how much of a learning curve it is. So there's this extended period of time where people are often working on initiatives alone.
One thing the community has facilitated is just knowing you're not alone. We've had a lot of one-off connections between people who've met through our workshop or through our Slack group who work on similar things and might be able to spitball ideas, or collaborate on projects. We also have people sharing resources, presenting at conferences, and pursuing more learning opportunities around these topics. This has been really, really cool to see.
LA Tech4Good has more workshops coming up soon. Join our mailing list for announcements.
