Best Practices for Fostering Diversity & Inclusion in Data Science

January 29, 2019

A blog post by Stuart Geiger and Orianna DeMasi, based on a conversation with Andreas Zoglauer, Diya Das, Fernando Hoces de la Guardia, Kellie Ottoboni, Marsha Fenner, Nelle Varoquaux, Rebecca Barter, Richard Barnes, Sara Stoudt, Stacey Dorton, and Stéfan van der Walt (names in alphabetical order), with additional contributions and links from Aaron Culich.

Introduction

This post is a summary of a recent BIDS Best Practices lunch, in which we bring people together from across the Berkeley campus and beyond to discuss a particular challenge or issue in doing data-intensive research. The goal of the series is to informally share experiences and ideas across disciplines and contexts on how to do data science well (or at least better), for many different senses of the term. The topic for this week was on best practices for achieving diversity and inclusion in data science groups, labs, and teams. This was a joint effort with the BIDS working group on Diversity and Inclusion in Data Science.

In brainstorming and planning for this session, we wanted to acknowledge that issues around diversity and inclusion are complex and multi-faceted. We do not have all the solutions or answers, and we only represented a small slice of the efforts around this topic. There are too many initiatives and efforts around diversity and inclusion to fully detail, even if we limited the discussion to what is happening on our campus --- which is an excellent problem to have.

Another aspect we discussed was that many of the issues we are concerned about require interventions and initiatives at many different levels, some of which we currently have little influence over. Most of us in attendance were not faculty or in senior administration, which is generally the case for our Best Practices meetings. It is important to find ways to involve administrators and faculty in such efforts, particularly around issues like admissions. However, in this meeting, we chose to focus on what we can do in our own research groups, labs, classes, departments, open-source software projects, and other spheres that we can influence.

Whether we recognize it or not, we all have spheres of influence in our own ways, and we have the responsibility to lead from where we are. Seemingly small actions that we take on a day-to-day basis can collectively have big impacts on a work environment. We also acknowledged that we came from quite different academic disciplines and departmental cultures, with some fields being more diverse along certain demographic areas than others. Finally, we began with some definitional work on these terms so that we were all speaking the same language, defining:

  • Diversity: Who is in the room? What is the demographic makeup of a group?
  • Inclusion: What is the cultural climate of a group? Does everyone feel like they are a part of the group once they’re in the room?
  • Equity: Does everyone in the group have the same opportunity to achieve at the same level, no matter where they started?

Fostering inclusive interpersonal dynamics in work environments and communities

One major area that we brainstormed was how to foster inclusive interpersonal dynamics in our various work cultures and environments. A common concern raised was how many of the individual and small group interactions in running meetings, workshops, classes, and talks can be exclusionary, particularly for underrepresented minorities. For example, in meetings, is the responsibility for taking notes and cleaning up equally distributed, or does it typically fall to the same kinds of people? Do a few people tend to dominate the meeting (or a Q&A session), or does everyone generally have the same chance to be heard? Is interrupting or speaking over someone else the ‘normal’ way people in a group must make themselves heard? We discussed several specific strategies for more inclusively engaging with each other, such as: formally delegating rotating responsibilities around taking notes and cleaning up, having agendas for meetings, or speaking up for someone else when they are trying to speak but finding it hard to cut in.

However, these ways to be supportive or an ally in the workplace raised that the work of ensuring an inclusive workplace is also a form of work, which is often not equitably distributed. An important question is if people have do this work independently by themselves, or does everyone in the organization share the responsibility for fostering an inclusive work environment and psychological safety (Edmondson 1999, Edmondson and Polzer, 2016). It may be easier for a third-party to point out a problem than it is for the person who was directly impacted, but it is still difficult and risky to point out that your colleagues has acted in a problematic way. This work of being “the mean person” can be quite exhausting, and it can also typically fall to the same kinds of people. We discussed how it was important to share the responsibility of both reporting issues and responding to them. Having multiple points of contact and modes of raising issues in the workplace crucial, as any single person cannot possibly take on the burden of facilitating an inclusive workplace.

Finally, our discussion of reporting problems in the workplace raised the issue of rules for respect and engagement, or codes of conduct. We quickly realized that we could spend days just talking about rules and codes, whether they are for traditional workplaces like labs, for co-located conferences, or for online communities like open-source software projects. We tabled many of those issues for later, but we felt it was important to have some kind of established ground rules or principles that everyone understands and commits to upholding and enforcing --- as opposed to a kind of ‘click-through’ document that is briefly shown but not enforced. For example, the Social Sciences D-Lab is developing prominent banners in its physical spaces which articulate the “community agreements,” based on work by the Anti-Oppression Resource and Training Alliance (AORTA, 2017) and the East Bay Meditation Center (EBMC, 2015).

Increasing diversity and inclusion in less-representative groups and disciplines

In this conversation about workplace practices, we recognized that the different disciplines we represented were quite different when it came to the existing levels of diversity. For example, in the biological sciences, 51.7% of U.S. Ph.D degrees in 2015-16 were awarded to women, in contrast to 22.8% in engineering and 25.8% in mathematics and computer science (Okahana and Zhou, 2017). Fostering inclusion should be an active and intentional process for all groups, but it can be challenging for research groups that have been more homogeneous on some dimensions, but are trying to increase their diversity of perspectives. Some of the people in our meeting asked how to support new members who do have different backgrounds, when they may not have any members of a particular background already in the group.

One approach is to hire a cohort of people from diverse perspectives, as being the only person from an underrepresented group can be difficult. Yet if this is not an option --- as it may not be in many relatively small research groups --- then working with new members to identify resources for their peer and professional support can be invaluable. Such resources could also include collaborators and co-advisors. We recognized that having open communication channels for both new members and established members to voice issues as the group dynamics evolve is also extremely beneficial. Multiple communication channels and reporting lines to raise and resolve various kinds of issues was also a strong recommendation. We also discussed the value of signaling intention to become a more inclusive environment through such approaches of helping a new member find additional outside resources and opening communication channels.

Organizing and messaging events for diversity and inclusion

We also discussed the many events and initiatives that have emerged to increase inclusive practices on a broader scale than a research group. These have widely ranging intentions, including setting up peer and career mentoring networks, elevating voices of role models, and roundtables or other forums for the discussion of change. Our group expressed much support for such events, but had recommendations to identify a specific and intended goal, then organize and message accordingly. Some of those in our discussion who did not feel they were a part of an underrepresented group --- but wanted to work towards diversity and inclusion --- expressed uncertainty about whether they ought to attend various “diversity” focused events and initiatives. Because of the different goals of such efforts, it is important to set intentions and audiences clearly, specifying them within advertising and event recruitment.

Events that target participation of individuals from not-underrepresented groups are valuable, but should clearly indicate who their target audience is. However, events that let people from a particular background share vulnerable experiences and receive support from people who are like them are also important. Broader, multiple-day events to increase diversity and inclusion should strongly consider whether to include sessions for members of not-underrepresented groups and how to engage them. Such sessions could be town halls to raise issues for discussion, trainings for challenging implicit bias, training in resources and techniques for supporting members of different backgrounds, mentor mixers, and research talks or poster sessions to raise visibility of research contributions.

Long-term efforts to champion change

Individual efforts to champion diversity, inclusion, and equity are important, but we ultimately felt that it was crucial to normalize communication and reflection about issues and inclusive environments. We felt it was important to mainstream discussions of workplace culture in all-hands events and meetings, going beyond holding individual one-off events that can have the unintentional effect of segmenting out diversity, inclusion, and equity issues. In this way, more individuals, including those who may not be members of under-represented groups, can participate and contribute to positive change. Another tactic we raised was that we should find and elevate good examples of teams, events, and organizations that have made excellent progress in diversity, inclusion, and equity issues and/or have thoughtful procedures or approaches for such issues.

Conclusion

As with many of the best practices discussions, we soon found ourselves at the end of our allotted time, but wanting to discuss far more issues. There are many more issues to raise and strategies to explore in the broad space of diversity, inclusion, and equity in academia. We also briefly raised issues around larger societal structural inequalities --- such as that ‘free time’ is not equitably distributed in society (Bianchi and Milkie, 2010) --- as well as issues with admissions and hiring. Overall, we felt that the kind of informal lunchtime meeting organized around what we can do in our spheres of influence was a good approach to raising and mainstreaming these issues.

References and Resources

Anti-Oppression Resource and Training Alliance (AORTA). (2017). “Anti-Oppressive Facilitation For Democratic Process: Making Meetings Awesome For Everyone.” 

Bears That Care, Unconscious Bias Project, and Expanding Potential. (2016). “Addressing Bias in STEM with Bystander Intervention: A 3 Page Handout.” 

Bianchi, S. M. and Milkie, M. A. (2010). “Work and Family Research in the First Decade of the 21st Century.” Journal of Marriage and Family, 72: 705-725. 

Duhigg, C. (2016). “What Google learned from its quest to build the perfect team.” The New York Times Magazine

Delizonna, L. (2017). “High-Performing Teams Need Psychological Safety. Here’s How to Create It.” Harvard Business Review

East Bay Meditation Center. (2015). “Agreements for Multicultural Interactions at EBMC.” 

Edmondson, A. (1999). “Psychological Safety and Learning Behavior in Work Teams.” Administrative Science Quarterly, 44(2), 350–383. 

Edmondson, A. and Polzer, J. (2016). “Why Psychological Safety Matters and What to Do About It.” Google re:Work. 

Okahana, H., & Zhou, E. (2017). Graduate enrollment and degrees: 2006 to 2016. Washington, DC: Council of Graduate Schools. 

Schneider, A. (2017). “Navigating Unconscious Bias.” Talk at DjangoCon EU 2017. 

UC-Berkeley Department of Chemical and Biomolecular Engineering. (2019). “Women in CBE resources.” 

___________________________

Preprint References

Best Practices for Fostering Diversity and Inclusion in Data Science: Report from the BIDS Best Practices in Data Science Series
February 1, 2019  |  SocArXiv
R. Stuart Geiger, Orianna DeMasi, Aaron Culich, Andreas Zoglauer, Diya Das, Fernando Hoces de la Guardia, Kellie Ottoboni, Marsha Fenner, Nelle Varoquaux, Rebecca Barter, Richard Barnes, Sara Stoudt, Stacey Dorton, Stéfan van der Walt

Download PDF (97 KB)



Featured Fellows

R. Stuart Geiger

Ethnographer

Orianna DeMasi

Electrical Engineering and Computer Sciences
Alumni - Data Science Fellow

Andreas Zoglauer

Space Sciences Laboratory
Alumni - DATA SCIENCE FELLOW

Diya Das

Molecular & Cell Biology
Alumni - DATA SCIENCE FELLOW

Kellie Ottoboni

Statistics
Alumni - DATA SCIENCE FELLOW

Marsha Fenner

Communications/Program Manager

Nelle Varoquaux

Statistics
Alumni - DATA SCIENCE FELLOW

Rebecca Barter

Statistics
Alumni - Data Science Fellow

Richard Barnes

Energy & Resources Group, Electrical Engineering and Computer Sciences

Sara Stoudt

Statistics

Stacey Dorton

Administrative/Operations Manager

Stéfan van der Walt

Senior Research Data Scientist

Aaron Culich

Berkeley Research Computing (BRC)
Research IT