Reflections on the 2023 Data Science Leadership Summit
July 13, 2023
In May, I had the opportunity to attend ADSA’s Data Science Leadership Summit, as a representative of the Career Development Network Steering Committee. The Leadership Summit is typically a meeting of the heads of data science institutes and academic programs (research institutes as well as degree programs). This was overall a fascinating experience for me, having previously attended a meeting that was co-located with the first leadership summit and watched all the institute heads disappear into another meeting. I’ve since transitioned to an industry role, but remain involved in the Career Development Network because I believe in broadening the array of career options available to graduate students and postdocs. This Leadership Summit was one of my first opportunities to return to an overwhelmingly academic setting as someone who has now been in industry for several years (and not doing what my former colleagues would consider research).
Topics throughout the summit spanned quite the range in just the pre-planned sessions, which included comparison of data science programs and institutes’ organizational structures, faculty hiring, workforce development, ramps and pathways in data science created by non-R1s, and industry partnerships. However, several themes echoed across multiple sessions and talks, and it’s those takeaways that I’ve highlighted below.
More than a few people are discussing how to ensure data science education is more accessible and reflects peoples’ lives outside the classroom
The first day of the summit, I walked over to Boston University’s new data science building and was immediately greeted by both some familiar faces, as well as some people whom I’d only met on Zoom. The summit opened with a keynote by Afua Bruce, who asked us “where are we now and where are we going?” Dr. Bruce specifically asked us to consider how data science is being integrated into not only classrooms, but also the world at large, and who gets to participate. She presented several case studies of how data science research that has tangible community outcomes can begin in the classroom – and also prepare students for the next steps outside the classroom.
Zarek Drozda (Data Science 4 Everyone) brought up the importance of data literacy even for non-data-scientists on Day 2, as part of the ramps and pathways session. Fellow panelists Ben Baumer (Smith College) and Helen Burn (Highline College) talked about removing administrative roadblocks to allow students to transfer data science course credits between educational institutions, with Dr. Burn again echoing a challenge to make data science education both relevant and accessible to high school students. An emerging challenge noted by both panelists is that high school students make choices that force them onto “tracks” that may not transfer and preclude them from further studies in data science – thus limiting the diversity of data science practitioners and resulting in “solutions” that are not developed by the communities they impact.
Speakers in this session and others also highlighted the importance of coalition building to provide cross-institutional resources to students. I was particularly interested to hear about the success of the HBCU Data Science Consortium, as presented by Jason Black, and the Atlanta University Center Data Science Initiative, as presented by director Talitha Washington. (The HBCU Data Science Consortium was catalyzed by the NSF South Big Data Hub and the AUC Data Science Initiative was co-founded by AUC Presidents George T. French Jr. (Clark Atlanta University), David Thomas (Morehouse College), Valerie Montgomery Rice (Morehouse School of Medicine), and Mary Schmidt Campbell (Spelman College) through a major investment from the UnitedHealth Group)). Both initiatives are less than 5 years old but have already arranged an impressive array of programming, including grants, workshops, REUs, and other programs targeted at K-12 students.
“Communication skills” and “practical experience” are universally declared to be important, and yet debate continues as to how to teach them and communicate that institutions are teaching them
Multiple sessions across the leadership summit raised the issue of what constitutes a “data science education,” particularly the workforce development session, but the topic arose even on day 1 in the discussion of organizational structures and again in ramps and pathways. Perhaps unsurprisingly, the development of new degree programs raises the questions of accreditation and competency frameworks. Are all of these emerging programs teaching the same core skills, and do potential employers recognize the experiences of these degree candidates? On the academic side, Rachel Saidi described how her community college program’s graduates often have significant work experience in a “domain” discipline, in addition to their data science master’s education, but that this was not necessarily apparent to industry employers. It appears that, across institutions, administrators are aiming to convince potential employers of the skills their programs provide by having students participate in capstone projects, typically done in partnership with community organizations or local governments. There was great interest among some attendees in expanding the capstone projects to work with industry employers, but it appears (to me) that there is yet work to be done on clarifying and aligning incentives to make potential partnerships successful.
The workforce development session was also interesting in that the panel organizers deliberately solicited input from several industry and government partners, who spoke at length about the skills they expect of potential employees. Both industry representatives spoke about the cross-functional teams (and the need for communication skills that bind together technical and business knowledge), which catalyzed many followup discussions across the following sessions about how to teach communication skills. I agree with my industry colleagues on the importance of communication, but I raised the possibility that university educations are already teaching communication and many other foundational skills, and that maybe the issue is demonstration of those skills via line items on a resume.
This discussion broadly speaks to a common problem in emerging degree programs: it can be difficult to convey the skill sets that new graduates possess.
Tenure still matters in academic data science programs, even as non-tenured research and teaching staff contribute to their success
One of the notable aspects of the ADSA Leadership Summit was the representation of non-tenure track research and teaching staff among attendees. I learned quite a bit in side conversations and lightning talks about the sheer amount of labor that goes into data science programs and institutes that is performed by non-tenured staff (again, perhaps unsurprising). The senior staff who attended highlighted the challenges and the responsibilities they felt to support the career trajectories of those they either supervised or mentored.
Unfortunately, due to the nature of their non-tenure-track positions vis-a-vis the session topics, these university staff were somewhat excluded from roles as panelists at the summit. Some attendees confessed that they could not speak to the programs at their own institutions, given their non-tenure track role, despite the lack of any tenured faculty in attendance to represent their university. I do want to acknowledge that the presence of non-tenured academic staff at a meeting like this is unusual for academia, and I commend the ADSA leadership team for extending the invitation, but it was somewhat unfortunate that their participation was so limited given the depth of their knowledge. Again, this is broadly reflective of academia at large, but it does raise the question of what constitutes “leadership” in an academic data science setting and whether we are fully utilizing the skills and knowledge of all of our colleagues.
There were many who weren’t in the room, who might be having different conversations
As the meeting organizers openly acknowledged, the leadership summit was perhaps not as representative of data science programs across the country as it might have been, despite organizers’ efforts. The meeting attendees skewed towards tenure-track academic leaders at R1 and R2 institutions. As the organizers exhorted us to remember: there is a lot being done in academic data science that was not represented in this meeting, and the leaders of those efforts are likely addressing challenges not discussed here.
Despite the diversity of institutions and positions that were represented in this meeting, I noticed an unusual number of stable groups and perhaps less intermingling than I expected. As with most meetings and conferences, I treated this summit as an opportunity to “get good ideas, for cheap” (i.e., the price of a conversation), and I was not disappointed. I’m now more familiar with the challenges faced by different groups within and adjacent to academic data science, even beyond that conferred by research projects I previously undertook on career paths in academic data science. I left the summit very impressed with the work that has been done, and hopefully with some new collaborators in paving the way for the next generation of graduate students and postdocs to choose from a broader range of data science careers.
NOTE: The opinions and observations reflected in this blog post are that of the author, and do not reflect official positions of any organization.
July 18, 2023: The statements about the HBCU Data Science Consortium and the AUC Data Science Initiative have been rewritten by the author to clarify the founders of the respective initiatives.