Comparative data science education policy and management: United States and Europe

• In Event: Case studies in educational policies: adult learning, data science, EdTech and competencies-based reform

Proposal

Scholars have examined the recent evolution of the “knowledge-based economy” to the “educational intelligent economy,” an application of knowledge using technology (Jules & Salajan, 2019). The emergence extends to undergraduate education. For example, interest in data science in undergraduate education is evidenced by the growth of courses such as the University of California, Berkeley’s lower-division Foundations of Data Science course, which increased from 100 students in 2013 to over 4,200 students in the 2023-2024 academic year (Kafka, 2018; Watanabe, 2023). Similar growth can be expected internationally, particularly in more developed, knowledge-based economies striving to improve their universities, as the U.S. higher education model is often emulated (Bok, 2013). As testament of its growing importance, the bipartisan-supported Data Science and Literacy Act introduced in 2023 would provide first-of-its-kind U.S. federal data science education legislation.

For this study, I am investigating the pressures driving curriculum development for degrees in an area that is seeing job market growth and evaluating these results considering comparative higher education perspectives. Specifically, I am critically evaluating what “good” undergraduate data science education and administration are, including a focus on ethical practices, and looking at big normative questions surrounding data science. For example, how should “good” be evaluated? What criteria shall we use: good in the sense that “good” data science is good for democratic society or good in the sense of students getting employment in their field of study?

To address these questions in this pilot study, I am surveying and interviewing data science educators and higher education administrators at four-year U.S. and European higher education institutions. I am recruiting participants through professional association listservs and social media, direct emails, in-person solicitations, and personal networking connections. For U.S. institutions, I add institutional data from the National Center for Education Statistics’ Integrated Postsecondary Education Data System and am seeking comparable data for European institutions. My survey and interview guide cover questions about curricula, including mathematics/statistics, computer science, domain, and ethics; administrative structure, fostering of transdisciplinarity, and integration of transfer students; training formats; and basic demographics. The survey questions were influenced by consideration of academic capitalism and isomorphism theories. I am also using these theories to set up hypotheses on social and market forces in higher education shaping undergraduate data science education.

Regarding survey data, I use a Kruskal-Wallis omnibus test for differences among groups; for subsequent tests where the omnibus test is significant, I use logistic regression. I predict that there will continue to be an emphasis on the need for a heavy dose of “foundational” courses in mathematics/statistics and computer science coming from educators and administrators with departmental homes in mathematics/statistics and computer science. For educators and administrators with departmental homes outside of mathematics/statistics and computer science, I predict that there will be a de-emphasis on mathematics requirements and a call for more applied, i.e., statistical, training along with significant coursework in domains outside of mathematics/statistics and computer science and different data science tracks. Interviews enable me to gain more detailed insight on data science education and administration.

Based on survey and interview analyses, I am making recommendations to improve data science curricula, program offerings, and organizational structure. For example, instead of requiring a two-year sequence of math courses, programs could offer a single course or a short-course sequence in the mathematical foundations of data science that covers multiple key concepts without going into as much detail. Programs may want to provide different tracks for data science that go into depth on certain subjects. Having too many prerequisites for higher-level data science application courses, including experiential learning opportunities, could be problematic and exclusionary. Integrating ethics into all data science curricula along with having a stand-alone course in ethics may be preferable.

Through comparing U.S. and European higher education institutions, topics to which I contribute include knowledge management across higher education institutions in the educational intelligent economy era; how data science education should work in practice; how disciplines can coexist with each other in this space; and the designing and implementation of public policies at the federal and local levels. I also add comparative evidence for social and market forces influencing curricular and administrative decision-making into academic capitalism and isomorphism theoretical literature.

Author

• Torbet McNeil, University of Arizona