Date/Time
Thursday
14 Nov 2024
4:00 pm - 5:30 pm

Location
470 Stephens Hall

Event Type
Colloquium

Shreeharsh Kelkar
Continuing Lecturer, University of California, Berkeley

The critical social science research on the emergence of “data scientists” is composed of many puzzles and somewhat contradictory findings, especially when it comes to their relationship with so-called “domain experts.” Some scholars have argued that data scientists display an imperialistic desire to create a domain-agnostic science while others have shown that they seek only to create tools that will serve existing science and domain experts. This paper argues that these contradictions stem from a methodological issue: researchers often start their investigations with experts who already identify as data scientists rather than looking at the emergence of data-driven analysis in a field of inquiry.

Through a historical and ethnographic analysis of an emerging field called “educational data mining,” I show that researchers within this field who described themselves as “data scientists” tended to work more inductively rather than deductively and, at the margin, distrusted established theory in a field; in other words, the line between “data scientist” and “domain expert” in the education data mining field was porous. At the same time, some researchers’ self-presentation as data scientists was more about strategy than methodology and reflected reactions to gatekeeping in the field. Based on this, I argue that the term “domain expert,” as it’s used in the literature, is somewhat incoherent, especially when used to refer to experts as diverse as researchers and frontline workers. The conflict, if there is one, is not between data scientists and domain experts but between data scientists and existing “meta-experts” in a domain.