Data Science Studies + QUAL Workshop – Wednesday, January 23rd 3:30-5:00 pm @ Data Science Studio
Qualitative Methods for Data Science
Following Stuart Geiger’s Data Science Studies talk, Data Science Studies and the QUAL Initiative will be convening a small group for a workshop-style discussion exploring how to incorporate qualitative methods into data science curricula and how to advance mixed-method research collaborations in data science. This event is by invitation, but we have a couple slots still available, so email Anissa Tanweer at firstname.lastname@example.org if you’re interested in the topic.
Workshop on Qualitative Methods for Data Science
Wednesday, January 23rd, 2019 from 2:30-5:00 pm
Data Science Studio, 6th floor of the Physics/Astronomy Tower
The occasion for this workshop is a visit to UW by R. Stuart Geiger, resident ethnographer at the UC Berkeley Institute for Data Science (BIDS). Stuart will present a study recently published with Aaron Halfaker that foregrounds the role of qualitative insights and contextual understanding in data-intensive computational research. The paper is based on their attempt to replicate a lauded study out of the Oxford Internet Institute on Wikipedia bots. That work, called “Even Good Bots Fight,” claimed to find evidence for rampant bot-on-bot conflict in the pages of Wikipedia and received substantial attention in the popular press. Geiger and Halkafer’s paper, however, demonstrates that Tsvetkova et al made a number of erroneous assumptions about what their data represented, leading to results and interpretations that bore little resemblance to what was actually happening on the ground. Geiger and Halfaker, in contrast, conducted a quantitative analysis that included the same digital traces used in the original study, but also incorporated insights from their long-term participant-observation among the Wikipedia community. This mixed methods approach yielded results that were in direct contradiction to the results published in the “Even Good Bots Fight” study, demonstrating instead that bots were reliable and constructive tools of governance on the platform, and rarely in true conflict.
As we witness the ascendancy of quantitative computational methods in the academy, and a simultaneous disinvestment in qualitative research methods training, Geiger and Halfaker’s work starkly reminds us that computational research often misses important insights that can only be gleaned through a qualitative lens. We will use their paper and Stuart’s lecture to launch a constructive discussion about the relationship between qualitative methods and data science. Our goal is to explore the importance of qualitative sensibilities and methods in data-intensive computational research, and to identify a path toward the development of curricula and collaborations that happily marry quantitative and qualitative approaches in data science.