Widening the Shrinking Pipeline: The Case of Data Science

Gender imbalance in STEM (Science, Technology, Engineering and Mathematics) studies and occupations is a well-known phenomenon with a large body of research that tries to explain it and offer remedial interventions. Data science is a new and interdisciplinary STEM-oriented domain, integrating knowledge and skills from computer science, mathematics, and statistics with an application domain, from which the data draw their context and meaning. Data science applications are relevant for various domains, and therefore, a variety of populations are increasingly attracted to learning it. Addressing the theme of the conference, i.e., “Women in Engineering”, in this paper we describe a data science workshop for social sciences and digital humanities researchers. A significant majority (83%) of the participants of this workshop self-identified as women. This gender proportion, the opposite of that prevailing in STEM studies, led us to examine the workshop from a gender perspective. Our results indicate that the women participating in the data science workshop perceived it as an opportunity to acquire research tools rather than programming tools. We suggest that framing the workshop as a research tool workshop and not as a programming workshop reduced prevalent gender barriers in STEM, encouraging a majority of women researchers to participate. In this paper, we elaborate on the participants' perceptions about data science and programming and analyze them based on three theoretical perspectives: expectancy value theory, the interdisciplinary perspective, and the epistemological perspective. Keywords-data science, data science education, gender balance, interdisciplinary of data science.