Exploring the Use of Text as Data in Political Science

In political science, as in other social sciences, the rise of text mining “has vastly broadened the scope of questions that can be investigated empirically” (Benoit, p. 461). As a political science librarian supporting a department with an emphasis on computational methods, I have encountered a steady rise in questions from researchers looking to use a variety of textual data sources. These range from whether library books can be digitized for computational analysis, to whether our existing databases allow for automated searching and bulk downloading, to requests for new licensed resources for text mining. Additionally, recent changes to the social media data landscape, such as the monetization of the Twitter API, are shaking up researchers’ approaches for finding and accessing data, causing them to turn to the library for guidance.

To better understand the specific needs and challenges of political science researchers using text mining methods, I conducted an analysis of recent dissertations from top political science programs. Specifically, I looked at what types of sources are used for text mining, how these sources are accessed, what challenges researchers encounter in working with large-scale textual data, and how libraries are involved in supporting this work. In this presentation I will highlight findings from this study, discuss how they have informed the way I and others at my institution are supporting text mining in the social sciences, and consider implications for libraries as we navigate the future of text as data.

Works Cited.

Benoit, K. (2020). Text as data: An overview. In L. Curini & R. Franzese (Eds.), The SAGE Handbook of Research Methods in Political Science and International Relations (pp. 461-497). SAGE Publications Ltd.