Epistemologies in Practice: A Review of the Uses of Big Data in the Political and Social Sciences

Discussions concerning Big Data epistemology have so far largely revolved around the question of whether a novel epistemological framework is required in order to understand the use of Big Data in research (e.g. Kitchin 2014; Frické 2015; Floridi 2012; Hey et al. 2009). For instance, Kitchin (2014) argues that Big Data, understood as a complex collection of technologies, methods, and practices of inquiry, enables a form of d ata -driven science, which challenges traditional conceptions of the relationship between theoretical and empirical research. On the other hand, Frické (2015) claims that reformulating epistemology in terms of data-driven science easily leads us to underrate the role of theory in research. As of yet, no consensual understanding has emerged of how research employing Big Data relates to existing epistemologies. This article proposes that a fruitful approach to studying Big Data epistemology is to examine how data-driven science is conducted and tho ught about i n practice. We argue that our epistemological conception of Big Data—regardless of its exact content—should be sensitive to the way in which Big Data is actually used in different scientific contexts. A focus on practices would both provide conceptual tools for formulating our epistemological perspective, and help avoid implausible extreme positions. Toward this end, we conduct a review of a selection of research articles which have appeared in recently published political and social science Big Data special issues. We selected the articles for our review by searching for journal special issues, symposia, or conference proceedings sections published during 2014-2017, which explicitly mention ‘Big Data’ as their topic. For instance, included in our selection are issues such as the ANNALS of the American Academy of Political and Social Science 659(1) , Journal of Communication 64(2) , Psychological Methods 21(4), and th e Political Science and Politics 48(1) symposium on Big Data. Basing on the reviewed articles, we develop a conceptual scheme which classifies them according to their field of study, epistemological perspective, conception of Big Data, and methods of analysis and data collection. The developed scheme shows us how the epistemologies and conceptions of Big Data adopted in different fields of research vary in relation to each other, and thus helps us assess the applicability of different epistemological frameworks. Our most important finding is that the epistemic views at play vary both across fields and within them. Thus, we should be cautious when developing an epistemological framework with the aim of applying it in the study of different scientific contexts. The heterogeneity of Big Data research practices suggest that epistemological discussions should be sensitive to the specific scientific context in question.