Reflexivity, Positionality, and Rigor in the Context of Big Data Research

Abstract

In this robust unstructured era of big data, neopositivistic empiricism asserts strict objectivity when manipulating data, yet big data are riddled with the subjective positions of those entering the data, those creating and maintaining the storage and retrieval mechanisms and those sifting through the data. Big data offer unique analytical opportunities to reveal patterns that may have gone unnoticed otherwise. The sheer amount and a variety of data being generated by a seemingly heterogeneous population and then collected primarily by businesses this relatively new phenomena, and it is tempting to assume data do not lie and are truths. The experiences of those who enter the data that then become amalgamated into big data, the experiences or subject positions, often referred to as positionality, of those manipulating and handling the data shape their understandings of the world, their epistemology. Positionality (in terms of race, age, socioeconomic status, ethnicity) of a researcher, scientist, database administrator and other actors influence what questions are and are not asked in data science. Knowledge is mediated and constructed through interactions with the world. The meanings extrapolated, the knowledge built from big data is limited by the questions that are asked. Here I impose a social constructivist critique, or rather a reflectivist theoretical stance, echoing the postulation of Thatcher inferring that the questions we ask, the analysis we conduct, and in turn, the patterns we find in big data are heavily influenced by our epistemologies, the lenses by which we view the world. Epistemologies describe the background, the perspective or lenses from which we study the world. We must consider not only the perspectives of individual researchers but also those of other individuals, such as the computer scientists and database administrators who maintain the data. The so-called black box is made up of a disconnected stream of bureaucracies, not a conspiring individual, but countless groups of individuals limited by their own knowledge, positionality, time and other constraints including those associated with actors up stream. Thus I ask: (how) do identities and experience (dis)appear from big data? This overarching question can help understand research design choices and what meaning is extracted from big data based on the identities and experiences of the researchers who enter the data and the database administrators who maintain them. This can be also applied to the identities and experience of the subjects whose data have been amalgamated into big data. As critical theorists, moving past positivistic assumptions, we understand that even the choice of mathematical methods by which researchers query, analyze, and display data are influenced by their positionality. While big data may seemingly be quantitative, the data are also overwhelmingly qualitative in nature, necessitating methodology distinctive to qualitative research.