Big Data, Outliers, and Group Marginalization

Michigan State University, East Lansing MI

Investigators

Abstract

This award supports a doctoral dissertation research project. The primary aim of the project is to contribute to understanding how diversity is navigated, compromised, and negotiated within big data practices. It will bring to light values and biases that privilege certain groups while marginalizing others, meaning those values that are unconsciously integrated into the design of technologies, such as data collection and sorting practices. Such results will have substantial social significance. They will serve to provide a better understanding of the various ways in which data can be different for different people, and of the potential biases that can result in inequity and injustice for the most vulnerable. The outcomes of this project also have the potential to address the social aspects of representational compression at the interface of humans and technology; they can be applied to integrating intersectional perspectives in technological design and technologically-driven policy decisions. The researcher will engage in an ethnographic study of how data may enact particular formulations and subjectivities of categorization. She grounds her theoretical approach in Science and Technology Studies to identify instances where dominant discourses of data come into confrontation with alternative narratives. Of particular interest are instances where the data produced is unusual; the data is sometimes referred to as an "outlier" or, in more derogatory terms, "dirty data" that needs "scrubbing." This project aims to analyze such "unusual" data through ethnographic investigation of self-tracking cultures. This approach will contribute to knowledge of how big data practices encourage and impede certain social positions and subjectivities. It will also contribute to eliciting the underlying values of big data practices; such values speak to how data is variously deployed for use in civilian projects, economics, governance, and knowledge production.