Understanding and Preventing the Reproduction of Gender and Racial Inequalities in the Big Data Era
This project investigates one of the social consequences of the Big Data revolution: the reproduction of social inequality. In particular, the research team will examine how data-driven predictive practices drive people to make unfair decisions, exacerbating existing gender and racial inequalities.
There are three core agendas – theoretical, empirical, and practical – of this project. Theoretically, the team will explain how data-based decision making becomes a critical mechanism for reproducing gender and racial inequality, via an interdisciplinary approach. Computer scientists have developed mathematical models to characterize when algorithmic predictions reflect the bias embedded in data, and social scientists have developed sociological theories of how social inequality is reproduced. Leveraging these two complementary frameworks, the project will develop a unified theory for explaining how data-driven decisions reinforce social inequality.
Empirically, the research team will demonstrate whether – if so, how – the existing hierarchy is reproduced through algorithmic predictions. Two topics of interest include the gender wage gap in the labor market and the racial gap in the criminal justice system. Employers and judges make decisions in these two social domains: One group about who to hire and how much wage to offer based on the predicted productivity, and the other group about bailing for a defendant based on the predicted recidivism probability. The two decision-making processes, however, share a commonality. Increasingly, employers and judges refer to data-driven predictions when making decisions. Since data-driven predictions reflect the bias in the data, both processes are expected to contribute to reproducing existing hierarchies and inequalities.
Practically, based on the critical reflection of the data collection and utilization processes, the team will design a new protocol, which, if adopted, can help to prevent the exacerbation of gender and racial inequalities. The study team will suggest ways to use big data and algorithmic predictions that do not reproduce existing social hierarchies and inequalities.
Kangwook Lee, assistant professor of electrical and computer engineering
Eunsil Oh, assistant professor of sociology