Faculty at SP2 are using big data in a variety of ways to advance research. From utilizing data to examine and inform social policy to advancing the ethical use of big data, SP2 faculty have positioned themselves as experts in the field of data analysis for social policy.
Dennis P. Culhane, PhD
Dr. Culhane researches the development, use, and innovation of integrated data systems (IDS) for policy analysis and program reform as well as the ethical use of administrative data in social science research. He leads the Actionable Intelligence for Social Policy (AISP) initiative as well as participates in the Administrative Data Research Facilities (ADRF) Network.
Ezekiel Dixon-Román, PhD
Dr. Dixon-Román is currently working on a book project on racializing algorithms. In this project, he is making both theoretical and empirical interventions on the ways in which and extent to which machine learning algorithms become imbued with sociopolitical values of hierarchizing difference. He draws connections between cybernectics, governmentality, biopolitics, and social policy. He is particularly interested in how these sociotechnical processes materialize in algorithmic governance, particularly in the areas of predictive policing, child welfare, and learning analytics. Finally, via what is learned from this investigation alongside algorithmic experimentations, he seeks to identify possibilities to mitigate reproduced or regenerated formations of difference.
Dr. Dixon-Román is also collaborating with Scott Dexter (of Brooklyn College) on a project using topic modeling, a machine learning method for text analysis, to analyze the ways in which race has been discursively constituted and reconfigured in policy-related text over several decades. In particular, they are taking all of the articles ever published in the top policy journal such as the Journal of Policy Analysis and Management, the flagship journal for the Association for Public Policy Analysis and Management, and employing latent dirichlet allocation to topic model these texts with a particular interest in if and how discourses on race appear as a latent topic and how it historically shifts over time.
As a more long term line of inquiry, we know little about the psychometric influences that underlie cybernetic systems of computational measurement and reasoning and the ways in which this may be contributing to the reconfiguring of education and society. Cybernetics is a science of the prediction and control of future events and actions, and has been influenced profoundly by psychology, psychiatry, psychoanalysis, and the cognitive sciences (Halpern, 2014; Hayles, 2001; Parisi, 2014). Yet, less is well known about the influence of psychometrics in cybernetics despite its obvious overlapping interest and endeavors in modeling human cognitive and behavioral processes. Scholarly links by way of Department of Defense research as well as computer adaptive testing go back several decades and include more recent developments in methods of assessment via video games, stealth assessment, and computational psychometrics. The latter are recent instantiations of the scholarly relationships between psychometrics and cybernetics and the sociotechnical movement toward what he is calling the new psychometrics. The new psychometrics raises important philosophical questions about measurement and society that have a long history in academic discussions. He is particularly interested in (1) studying the influences between psychometrics and cybernetics; (2) exploring how the developments in cybernetics and computational methods have led toward both the ubiquity of new forms of psychometrics in society while also transforming the technologies and practices of psychometrics in education; (3) investigating what this genealogy might illuminate about the potential sociopolitical relations imbued in the sociotechnical assemblages of the new psychometrics and how might they be reconfiguring power relations in society in real time; and (4) what human and/or algorithmic forms of intervention and practice might help mitigate the potential reconfiguring of education and society.
Amy Castro Baker, PhD, MSW
Women experience financial markets differently from their male peers, but wealth building interventions rarely account for the invisible ways gender shapes market outcomes. Women work less than men and make less over the lifespan, but that information is typically not included in the calculators used by defined benefit plans whose default settings match male lifespans and earning curves. Women typically work only 75% of the years men work because they are more likely to be caring for children or ill family members. This means the timing of their investment strategies and the mix of their asset class should look different from that of their male peers. Dr. Castro Baker is currently working to develop models for grantmakers and philanthropists that account for and reverse algorithmic bias.
Ioana E. Marinescu, PhD
Dr. Marinescu uses data to understand social and economic trends. A product market is concentrated when a few firms dominate the market. Similarly, a labor market is concentrated when a few firms dominate hiring in the market. Using big data from the leading employment website CareerBuilder, she calculates labor market concentration for over 8,000 geographic-occupational labor markets in the US. Based on the DOJ-FTC horizontal merger guidelines, the average market is highly concentrated. Using a panel IV regression, Dr. Marinescu’s group of economists shows that going from the 25th percentile to the 75th percentile in concentration is associated with a 17% decline in posted wages, suggesting that concentration increases labor market power. More information on her research is available here.