Research
I study social measurement. Most of my work is concerned with the history and practice of methodology (i.e. rules and rulemaking about knowledge) in the domain of attaching numbers to what people do, say, and think.
My current applied methodological work focuses on the interpretation of low-rank approximations in quantitative and computational cultural analysis. A working paper on multivariate word association measurement with word embeddings is available here.
My research in historical sociology examines public displays of expert knowledge and the formal organization of technical authority. Most of my work employs archival data, which I analyze using a range of methods (statistical, historical, computational, interpretive). I am currently studying the development of research methodology during US academic expansion in the postwar period (1945-1970s). My first paper from this line of work, written as part of a research group at MPIWG on the history of validation in regulatory and biomedical science, investigates the origins of psychometric validity theory (i.e. internal, external, construct) and its entanglements with the cultivation of academic reputation in postwar US psychology.
My dissertation project, Observing Prestige, examined the measurement of occupational and professional prestige in sociology. I developed two extended case studies — estimating the causal effect of LaTeX typesetting on the evaluation of professional knowledge in economics; describing the co-portrayal of occupational incumbents on the US television game show Jeopardy! — to discuss the conceptual and mathematical limitations of standard prestige measures. Working papers associated with these projects will be available in the future.
In prior work, I’ve developed a historical sociology of educational expansion (with Mitchell Stevens) and assessed the predictability of life outcomes in longitudinal social survey data (with Matt Salganik, Ian Lundberg, Sara McLanahan, and more than 100 collaborators). I have also led and contributed to collaborative projects on team-based coding software for forum data, variable selection in longitudinal data systems, and reporting guidelines for multi-analyst studies.