TRIADS Speaker Series: Modeling the Speech Audio in the Social Sciences
Join Christopher Lucas, Assistant Professor of Political Science, for a presentation on Modeling the Speech Audio in the Social Sciences.
Lunch will be served.
Abstract: Qualitative accounts of politics regularly reference tough-to-measure concepts about political speakers: an "inspiring" political candidate, a "passionate" local official, or a Supreme Court Justice who just seemed "a bit skeptical" of an argument. Because these concepts are difficult to observe, they have largely eluded quantitative inquiry. Here, we unify existing qualitative understanding with quantitative measurement and testing, proposing new methods for the statistical analysis of political speech. First, we develop a statistical model of political speech, in which political speech is modeled as a stochastic process shaped by fixed and time-varying covariates, including the history of the conversation itself. We then develop a unified causal framework for studying the effects of text, audio, and visual speech components. In sum, we lay methodological foundations for a broad research agenda into the measurement of tough-to-observe concepts in unstructured data, as well as the causal effects of those concepts on core social science behaviors.