CTSI BERD Webinar: 'Identification and Semiparametric Estimation of Conditional Means from Aggregate Data'

Often, researchers want to estimate the relationship between two variables, but only have aggregate data on each variable separately. For example, in analyzing voter preferences by race, researchers may only have county-level election results and county-level demographics. In this webinar, Cory McCartan will discuss how common approaches to this problem rely on overly strong assumptions, and will present a more flexible, machine learning–based alternative.

Using simulations and real-world examples, McCartan will show how this method improves accuracy and allows researchers to consider the effect of key assumptions being violated. The new methods are available in open-source software, which will be shared so attendees can apply the approach in their own work.

Register here.

The Biostatistics, Epidemiology and Research Design (BERD) Recent Topics in Research Methods seminar series is held during the academic year. Lectures on fundamental research methods are given in the fall semester, with more advanced topics presented in the spring semester. Statisticians and methodologists from multiple Penn State departments present these seminars.