CHDH Seminar Series 2020: Empirical Dynamic Modeling Approaches to Causal Inference — Michael Zyphur (Part 1)

Presenter: Michael Zyphur

Date: Monday 9 March

Time: 12 - 1 PM

Location: Lowe Theatre (Room 103 Redmond Barry Building, University of Melbourne)

Seminar Description: Experiments and model-based regressions have been the two dominant methodological workhorses for over 60 years in the social and health sciences. However, these approaches have limitations that make them unsuitable for modeling complex dynamic systems. These systems involve interactions and feedback loops that can produce seemingly unpredictable behavior that is hard to study experimentally and arguably even harder to model using typical regression methods. As an alternative, physical sciences such as ecology have recently witnessed the advent of revolutionary new methods designed to study such complex dynamic systems. Referred to as empirical dynamic modeling or EDM, these methods allow 1) characterizing the complexity of a system and the degree of nonlinearity that defines its evolution over time, while also enabling 2) distinguishing correlation from causation using a non-parametric method for causal inference (i.e., a truly data-driven method for causal discovery). My talk introduces EDM and applies it to study the relationship between daily temperate and crime in Chicago, showing a causal effect of temperature on crime but not the reverse -- there are juicier results about companies and individual emotions that I can also describe after the initial talk. By using new methods such as EDM, social and health scientists will be better equipped to study a variety of real-world complex dynamic systems.

Presenter Bio: Michael Zyphur graduate in 2006 with a PhD in industrial and organizational psychology from Tulane University (New Orleans). He is currently an associate professor of management and marketing in the faculty of business and economics at the University of Melbourne. His research focuses on the application of quantitative methods to answer questions in the social and health sciences. His new EDM Stata package is now available for free download at https://jinjingli.github.io/edm/.

See Michael's Google Scholar page here: https://scholar.google.com/citations?user=J2VA-RMAAAAJ&hl=en