Dr. Gregori’s research focuses on improving the reliability of marketing insights through advanced experimental design. His latest work introduces the "Proximal Difference-in-Differences" estimator, a semiparametric approach designed to correct for biases caused by unobserved individual characteristics—factors that often lead to inaccurate results in standard marketing studies.
Applying this method to a large-scale longitudinal study of "Buy-Now-Pay-Later" programs, the research demonstrates that traditional estimation methods can overestimate treatment effects by as much as 28%. By leveraging negative control variables to account for hidden factors like disposable income, Dr. Gregori provides a more robust framework for evaluating consumer behavior. To support the academic and practitioner community, he has released the "ProxDiD" R package for implementing these advanced causal inference techniques.
