I am interested in developing and evaluating methods to assess the effectiveness and safety of health care interventions, particularly considering treatment effect modifiers and heterogeneity. My primary area of expertise is devising methods to jointly analyze data from multiple sources, and related topics such as handling missing data, causal inference, and probability sampling.

Perhaps the most traditional purpose for multi-source statistical analysis is to use data from a collection of similar randomized controlled trials in what is known as meta-analysis. Meta-analysis has historically been used to facilitate more precise estimation of treatment effects, but has more recently been tapped to explore the variability of treatment effects across individuals/studies/treatments with different characteristics. Although meta-analysis is not a new statistical idea, there are still a plethora of open questions, including methods for synthesizing categorical data with rare events, best practices for handling missing data and relating relevant trial-estimated treatment effects to populations or individuals of interest.