Research Areas
Combining individual-level data sources to estimate contact network parameters:
Most high-level network features are difficult or impossible to measure from egocentric reporting. I develop methods to assess whether multiple data sources, when combined, provide enough information to estimate properties such as degree assortativity or clustering in both static and dynamic networks.
Measuring and accounting for cross-community interaction in CRT:
Methodological development to account for influences of the underlying contact network on the outcome of a large, community-randomized trial of HIV prevention. Includes methods for estimation of causal effects of treatment in the presence of network interference and methods for use of viral genetic data as a proxy measure for cross-community interaction.
Assessing sensitivity of causal inferences to unmeasured confounding:
Developing interpretable tools to assess robustness to confounding for a variety of analytic methods. This includes development of software to make implementation of sensitivity analysis accessible for researchers across disciplines.