About

I am a researcher at the Australian Centre for Evaluation in the Australian Treasury and the RMIT Department of Economics, Finance, and Marketing. Across both roles I work with large, linked, administrative data on impact evaluations. I recently received my PhD from the Australian National University where my thesis focused on the use of a causal machine learning method called the causal forest in program evaluation.

The causal forest is a causal machine learning method for finding drivers of treatment effect heterogeneity — how a given treatment affects different people differently. It is useful because standard approaches rely on pre-specifying hypotheses and then testing them, while the causal forest allows for a statistically valid, data-driven, flexible way to explore heterogeneity. The causal forest, if used well, lets evaluators search for heterogeneity across an arbitrarily large number of variables, including across interactions between them. This helps to make sure unintended consequences of a program will not be missed.

Prior to my current roles I was as a data scientist / quantitative researcher working for government and not-for-profits organisations.

My recent paper in the International Statistical Review called How Do Applied Researchers Use the Causal Forest? A Methodological Review includes a web appendix detailing the 133 papers identified.