IMPACT Design and Statistics Core member Keith Goldfeld, DrPH, MS, MPA, has created a blog series exploring an analytic method called Targeted Minimum Loss Estimation (TMLE). TMLE is among causal inference approaches—methods first developed for observational data, but increasingly relevant for randomized trials as well. In particular, Dr. Goldfeld is digging into what TMLE can (and can’t) offer for cluster randomized trials, including stepped-wedge designs.

In this series, Dr. Goldfeld explores TMLE’s use in the context of cluster randomized trials generally and stepped-wedge trials more specifically. One of the challenges of this method for regular users is the underlying theory of the method and the math that goes along with it. In this blog series Dr. Goldfeld works through the theory to get a better understanding of it.

Getting to the bottom of TMLE: influence functions and perturbations | Posted on February 5, 2026

Getting to the bottom of TMLE: the (almost) vanishing nuisance interaction | Posted on March 2, 2026

Getting to the bottom of TMLE: forcing the target to behave | Posted on March 9, 2026

Read the ongoing series here.