Heterogeneous treatment effects with `grf`
This post summarizes an email exchange between the two of us on how to use the grf package to assess treatment effect heterogeneity that we…
This post summarizes an email exchange between the two of us on how to use the grf package to assess treatment effect heterogeneity that we thought could be useful more broadly.
banditsCI: Inference tools for adaptive experiments
This R statistical software package provides functions for conducting frequentist inference on adaptively generated data. These functions produce point estimates and confidence intervals, using the…
This R statistical software package provides functions for conducting frequentist inference on adaptively generated data. These functions produce point estimates and confidence intervals, using the methods proposed in Zhan, Ruohan, et al. (2021) and Hadad, Vitor, et al. (2021). The code in this package is directly adapted from the original python code for those publications.
Hierarchical ridge regression tutorial
Suppose we have a factorial experiment, where we want to account for two-way and higher-order interactions. We may think that interaction effects will be small…
Suppose we have a factorial experiment, where we want to account for two-way and higher-order interactions. We may think that interaction effects will be small but not exactly equal to zero, and higher-order interactions will tend to be associated with smaller effects relative to lower order interactions. Accounting for all interactions in a standard linear model may be costly in terms of variance, so we want to use some form of regularization. Hierarchical ridge regression facilitates penalization that is increasing with degree of complexity of interactions.
Adaptive experimentation tutorial
Much of my recent work has been on developing tools for designing and analyzing data from adaptive experiments. This tutorial provides an overview of adaptive…
Much of my recent work has been on developing tools for designing and analyzing data from adaptive experiments. This tutorial provides an overview of adaptive experimental design, descibes some basic algorithms for treatment assignment, and discusses considerations for inference.