Home

Welcome to the User Manual for hetviz, an R package designed to facilitate exploratory, hypothesis-generating analyses of treatment effect heterogeneity (TEH).

---

We mean (average) treatment effect in the causal sense, as a comparison of what an outcome would have potentially been under treatment & what the outcome would have potentially been under no treatment.

• Note that this software assumes a binary treatment.

In a population, there may be subpopulations for whom their (average) treatment effect is systematically different. We refer to this systematic variability as TEH. It would be of great use to be able to empirically characterize these subpopulations in the following ways:

1. estimation of the true number of underlying subgroups
2. identification of which covariates and what values determine subpopulation membership
3. estimation of the subpopulation average treatment effect

This software allows analysts to interact with their data in ways that will allow TEH to reveal itself, via

• forest plots to check for general patterns that would suggest heterogeneity
• subgroup profiles to visually compare subgroups to identify distinguishing features.
• viz(ualization) by subgroup to visually evaluate the covariate distributions of a specific subgroup.
• covariate profiles to visually compare how covariate distributions change as a function of the subgroup-specific treatment effect.
• viz(ualization) by covariate to visually evaluate how the distribution of a specific covariate changes as a function of the subgroup-specific treatment effect.

---

Install + Deploy

1. Install and load the devtools package in R, which will facilitate the installation of this package.
2. Install hetviz.
3. Load hetviz.
4. Run hetviz.

# step 1
install.packages("devtools")
# step 2
devtools::install_github("sanoke/hetviz")
# step 3
library(hetviz)
# step 4
hetviz()