Sensitivity Analysis for Weighted Estimators

Provides tools to conduct interpretable sensitivity analyses for weighted estimators, introduced in Huang (2024) and Hartman and Huang (2024) . The package allows researchers to generate the set of recommended sensitivity summaries to evaluate the sensitivity in their underlying weighting estimators to omitted moderators or confounders. The tools can be flexibly applied in causal inference settings (i.e., in external and internal validity contexts) or survey contexts.

senseweight

senseweight implements a set of sensitivity functions and tools to help researchers transparently conduct sensitivity analyses for weighted estimators. senseweight allows researchers to assess the sensitivity present in their weighted estimates to omitted confounders. Specific methods provided in senseweight include the following: (1) visualization tools to summarize sensitivity; (2) summary tables containing necessary sensitivity statistics; (3) formal benchmarking methods which allow researchers to use observed covariates to assess the plausibility of different confounders.

Installation

You can install the development version of senseweight from GitHub with:

# install.packages("devtools")
devtools::install_github("melodyyhuang/senseweight")

References

The package implements a series of methods developed in the following papers.

For the technical introduction of the sensitivity tools:

For less technical introductions with interesting applications and best practice:

Huang, Melody and Hartman, Erin. “Assessing Nonignorable Response: Sensitivity Analysis for Survey Weighting, with Applications to Survey Estimates of COVID-19 Vaccination Uptake.” Working paper.
Bailey, Michael. “Polling at a Crossroads.” (Chapter 7)