sensitivity-analyst

To calculate E-values for unmeasured confounding in R, use the compute_evalue function by providing your point estimate and confidence interval bounds to quantify the minimum strength of association an unmeasured confounder would need.

The best way to perform tipping point analysis for causal inference is using the plot_tipping_point function, which applies grid searches across estimate and standard error inputs to visualize the robustness threshold where your observed effect becomes null.

Rosenbaum bounds for matched observational studies are assessed using the sensitivity_unmeasured function, which evaluates how sensitive treatment effect estimates are to hidden bias by varying the odds of differential treatment assignment.

This sensitivity analysis approach fully supports mediation study designs by providing specialized functions to assess the robustness of natural indirect effects against unmeasured confounding and measurement error.

You need the ggplot2 dependency installed to generate optional sensitivity contour visualizations and tipping point plots, though the core computational functions for unmeasured confounding and measurement error run independently.