Sensitivity Analysis

Sensitivity analysis is a procedure to evaluate how the conclusions of causal inference estimation would change if one or more of the causal assumptions (see Assumptions) were violated in specific ways. Sensitivity analysis is available for some assumptions of some causal inference methods. Check the Methods page to see which options are available for each method.

The core function across submodules of causalinf to run sensitivity analysis is sensitivity_analysis(<args>). For instance, if one is using Selection on Observables (SoO), sensitivity analysis can be conducted as follows:

1
2
3
from causalinf import soo

sens = soo.sensitivity_analysis(<args>)

<args> vary across submodules depending on the method used.