Robust variance-moderated testing of linear model parameters

[nhejazi]

The methodology implemented in this software package performs variance-moderated hypothesis testing of estimators of causal parameters (currently, only the ATE) in the nonparametric statistical model based on the values of the efficient influence function evaluated at each observation. Within the context of parametric models (e.g., the \beta coefficient of a linear model), it is possible to perform the same decomposition (since \hat{\beta} is an asymptotically linear estimator of \beta) and thus perform the same robust variance-moderated hypothesis test.

[nhejazi]

The purpose of this package is to implement the supervised variance moderation procedure to stabilize small-sample behavior of nonparametric or semiparametric-efficient estimators. It is outside of the scope of this R package to implement analogous methods based on linear models.