Treatment effect estimators and marginal-effects extenders

Two function families live here:

Details

Treatment-effect estimators (legacy)

• estimate_ate() – IPW-weighted OLS ATE.

• estimate_plr() – Partially Linear Regression, native cross-fit (cross-validated against DoubleML).

• estimate_pliv() – Partially Linear IV (LATE), native (cross-validated against DoubleML).

• estimate_ate_gcomputation() – G-computation ATE with stdReg's sandwich SE, native.

• sensitivity_rosenbaum() – Rosenbaum bounds, native normal-approximation Wilcoxon signed-rank bounds.

• e_value() – VanderWeele-Ding E-value. Thin wrapper native VanderWeele-Ding continuous-scale E-value (module 26).

Marginal effects (native; conventions of emmeans and Vincent Arel-Bundock's marginaleffects, which are used only by the cross-validation tests):

• morie_effects_emmeans() – estimated marginal means.

• morie_effects_predictions() – adjusted predictions.

• morie_effects_comparisons() – counterfactual contrasts.

• morie_effects_slopes() – partial derivatives.

• morie_effects_tidy() – native coefficient table with broom::tidy()'s columns.

They run natively for lm / glm fits (analytic delta-method standard errors) and return plain data frames.

References

Chernozhukov et al. (2018); Robins (1986); VanderWeele & Ding (2017); Rosenbaum (2002); Arel-Bundock (2024, marginaleffects); Lenth (2024, emmeans); Robinson, Hayes & Couch (2024, broom).