This function implements inverse probability weighted (IPW) estimation of population means with missing data, provided fitted propensity scores.
mn.ipw(y, tr, fp)An \(n\) x \(1\) vector of outcomes with missing data.
An \(n\) x \(1\) vector of non-missing indicators (=1 if y is observed or 0 if y is missing).
An \(n\) x \(1\) vector of fitted propensity scores.
The direct IPW estimate of 1.
The ratio IPW estimate.
The ratio IPW estimate is the direct IPW estimate divided by that with y replaced by a vector of 1s. The latter is referred to as
the direct IPW estimate of 1.
Tan, Z. (2020a) Regularized calibrated estimation of propensity scores with model misspecification and high-dimensional data, Biometrika, 107, 137<U+2013>158.
Tan, Z. (2020b) Model-assisted inference for treatment effects using regularized calibrated estimation with high-dimensional data, Annals of Statistics, 48, 811<U+2013>837.