PSW (version 1.1-3)

psw.aug: Propensity score weighting with augmented estimation

Description

psw.aug is the function to estimate the augmented estimator for mean difference (mean outcome difference for "gaussian" family and risk difference for "binomial" family). The augmented estimator is consistent for the estimand defined by the corresponding propensity score model.

Usage

psw.aug(data, form.ps, weight, form.outcome, family = "gaussian", K = 4)

Arguments

data

data frame to be used.

form.ps

propensity score model.

weight

weighting method to be used. Available methods are "ATE", "ATT", "ATC", "MW", "OVERLAP", and "TRAPEZOIDAL".

form.outcome

outcome model.

family

outcome family, either "gaussian" or "binomial". family="gaussian" by default.

K

value of \(K\) in \(\omega(e_i) = min(1, K min(e_i, 1-e_i)) \) for "TRAPEZOIDAL" weight. The estimand is closer to the average treatment effect (ATE) with larger value of K. K=4 by default.

Value

A list of weighting method, fitted propensity score model, estimated propenstity scores, estimated propensity score weights, augmented estimator and associated standard error.

weight

weighting method.

ps.model

object returned by fitting the propensity score model using glm with "binomial" family.

ps.hat

estimated propensity score.

W

estimated propensity score weight.

est.aug

augmented estimator for mean difference when family = "gaussian".

std.aug

standard error for est.aug.

est.risk.aug

augmented estimator for risk difference when family = "binomial".

std.risk.aug

standard error for est.risk.aug.

Details

psw.aug is used to estimate the augmented estimator, \(\hat{\Delta}_{aug}\), and make inference using the sandwich variance that adjusts for the sampling variability in the estimated propensity score.

Examples

# NOT RUN {
# Load the test data set
data(test_data);
# Propensity score model
form.ps <- "Z ~ X1 + X2 + X3 + X4";
# Outcome model
form.out <- "Y ~ X1 + X2 + X3 + X4";
tmp <- psw.aug( data = test_data, form.ps = form.ps, weight = "ATE",
form.outcome = form.out, family="gaussian" );

# }