plot.ATE: Plots of Empirical CDF and Weighted Empirical CDF for...

In asadharis/ATE: Inference for Average Treatment Effects using Covariate Balancing

Description

Plot function for class 'ATE'.

Usage

## S3 method for class 'ATE'
plot(x, ...)

Arguments

x	An object of type 'ATE'.
...	Further arguments passed to or from the function.

Details

This function plots the empirical CDF and weighted empirical CDF for each covariate to demonstrate the effect of covariate balancing and for graphical diagnostics. In observational studies with confounding, the covariate distributions are different for each treatment arms. Comparisons of unweighted empirical CDF would demonstrate this difference. The balancing weights constructed by balancing moments of covariate distributions, and the weighted CDF would show an improved balance.

Author(s)

Asad Haris, Gary Chan

See Also

ATE

Examples

library(ATE)
# Binary treatment.
set.seed(25)
n <- 200
Z <- matrix(rnorm(4 * n), ncol = 4, nrow = n)
prop <- 1 / (1 + exp(Z[, 1] - 0.5 * Z[, 2] + 0.25 * Z[, 3] + 0.1 * Z[, 4]))
treat <- rbinom(n, 1, prop)
Y <- 200 + 10 * treat + (1.5 * treat - 0.5) *
     (27.4 * Z[, 1] + 13.7 * Z[, 2] + 13.7 * Z[, 3] + 13.7 * Z[, 4]) + rnorm(n)
X <- cbind(exp(Z[, 1]) / 2, Z[, 2] / (1 + exp(Z[, 1])),
           (Z[, 1] * Z[, 3] / 25 + 0.6) ^ 3, (Z[, 2] + Z[, 4] + 20) ^ 2)

# Estimation of average treatment effects (ATE).
fit1 <- ATE(Y, treat, X)
summary(fit1)
## NOT RUN
# plot(fit1)

# Estimation of average treatment effects on treated (ATT).
fit2 <- ATE(Y, treat, X, ATT = TRUE)
summary(fit2)
## NOT RUN
# plot(fit2)


# Four treatment groups.
set.seed(25)
n <- 200
Z <- matrix(rnorm(4 * n), ncol = 4, nrow = n)
prop1 <- 1 / (1 + exp(1 + Z[, 1] - 0.5 * Z[, 2] + 0.25 * Z[, 3] + 0.1 * Z[, 4]))
prop2 <- 1 / (1 + exp(Z[, 1] - 0.5 * Z[, 2] + 0.25 * Z[, 3] + 0.1 * Z[, 4]))

U <- runif(n)
treat <- numeric(n)
treat[U > (1 - prop2)] = 2
treat[U < (1 - prop2) & U > (prop2 - prop1)] = 1

Y <- 210 + 10 * treat + (27.4 * Z[, 1] + 13.7 * Z[, 2] + 13.7 * Z[, 3] +
                        13.7 * Z[, 4]) + rnorm(n)
X <- cbind(exp(Z[, 1]) / 2, Z[, 2] / (1 + exp(Z[, 1])),
           (Z[, 1] * Z[, 3] / 25 + 0.6) ^ 3, (Z[, 2] + Z[, 4] + 20) ^ 2)

fit3 <- ATE(Y, treat, X)
summary(fit3)
## NOT RUN
# plot(fit3)

asadharis/ATE documentation built on Nov. 14, 2020, 2:27 a.m.