grf (version 1.2.0)

average_treatment_effect: Estimate average treatment effects using a causal forest

Description

Gets estimates of one of the following.

• The (conditional) average treatment effect (target.sample = all): sum_i = 1^n E[Y(1) - Y(0) | X = Xi] / n

• The (conditional) average treatment effect on the treated (target.sample = treated): sum_Wi = 1 E[Y(1) - Y(0) | X = Xi] / |i : Wi = 1|

• The (conditional) average treatment effect on the controls (target.sample = control): sum_Wi = 0 E[Y(1) - Y(0) | X = Xi] / |i : Wi = 0|

• The overlap-weighted (conditional) average treatment effect sum_i = 1^n e(Xi) (1 - e(Xi)) E[Y(1) - Y(0) | X = Xi] / sum_i = 1^n e(Xi) (1 - e(Xi)), where e(x) = P[Wi = 1 | Xi = x].

This last estimand is recommended by Li, Morgan, and Zaslavsky (JASA, 2017) in case of poor overlap (i.e., when the propensities e(x) may be very close to 0 or 1), as it doesn't involve dividing by estimated propensities.

Usage

average_treatment_effect(
  forest,
  target.sample = c("all", "treated", "control", "overlap"),
  method = c("AIPW", "TMLE"),
  subset = NULL
)

Arguments

forest

The trained forest.

target.sample

Which sample to aggregate treatment effects over.

method

Method used for doubly robust inference. Can be either augmented inverse-propensity weighting (AIPW), or targeted maximum likelihood estimation (TMLE).

subset

Specifies subset of the training examples over which we estimate the ATE. WARNING: For valid statistical performance, the subset should be defined only using features Xi, not using the treatment Wi or the outcome Yi.

Value

An estimate of the average treatment effect, along with standard error.

Details

If clusters are specified, then each unit gets equal weight by default. For example, if there are 10 clusters with 1 unit each and per-cluster ATE = 1, and there are 10 clusters with 19 units each and per-cluster ATE = 0, then the overall ATE is 0.05 (additional sample.weights allow for custom weighting). If equalize.cluster.weights = TRUE each cluster gets equal weight and the overall ATE is 0.5.

Examples

# NOT RUN {
# Train a causal forest.
n <- 50
p <- 10
X <- matrix(rnorm(n * p), n, p)
W <- rbinom(n, 1, 0.5)
Y <- pmax(X[, 1], 0) * W + X[, 2] + pmin(X[, 3], 0) + rnorm(n)
c.forest <- causal_forest(X, Y, W)

# Predict using the forest.
X.test <- matrix(0, 101, p)
X.test[, 1] <- seq(-2, 2, length.out = 101)
c.pred <- predict(c.forest, X.test)
# Estimate the conditional average treatment effect on the full sample (CATE).
average_treatment_effect(c.forest, target.sample = "all")

# Estimate the conditional average treatment effect on the treated sample (CATT).
# We don't expect much difference between the CATE and the CATT in this example,
# since treatment assignment was randomized.
average_treatment_effect(c.forest, target.sample = "treated")

# Estimate the conditional average treatment effect on samples with positive X[,1].
average_treatment_effect(c.forest, target.sample = "all", subset = X[, 1] > 0)
# }
# NOT RUN {
# }