estimation: Cross-fitting estimation of the GATEs

In farbmacher/LATEtest: Local Instrument Validity Tests with Causal Forests

Description

Grow pruned tree to identify heterogeneous subgroups and picks promising leaves (based on training data) Calculate magnitude of potential violations within promising leaves (in estimation sample)

Usage

estimation(tra, est, y, covars, type, minsize, cp)

Arguments

tra	training sample
est	estimation sample
y	relevant interval of Y
covars	provides names of covariables
type	indicates whether we consider D=1 or D=0
minsize	pruned tree insists on at least 2*"minsize" observations in the additional trees to identify the promising subgroups
cp	sets complexity parameter which rpart uses to fit the tree before pruning; default=0

Value

augmented inverse probability weighting scores, indicator for promising leaves

farbmacher/LATEtest documentation built on Nov. 20, 2020, 10:13 a.m.