Overlap weights (Li, Morgan & Zaslavsky 2018)

Weight = 1 - p(X) for treated units and p(X) for controls: targets the subpopulation with the best covariate overlap and is bounded by construction (no extreme weights).

Usage

morie_weight_ow(data, treatment, covariates)

Arguments

data

Data frame.

treatment

Binary treatment column (0/1).

covariates

Covariate column names.

Value

A morie_weight object (estimand "ATO").

References

Li, Morgan & Zaslavsky (2018) JASA 113(521).

Examples

d <- data.frame(t = rbinom(100, 1, 0.4), x = rnorm(100))
morie_weight_ow(d, "t", "x")
#> morie_weight: overlap weights (Li et al. 2018) (estimand ATO)
#>   n = 100  ESS = 98.8  weight range [0.359, 0.618]