This R package implements the Generalized Oaxaca Blinder framework described in Guo and Basse (2020). It provides utilities for computing point estimates and confidence intervals using common regression functions, and for constructing custom regression adjustment.
Install the package from Github with:
# install.packages("devtools")
devtools::install_github("gwb/RGOB")The main entry point is the gob function, which computes the point estimate
and associated confidence interval, using the specified regression function.
The code below gives a simple example of how to adjust using a both linear
and non-linear models:
library(RGOB)
set.seed(123) # for replicability
## Generates synthetic data
N <- 100 # the number of units
X <- rnorm(N); B <- rnorm(N); C <- rnorm(N)
Z <- sample(c(0,1), N, replace=T) # the binary assignment vector
Y <- 1 + 0.2 * X + 0.5 * Z + rnorm(N, sd=0.1) # the observed outcomes
Yb <- ifelse(Y > 1.2, 1, 0) # a binary observed outcome
Yp <- rpois(N, lambda=Y) # integer observed outcomes
## Computes the point estimate and associated confidence interval
## using the separate slope adjustment (same as in Lin 2013)
gob(lm(Y ~ X + B), Z)
## Logistic regression adjustment
gob(glm(Yb ~ X + B, family=binomial(link="logit")), Z)
## Poisson regression adjustment
gob(glm(Yp ~ X + B, family=poisson), Z)- add check to
gobfunction to make sure there is no issue with assignment in formula