README.md

In ehsanx/iptw: Inverse Probability of Treatment Weighting

iptw

R package for calculating Inverse Probability of Treatment Weights

• This package calculates various measures of association and helps understand and visualize the link between causal models such as Marginal Structural Models and Standerdized measures.

Installation

library(devtools)
install_github("ehsanx/iptw")

Loading the package

require(iptw)

Pulling the help file

help(package="iptw")

Using this package

r1 <- .9 # Risk of the treated subjects
r0 <- .5 # Risk of the untreated subjects
measures.calc(r1,r0)
# Calculates risk measures

form.table.object <- form.table(Y1A1L1=150, 
                                Y1A0L1=45, 
                                Y1A1L0=20, 
                                Y1A0L0=5, 
                                Y0A1L1=300, 
                                Y0A0L1=10, 
                                Y0A1L0=40, 
                                Y0A0L0=55)
form.table.object
# Forms the crude, stratified tables suitable 
# for using in other functions of this package.

ungrouped.data.object <- ungrouped.data(form.table.object)
# Creates ungrouped data suitable for performing regression on it
head(ungrouped.data.object)

associational(form.table.object) 
# Calculates measures of associations
standardization(form.table.object) 
# Calculates Standardized measures
iptw(form.table.object, type = "sw") 
# Standardized measures are calculated by this function 
# using inverse probability of treatment weights (IPTW).

iptw.regression(form.table.object, type = "sw") 
# Calculates IPTW estimates from regression
# requires grid and survey package

graph(form.table.object, type = "sw")
# Maps probabilities and frequesncies

Author

• Ehsan Karim :octocat: Feel free to report any errors / update suggestions.

Useful Reference

• [x] Hernán MA, Robins JM (2016). Causal Inference. Boca Raton: Chapman & Hall/CRC, forthcoming.

Related web-Apps

• [x] Causal Inference Web-Apps

ehsanx/iptw documentation built on May 16, 2019, 1:21 a.m.