Choose a method

Start with the study question, exposure contrast, outcome and adjustment set. Method availability depends on that combination. The original eight estimator families are retained here, with examples aligned to the current public wrappers.

Method	Main use	Read next
GLM	Adjusted outcome regression	GLM
G-Computation	Average modeled outcomes under exposure contrasts	G-Computation
AIPW	Outcome and propensity models with optional cross-fitting	AIPW
IPW	Propensity-based weighting	IPW
TMLE	Targeted outcome-model updating	TMLE
PSM	Propensity-score matching	PSM
Causal Forest	Exploratory effect heterogeneity	Causal Forest
GPS	Dose contrasts for a continuous exposure	GPS

Multi-method comparison is a comparison utility, not a ninth estimator or a vote on causality. Compare only compatible targets and scales.

Synthetic example data

Download methods-data.R, save it in your working directory, and run:

library(IOBRcausal)
source("methods-data.R")
str(data)

The script creates 240 fictional subjects and defines data, including the columns used on the method pages. Immunotherapy is a simulated binary exposure; TMB is a simulated continuous exposure. The response, continuous outcome and survival times are also artificial. Use the examples to learn the interface, not to make clinical claims or calibrate a method.

Read the returned contract

Check Scale, contrast, uncertainty, method status, support and diagnostics. Binary, continuous and survival refer to outcome types; the exposure has its own type. Continuous-exposure policies and survival horizons can require the core iobr_causal_estimate() interface rather than a convenience wrapper.

Use the support reference for implementation and validation limits. For complete analysis delivery through one request, return to the workflow chapters.