R, Julia, and JAGS code to accompany the paper "Bayesian G-Computation to Estimate Impacts of Interventions on Exposure Mixtures: Demonstration with Metals from Coal-fired Power Plants and Birthweight"

NOTE: these programs specifically do not sample from the baseline distribution of covariates (e.g. via bootstrapping, bayesian boostrap, or parametric modeling). This is a large sample approximation (which seems wise in this example), and results may vary at small sample sizes (in simulations this effect is typically negligible).

File manifest:

• Julia programs (v. 1.3.0)

  • Gibbs.jl: Julia functions to complete a Bayesian hierarchical modeling analysis of simulated data using Hierarchical models within a Bayesian g-computation framework using Gibbs sampling based MCMC methods
  • sims_bayes.jl: a Julia program to perform parallel Bayesian analysis of multiple simulated datasets using the functions from Gibbs.jl
  • sims_ml.jl: a Julia program to perform parallel maximum likelhood analysis of multiple simulated datasets using the GLM package in Julia

• R program (v. 3.6.0)

  • sims_bma.R: an R program to perform parallel Bayesian analysis of multiple simulated datasets using Bayesian model averaging within a Bayesian g-computation framework

• JAGS programs (v 4.3.0)

  • No files included, but JAGS programs are contained as character strings in sims_bma.R