Skip to contents

Automated Mediation Analysis

Source: R/because_mediation.R
because_mediation.Rd

Decomposes the total effect of an exposure on an outcome into direct and indirect pathways by tracing paths in the fitted DAG and multiplying posterior coefficients.

Usage

because_mediation(fit, exposure, outcome, prob = 0.95)

Arguments

fit

A fitted object from because().

exposure

Character string; the name of the exposure variable.

outcome

Character string; the name of the outcome variable.

prob

Numeric; probability mass for the credible interval (default 0.95).

Value

A list containing:

  • paths: A data frame summarizing each path (Path string, Mean, SD, CI).

  • summary: A data frame summarizing Total, Direct, and Total Indirect effects.

  • samples: A matrix of posterior samples for each path and the total effect.

Details

This function reconstructs the causal graph from the parameter_map stored in the fit object. It uses igraph to find all simple paths from exposure to outcome. For each path, it identifies the corresponding regression coefficients (beta_...) and computes their product across all MCMC samples.

Note: This assumes linear relationships for indirect effects (\(\beta_1 \times \beta_2\)). For non-linear models, this is an approximation of the average causal effect.

Examples

if (FALSE) { # \dontrun{
# Assuming a dataset 'df' exists with variables X, M, and Y
equations <- list(M ~ X, Y ~ M + X)
fit <- because(equations, data = df)
med_results <- because_mediation(fit, exposure = "X", outcome = "Y")
print(med_results$summary)
} # }