Skip to content

Data-Wise/mediationverse

BranchesTags

Repository files navigation

mediationverse

Unified ecosystem for mediation analysis in R

Lifecycle: experimental Repo Status R-CMD-check Website Status

Overview

The mediationverse is a collection of R packages for mediation analysis, providing a unified ecosystem for:

  • 🔧 Model fitting and extraction
  • 📊 Effect size computation
  • 📈 Confidence interval estimation
  • 🛡️ Sensitivity analysis
  • 🔬 Simulation studies

Quick Start

# Install mediationverse
pak::pak("data-wise/mediationverse")

# Load foundation package + packages you need
library(mediationverse)  # Loads medfit (foundation)
library(probmed)         # For P_med effect size
library(RMediation)      # For confidence intervals
library(medrobust)       # For sensitivity analysis

# Fit mediation models
fit_m <- lm(M ~ X + C, data = mydata)
fit_y <- lm(Y ~ X + M + C, data = mydata)
med_data <- extract_mediation(fit_m, model_y = fit_y,
                               treatment = "X", mediator = "M")

# Get P_med effect size, confidence intervals, and sensitivity analysis
pmed_result <- compute_pmed(med_data)
ci_result <- ci(med_data, type = "dop")
sensitivity <- sensitivity_analysis(med_data)

Ecosystem Status Dashboard

📊 View Full Dashboard | 🗺️ Development Roadmap

Package Overview

Package Status Build Website Role
medfit Lifecycle Build Docs Foundation
probmed Lifecycle Build Docs Effect Size
RMediation Lifecycle Build Docs Confidence Intervals
medrobust Lifecycle — Docs Sensitivity Analysis
medsim Lifecycle Build Docs Simulation Infrastructure

Development Progress

Current Quarter (Q4 2025): Ecosystem Development Phase

Milestone Status Progress
medfit Phase 3 (Extraction) ✅ Complete 100%
medfit Phase 4 (Model Fitting) 🚧 In Progress 60%
probmed Integration ⏳ Pending 0%
RMediation Integration ⏳ Pending 0%
medsim Core Implementation ✅ Complete 100%
Comprehensive Documentation ✅ Complete 100%

Next Quarter (Q1 2026): Integration & Release

  • Complete medfit Phase 4-5 (Bootstrap, Testing)
  • Integrate probmed and RMediation with medfit
  • CRAN submissions for medfit, probmed
  • Finalize mediationverse meta-package

Package Ecosystem

The mediationverse follows a modular architecture with selective loading inspired by tidyverse and easystats:

Architecture Diagram

┌──────────────────────────────────────────────────────────────────────┐
│                        mediationverse                                │
│                      (Meta-Package Loader)                           │
│                                                                      │
│  library(mediationverse)  →  Loads only medfit (foundation)         │
│  library(probmed)         →  Explicit loading for effect sizes       │
│  library(RMediation)      →  Explicit loading for CIs                │
│  library(medrobust)       →  Explicit loading for sensitivity        │
│  library(medsim)          →  Explicit loading for simulation         │
└──────────────────────────────────────────────────────────────────────┘
                                    │
                                    │ Always loads
                                    ▼
┌──────────────────────────────────────────────────────────────────────┐
│                            FOUNDATION LAYER                          │
│  ┌────────────────────────────────────────────────────────────┐     │
│  │                         medfit                             │     │
│  │  • S7 classes (MediationData, SerialMediationData)         │     │
│  │  • extract_mediation() - Extract from lm/glm/lavaan        │     │
│  │  • fit_mediation() - Formula interface                     │     │
│  │  • bootstrap_mediation() - Bootstrap inference             │     │
│  └────────────────────────────────────────────────────────────┘     │
└──────────────────────────────────────────────────────────────────────┘
                                    │
                    ┌───────────────┼───────────────┐
                    │               │               │
                    ▼               ▼               ▼
┌──────────────────────────────────────────────────────────────────────┐
│                      SPECIALIZED PACKAGES LAYER                      │
├────────────────┬─────────────────┬────────────────┬─────────────────┤
│   probmed      │  RMediation     │  medrobust     │    medsim       │
│ ┌────────────┐ │ ┌─────────────┐ │ ┌────────────┐ │ ┌─────────────┐ │
│ │ Effect     │ │ │ Confidence  │ │ │Sensitivity │ │ │ Simulation  │ │
│ │ Sizes      │ │ │ Intervals   │ │ │ Analysis   │ │ │Infrastructure│ │
│ ├────────────┤ │ ├─────────────┤ │ ├────────────┤ │ ├─────────────┤ │
│ │• P_med     │ │ │• DOP        │ │ │• Bounds    │ │ │• Data gen   │ │
│ │• Scale-free│ │ │• Monte Carlo│ │ │• Falsify   │ │ │• Parallel   │ │
│ │• Continuous│ │ │• MBCO       │ │ │• Partial ID│ │ │• Caching    │ │
│ │• Binary    │ │ │• Asymmetric │ │ │• Plots     │ │ │• Tables     │ │
│ └────────────┘ │ └─────────────┘ │ └────────────┘ │ └─────────────┘ │
└────────────────┴─────────────────┴────────────────┴─────────────────┘
                                    │
                                    │ All packages use
                                    ▼
                        ┌─────────────────────┐
                        │  MediationData (S7) │
                        │  Unified Interface  │
                        └─────────────────────┘

Data Flow

User Data
   │
   ▼
fit_mediation() or lm()/glm()/lavaan::sem()
   │
   ▼
extract_mediation()  ──────►  MediationData object
   │                                  │
   │                                  │
   ├──────────────┬──────────────┬────┴────────────┐
   │              │              │                 │
   ▼              ▼              ▼                 ▼
probmed::     RMediation::   medrobust::      medfit::
compute_pmed() medci()        sensitivity()   bootstrap_mediation()
   │              │              │                 │
   │              │              │                 │
   └──────────────┴──────────────┴─────────────────┘
                        │
                        ▼
                Results & Inference

Core Packages

medfit: Foundation Infrastructure

Purpose: Unified S7-based infrastructure for mediation analysis Status: Phase 4 in progress Provides:

  • S7 classes (MediationData, SerialMediationData, BootstrapResult)
  • Model extraction from lm/glm/lavaan
  • Bootstrap inference (parametric, nonparametric, plugin)
  • Type-safe data structures

Links: Website | GitHub | Issues

probmed: Probabilistic Effect Sizes

Purpose: Compute P_med, a scale-free probabilistic effect size Status: Stable, ready for integration Provides:

  • P_med computation for continuous and binary outcomes
  • GLM support (logistic, Poisson)
  • Integration with lavaan and mediation packages
  • Bootstrap confidence intervals

Links: Website | GitHub | Issues

RMediation: Confidence Intervals

Purpose: Asymmetric confidence intervals via Distribution of Product Status: Stable on CRAN Provides:

  • Distribution of Product (DOP) method
  • Monte Carlo (MC) confidence intervals
  • MBCO bootstrap method
  • Integration with standard R models

Links: CRAN | Website | GitHub

medrobust: Sensitivity Analysis

Purpose: Partial identification under differential misclassification Status: In development Provides:

  • Partial identification bounds for NDE/NIE
  • Data-driven falsification tests
  • Synthetic data generation
  • Publication-ready visualizations

Links: Website | GitHub | Issues

medsim: Simulation Infrastructure

Purpose: Standardized infrastructure for Monte Carlo studies Status: Core implementation complete Provides:

  • Environment-aware execution (local/HPC)
  • Parallel processing with progress bars
  • Ground truth caching
  • Publication-ready figures and LaTeX tables

Links: Website | GitHub | Issues

Installation

From GitHub (Development)

# Install all packages
pak::pak("Data-Wise/mediationverse")

# Or install individually
pak::pak("Data-Wise/medfit")
pak::pak("Data-Wise/probmed")
pak::pak("Data-Wise/medrobust")
pak::pak("Data-Wise/medsim")

# RMediation from CRAN
install.packages("RMediation")

From CRAN (Future)

# Planned for Q2-Q3 2026
install.packages("mediationverse")

Usage

Loading Packages

The mediationverse uses selective loading: only the foundation package (medfit) is loaded by default. Load other packages as needed for your analysis.

library(mediationverse)
#> ── Attaching mediationverse 0.0.0.9000 ──
#> ✔ medfit 0.1.0 (foundation package)
#> ℹ Use library(probmed) for P_med effect size
#> ℹ Use library(RMediation) for DOP/MBCO inference
#> ℹ Use library(medrobust) for sensitivity analysis
#> ℹ Use library(medsim) for simulation utilities
#> ──────────────────────────────────────────────────────

# Load additional packages as needed
library(probmed)      # For probabilistic effect sizes
library(RMediation)   # For confidence intervals
library(medrobust)    # For sensitivity analysis
library(medsim)       # For simulation studies

Why selective loading?

  • Clean namespace (avoids function conflicts)
  • Only load what you need for your analysis
  • medfit is always available (foundation for all packages)

Package Management

# List installed packages and versions
mediationverse_packages()

# Update all packages
mediationverse_update()

# Check for function conflicts
mediationverse_conflicts()

Complete Analysis Workflow

# Load foundation and packages you need
library(mediationverse)  # Loads medfit (foundation)
library(probmed)         # For P_med effect size
library(RMediation)      # For confidence intervals
library(medrobust)       # For sensitivity analysis

# 1. Fit mediation models (medfit - already loaded)
fit_m <- lm(M ~ X + C, data = mydata)
fit_y <- lm(Y ~ X + M + C, data = mydata)

# 2. Extract mediation structure (medfit)
med_data <- extract_mediation(fit_m, model_y = fit_y,
                               treatment = "X", mediator = "M")

# 3. Bootstrap inference (medfit)
boot_result <- bootstrap_mediation(med_data, n_boot = 2000)

# 4. Compute probabilistic effect size (probmed)
pmed_result <- compute_pmed(med_data)

# 5. Get confidence intervals (RMediation)
ci_result <- ci(med_data, type = "dop")

# 6. Sensitivity analysis (medrobust)
robust_result <- sensitivity_analysis(med_data)

# 7. Run simulation study (medsim - load if needed)
library(medsim)
sim_results <- medsim_run(
  method = my_method,
  scenarios = medsim_scenarios_mediation(),
  config = medsim_config("local")
)

Design Philosophy

Following principles from tidyverse and easystats:

  1. Foundation Package: medfit provides shared S7 infrastructure
  2. Specialized Packages: Each package focuses on one methodological contribution
  3. Consistent API: Unified interfaces across all packages
  4. Type Safety: S7 classes ensure data integrity
  5. Low Dependencies: Minimal external dependencies for stability
  6. Comprehensive Testing: >90% code coverage across packages

Documentation

Getting Started

Status & Planning

Package Documentation

Contributing

The mediationverse is in active development. Contributions are welcome!

Development Workflow:

  1. Fork the repository
  2. Create a feature branch from dev
  3. Make your changes with tests
  4. Submit a pull request to dev

See CONTRIBUTING.md for detailed guidelines.

Citation

If you use packages from the mediationverse in your research, please cite the individual packages:

Tofighi, D. (2025). medfit: Infrastructure for mediation analysis in R.
R package version 0.1.0. https://github.com/data-wise/medfit

Tofighi, D. (2025). probmed: Probabilistic effect sizes for mediation analysis.
R package version 0.1.0. https://github.com/data-wise/probmed

Full citations available via citation("packagename").

Related Ecosystems

The mediationverse draws inspiration from:

  • tidyverse - Cohesive data science packages
  • easystats - Statistical modeling ecosystem
  • mlr3 - Machine learning framework

Code of Conduct

Please note that the mediationverse project is released with a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.

License

GPL (>= 3)

Contact

Maintainer: Davood Tofighi (dtofighi@gmail.com) ORCID: 0000-0001-8523-7776

Sources:

About

Meta-package for the mediation analysis ecosystem in R

data-wise.github.io/mediationverse/

Topics

mediation-analysis sensitivity-analysis causal-inference causal-models

Resources

Readme

License

GPL-3.0 license
Activity

Stars

1 star

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •  

Languages