Marginal Mediation

Provides the ability to perform "Marginal Mediation"--mediation wherein the indirect and direct effects are in terms of the average marginal effects (Bartus, 2005, <>). The style of the average marginal effects stems from Thomas Leeper's work on the "margins" package. This framework allows the use of categorical mediators and outcomes with little change in interpretation from the continuous mediators/outcomes. See for more details on the method (peer reviewed articles forthcoming).

CRAN Rdoc Downloads BuildStatus

The MarginalMediation package provides the ability to perform marginal mediation analysis. It provides a useful framework from which to interpret the coefficients in a mediation analysis, especially when the mediator(s) and/or outcome is binary or a count (other types of outcomes will be added).

You can install it via:




The main function is mma():

    ind_effects = c("apath-bpath"))

where ... consists of 2 or more model objects. The first is the b and c' path model, while the others are the a path models.

The ind_effects is a vector of requested mediated paths. These estimates are in terms of the average marginal effects using the a x b method of estimating indirect paths. Any number of these can be included, although it is limited to the number of variables available in the models.

A Quick Example

Below is an example, where the theoretical backing of such a model is not very stable, but it is useful to show how to use the function and the output.

## Data for the example
#> furniture 1.7.6: learn more at
## The MarginalMediation package
#> MarginalMediation 0.5.1: This is beta software.
#> Please report any bugs ([email protected]).
pathbc = glm(marijuana ~ home_meals + gender + age + asthma, 
             data = nhanes_2010, 
             family = "binomial")
patha = glm(home_meals ~ gender + age + asthma,
            data = nhanes_2010, 
            family = "gaussian")
mma(pathbc, patha,
    ind_effects = c("genderFemale-home_meals",
    boot = 500)
#> calculating a paths... b and c paths... Done.
#> ┌───────────────────────────────┐
#> │  Marginal Mediation Analysis  │
#> └───────────────────────────────┘
#> A marginal mediation model with:
#>    1 mediators
#>    3 indirect effects
#>    3 direct effects
#>    500 bootstrapped samples
#>    95% confidence interval
#>    n = 1417 
#> Formulas:
#>    ◌ marijuana ~ home_meals + gender + age + asthma
#>    ◌ home_meals ~ gender + age + asthma 
#> Unstandardized Effects
#> ⎺⎺⎺⎺⎺⎺⎺⎺⎺⎺⎺⎺⎺⎺⎺⎺⎺⎺⎺⎺⎺⎺
#> ── Indirect Effects ──
#>                           A-path   B-path Indirect    Lower   Upper
#> genderFemale-home_meals -1.34831 -0.00973  0.01312  0.00429 0.02562
#> age-home_meals          -0.05689 -0.00973  0.00055  0.00003 0.00139
#> asthmaNo-home_meals     -0.00428 -0.00973  0.00004 -0.00639 0.00672
#> ── Direct Effects ──
#>                Direct    Lower   Upper
#> genderFemale  0.10430  0.04813 0.15967
#> age           0.00066 -0.00603 0.00848
#> asthmaNo     -0.00172 -0.06947 0.07061
#> -----

The print method provides:

  1. the a path,
  2. the b path,
  3. the indirect effect with the confidence interval, and
  4. the direct effect with the confidence interval.

These are all average marginal effects, and are, therefore, in terms of the corresponding endogenous variable’s units.


This is currently beta but I am excited to provide an initial working release. Let me know if you find any bugs or want to discuss the method ([email protected]).

Final Notes

More will be done to MarginalMediation as it is under development, including:

  1. More bootstrapping options (e.g., Bias-Corrected Bootstrap)
  2. More distributional options as currently only gaussian, binomial, and Poisson are available (e.g., zero-inflated distributions, multinomial distributions)



Version 0.5.1 (Initial CRAN Release)

This is the initial release. The main function is mma(), which performs the marginal mediation analysis. Other functions are mainly helper functions in interpreting and pulling information out of a mma object.

This is beta software, so please report any issues!

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.


0.5.1 by Tyson S Barrett, 10 months ago

Browse source code at

Authors: Tyson S Barrett [aut, cre] , Angelo Canty [ctb] (S original bootstrap method , , Brian D. Ripley [ctb] (R port , bootstrap method) , Thomas Leeper [ctb] (Vectorized marginal effects ,

Documentation:   PDF Manual  

GPL-2 license

Imports stats, magrittr, boot, purrr

Suggests testthat, knitr, rmarkdown, furniture, margins

See at CRAN