Correcting Misclassified Mediation Analysis

Use three methods to estimate parameters from a mediation analysis with a binary misclassified mediator. These methods correct for the problem of "label switching" using Youden's J criteria. A detailed description of the analysis methods is available in Webb and Wells (2024), "Effect estimation in the presence of a misclassified binary mediator" .

COMMA

COMMA: COrrecting Misclassified Mediation Analysis

Overview

COMMA provides a set of functions for the analysis of mediation models with binary mediator misclassification.

The two main parts are:

• Classification probability calculations
• Parameter estimation

Classification probability calculations

The package allows users to compute the probability of the latent true mediators and the conditional probability of observing a mediator value given the latent true mediator, based on parameters estimated from the COMMA_EM, COMMA_PVW, and COMMA_OLS functions.

Parameter estimation

Jointly estimate parameters from the true mediator, observed mediator, and outcome mechanisms, respectively, in a mediation analysis with a binary misclassified mediator variable. Three methods are provided for parameter estimation: an ordinary least squares correction procedure, a predictive value weighting approach, and an expectation-maximization algorithm.

Installation

# Install from CRAN:
install.packages("COMBO")

# Install the development version from GitHub:
# install.packages("devtools")
devtools::install_github("kimberlywebb/COMMA")