RMediation: an R package for mediation analysis confidence intervals.

Area of Science:

• Statistics
• Psychometrics
• Data Analysis

Background:

• Mediation analysis is crucial for understanding indirect effects in statistical models.
• Existing methods for constructing confidence intervals (CIs) for mediated effects have limitations.
• The PRODCLIN program, while cited, is cumbersome and has case limitations.

Purpose of the Study:

• To introduce the RMediation package for calculating confidence intervals for mediated effects.
• To offer a more robust and user-friendly alternative to existing software.
• To implement and compare various methods for CI estimation in mediation analysis.

Main Methods:

• The RMediation package utilizes the distribution-of-the-product method, Monte Carlo simulations, and asymptotic normal distribution.
• It generates percentiles, quantiles, and plots for visualizing the distribution and CI of the mediated effect.
• The package is built using the widely available R software environment.

Main Results:

• The distribution-of-the-product method demonstrates superior statistical performance for CI construction.
• RMediation provides accurate results, addressing limitations encountered with PRODCLIN.
• The package offers enhanced capabilities beyond those available in PRODCLIN.

Conclusions:

• RMediation is a powerful and accessible tool for researchers conducting mediation analysis.
• It improves the accuracy and ease of calculating confidence intervals for mediated effects.
• The package facilitates a deeper understanding of indirect effects in statistical modeling.