Two-level moderated mediation models with single-level data and new measures of effect sizes

Area of Science:

• Psychometrics
• Statistical Modeling
• Quantitative Psychology

Background:

• Traditional moderated mediation (moME) models can be limited in accuracy.
• There is a need for improved methods to quantify moderated mediation effect sizes.
• Bayesian approaches offer robust estimation techniques.

Purpose of the Study:

• To propose a novel two-level moderated mediation (2moME) model for single-level data.
• To develop and validate measures for quantifying moderated mediation effect sizes (ES).
• To compare the performance of the 2moME model against the conventional moME model.

Main Methods:

• Development of a two-level moderated mediation (2moME) model.
• Application of a Bayesian approach for parameter estimation and hypothesis testing.
• Monte Carlo simulations to evaluate model accuracy and reliability.
• Development of interpretable effect size measures for moderated mediation.

Main Results:

• The 2moME model provides more accurate parameter estimates than the conventional moME model.
• The 2moME model's credibility intervals more accurately cover moderated mediation effects and effect sizes.
• Statistical tests using the 2moME model are more reliable in controlling Type I errors, particularly with heteroscedasticity.
• Developed effect size measures offer enhanced interpretability regarding moderator influence.

Conclusions:

• The proposed 2moME model represents an advancement in analyzing moderated mediation.
• The developed effect size measures provide clearer insights into the role of moderators.
• The Bayesian approach facilitates robust estimation and testing of moderated mediation effects and their sizes.