Demonstration and evaluation of a method for assessing mediated moderation

Area of Science:

• Psychology
• Statistics
• Social Sciences

Background:

• Mediated moderation describes a complex interaction where variable relationships influence a mediator, which subsequently impacts a dependent variable.
• Understanding this model is crucial for analyzing intricate causal pathways in various research fields.

Purpose of the Study:

• To describe and statistically evaluate the mediated moderation model.
• To assess the accuracy of estimation methods for mediated moderation.
• To identify conditions that may compromise the power to detect mediated moderation effects.

Main Methods:

• Statistical simulation using an adaptation of product-of-coefficients methods.
• Evaluation of relative bias in point estimates and standard errors.
• Examination of power to detect mediated moderation under varying conditions.

Main Results:

• Relative bias in point estimates and standard errors remained within acceptable levels (+/- 10%) across most simulated conditions.
• Bias was systematically influenced by parameter size, sample size, and the presence of direct effects.
• Statistical power to detect mediated moderation was significantly reduced when interaction components were correlated and partial mediated moderation was present.

Conclusions:

• The adapted product-of-coefficients method provides a viable approach for estimating mediated moderation.
• Researchers must be cautious about statistical power when dealing with correlated predictors in mediated moderation models.
• The findings have implications for the accurate estimation of mediated moderation in both experimental and nonexperimental research designs.