New recommendations for testing indirect effects in mediational models: The need to report and test component paths

Area of Science:

• Psychology
• Statistics
• Research Methodology

Background:

• Concerns regarding replicability and false-positive findings in psychology are prevalent.
• Current mediation analysis methods may contribute to these issues.

Purpose of the Study:

• To compare Type I errors and statistical power between the component and index approaches for mediation assessment.
• To highlight potential risks associated with the commonly used index approach.

Main Methods:

• Simulation studies were conducted to evaluate two distinct mediation assessment approaches: the component approach and the index approach.
• The simulations examined Type I error rates and statistical power under various conditions.

Main Results:

• The index approach, particularly common tests, demonstrated inflated Type I error rates compared to the component approach's joint-significance test.
• Reporting a single mediational index without examining individual causal paths is problematic.

Conclusions:

• Testing individual components of an indirect effect is recommended to establish its presence before calculating its size.
• These findings extend to within-participant and moderated mediation, with a new R-package offered for implementation.