Regression-based statistical mediation and moderation analysis in clinical research: Observations, recommendations,...

Area of Science:

• Clinical Psychology
• Quantitative Psychology
• Statistical Methods

Background:

• Mediation and moderation analyses are crucial for understanding psychological mechanisms.
• Traditional linear regression approaches for these analyses are increasingly outdated.
• Clinical researchers need updated methods for hypothesis testing regarding effects.

Purpose of the Study:

• To address and update the practice of mediation and moderation analysis in clinical research.
• To offer recommendations and debunk myths surrounding traditional methods.
• To introduce modern techniques, including conditional process analysis.

Main Methods:

• Review of mediation and moderation analysis in Behaviour Research and Therapy.
• Application of linear regression for hypothesis testing.
• Demonstration of conditional process analysis using the PROCESS macro for SPSS and SAS.

Main Results:

• Identified limitations in historically significant but older analytical approaches.
• Provided updated recommendations and debunked common misconceptions.
• Illustrated the integration of mediation, moderation, and conditional process analysis.

Conclusions:

• Clinical researchers should move beyond traditional mediation/moderation methods.
• Modern statistical approaches offer more nuanced understanding of effects.
• Conditional process analysis represents a significant advancement in analyzing mechanisms and contingencies.