Design approaches to experimental mediation.

Area of Science:

• Experimental social psychology
• Causal inference methods

Background:

• Mediation analysis is central to experimental social psychology, with editors encouraging innovative methods.
• Current "measurement-of-mediation" designs often rely on correlational data for mediator-dependent variable relationships, risking confounding.
• Experimentally manipulating the mediator offers a solution but presents its own challenges.

Purpose of the Study:

• To provide a practical guide to manipulation-of-mediator designs.
• To encourage more rigorous mediation approaches in experimental social psychology.
• To strengthen the ability to infer causality of mediating variables on dependent variables.

Main Methods:

• Describes experimental manipulations targeting the mediator's causal effect.
• Details experimental designs: double randomization, concurrent double randomization, and parallel designs.
• Provides published examples and discusses strengths and challenges of each design.

Main Results:

• Manipulation-of-mediator designs enhance causal inference for mediation.
• Different designs offer varying strengths in establishing mediator causality.
• Practical guidance is offered for implementing these rigorous designs.

Conclusions:

• Manipulation-of-mediator designs are crucial for robust causal claims in mediation research.
• Researchers should adopt these methods to overcome limitations of traditional mediation designs.
• This approach strengthens the understanding of causal mechanisms in social psychology.