Mediation

Mediation asks how or why an effect happens. X influences M, and M in turn influences Y, so part of the X-to-Y relationship travels through the mediator. A mediator is a mechanism sitting between cause and outcome — distinct from a moderator, which changes the strength of an effect rather than carrying it.

Current practice tests the indirect effect (the product of the a and b paths) directly using bootstrapped confidence intervals, rather than the older Baron and Kenny causal-steps procedure. If the bias-corrected bootstrap interval for the indirect effect excludes zero, mediation is supported — whether or not the total effect was significant.

Many methodologists now discourage the full-versus-partial mediation distinction; report the indirect, direct and total effects and let the reader judge. Remember that mediation is a causal claim, so cross-sectional data collected at a single point in time supports it only weakly. Acknowledge that limitation explicitly.