Dear list: I just conducted a mediational analysis
(X-->M-->Y) and discovered that the standardized partial regression
coefficient for the path between the Mediator and the Outcome variable is .77,
whereas the path from the predictor to the mediator is -.45 and the original
beta coefficient between the predictor and the outcome is -.33 (which reduces
to zero when both the predictor and mediator are controlled). My
questions is whether the path from the mediator to the outcome is too large for
a proper test for mediation. Thoughts appreciated. thanks
Martin F. Sherman, Ph.D.
Professor of
Psychology
Director of
Masters Education: Thesis Track
Loyola College of Arts and
Sciences
Locked
