Medium
Mediation Analysis
Free
A researcher tests whether study time (X, hours/week) affects exam score (Y) through test anxiety (M). They run three regressions (all variables are continuous):
- M on X: b = -0.60, p < .001
- Y on X: b = 4.00, p < .001
- Y on X and M: b(X) = 1.50, p = .040; b(M) = -4.20, p < .001 Which conclusion best matches these results (using the classic regression-based mediation logic)?
A
No mediation: because X is still significant in Model 3, M cannot be a mediator.
B
Full mediation: because b(M) is significant and X becomes smaller, mediation is complete.
C
Partial mediation: X predicts M, M predicts Y controlling X, and the effect of X on Y shrinks but remains significant.
D
Suppression (not mediation): because the sign of b(X) changed from Model 2 to Model 3.