Medium
Mediation Analysis
Free
A student says: “To prove mediation in SPSS, I only need to show that X predicts Y, and then when I add M into the regression, X becomes non-significant. That’s mediation.” Which response is the best correction for an advanced statistics course (no SEM)?
A
They’re correct. If X becomes non-significant after adding M, mediation is proven, and you don’t need any additional tests.
B
They’re partially correct, but the key evidence of mediation is testing the indirect effect (e.g., bootstrapped CI for a*b). X does not have to become non-significant for mediation to exist.
C
They’re wrong because mediation can only be tested with ANOVA, not regression.
D
They’re correct, but only if the Sobel test is used instead of bootstrapping.