Hi all
I wonder whether anyone can advise me please.
I am testing 234 couples to see whether interpersonal conflict mediates the
association between their personalities and their relationship satisfaction.
I have following for neuroticism, conflict and satisfaction:
A = effect of neuroticism on conflict
B = effect of conflict on satisfaction
C = unmediated effect of neuroticism on satisfaction
C' = mediated effect of neuroticism on satisfaction
A = .26 (p < .001)
B = -.16 (p < .001)
Indirect effect (A*B) = -.04 (Sobel: p < .001)
C' = .08 (p < .05)
Total effect = C' + A*B = .04
C = -.084 (not significant)
My questions are:
1. A*B has an opposite sign to C'. Is this an example of suppression even
though C is not significant?
2. If C' were not significant (as is the case with some traits), is it still
valid to use it to create the total effect? I ask because significance in
mediation testing as it is rarely referred to.
Many thanks
Max
|