1

I'm running a SEM model with one mediator variable (cognitive ability). Using lavaan I report within the code the indirect and total effects to test them. However, I have some doubts on how to interpret them. Specifically, I have 2 cases in which the indirect effect is not significant, but the direct and total effects are. Is it correct to claim, in this case, that no mediation effect exists? On the contrary, I have two other cases where the indirect and direct effects are significant, but not the total effect and specifically the indirect has an opposite sign of the direct effect. In this case, how should I interpret them?

I attach the table with the decomposition. The first case refers to neuroticism an agreeableness, while the second to extroversion and openness.

enter image description here

Thanks for your help

Community
  • 1
Luca Giangregorio
  • 115
  • 7
  • 1
    you may find the following useful: https://stats.stackexchange.com/questions/185626/what-if-path-c-isnt-significant-but-paths-a-and-b-are-indirect-effect-in-medi/431714#431714 – PsychometStats Feb 16 '20 at 14:59
  • @PsychometStats thanks, I read the paper and seems very interesting. I will use Zhao et al. for the interpretation, as I also used SEM and not the standard 3-steps. – Luca Giangregorio Feb 16 '20 at 16:44
  • I'm glad you've found it useful! Also, well done! Using Kenny's 3-step approach is outdated and methodologically flawed – PsychometStats Feb 16 '20 at 17:51

0 Answers0