I ran initially a causal steps approach between a variable (A), a predictor (I), and a moderator (G).
Using this approach i found that A correlates to I; I correlates to G; A correlates to G
and entering everything into the model shows a full mediation of I through G for A (i.e. I is no longer sig for A).
Then I decided to use PROCESS and it did not find the initial I and G effect to be significant, and the indirect effect was not significant. finding a single effect of G on A.
I know it's confusing, but which one should I trust? maybe I did something wrong in PROCESS?
Help!
Due to different algorithm, it's common to get different results from different approaches. At the moment Preacher and Hayes is the approach commonly used for mediation tests. Look at the LLCI and ULCI of your PROCESS output to determine if there is mediation effect.
Thank you, I found it was due to selecting the heteroskedasticity option for SE.
Given that the "causal steps" approach has some serious problems, I would hesitate to rely on it.
For an overview, see:
MacKinnon, D. P., Coxe, S., & Baraldi, A. N. (2012). Guidelines for the investigation of mediating variables in business research. Journal of Business and Psychology, 27(1), 1-14.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis