Dear all,
I would like to perform a mediation analysis and test direct effect of X on Y and verify if this relation could be mediate by M. I download the macro process for SPSS from Andrew Hayes.
Before running the process, I verified my homogeneity and my data are non-parametric. I found Significant correlation between the three variables with spearman test. However, when I run the macro process with the modele 4, I don't find the same result.....
I suppose that this process run for homogeneity data... so I would like to down to perform the mediation analysis with non-parametric data... I share my results in figure enclosed.
Thanks you in advance.
Bests
Olivier Dupuy
Article Toward non-parametric and clinically meaningful moderators a...
Article Incorporating nonlinearity into mediation analyses
http://www.math.sciences.univ-nantes.fr/~philippe/Stat_&_mediation_files/np_preprint.pdf
Hello Olivier,
Are your scores ordinal (as the reason for your declaring the data to be non-parametric), or is there some issue with bivariate normality?
Hayes's PROCESS macro, model 4 ("simple mediation"), presumes interval strength scale (and parametric tests for relationships). That alone could be the reason for some apparent discrepancy. The results you posted (if from Spearman correlations) would suggest that all three paths are non-zero, and that there is evidence for indirect effect of A on B via M (as well as direct effect).
Good luck with your work.
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publications