Basically, my study has a mixed design. I have a two-level X as between-subject. I measured Y twice (pre- and post-) (i.e. time as the within-subject) I measured the supposed mediator M once.
I'm seeking help for how to conduct the mediation analysis in my case.
Currently, I'm thinking of two ways:
1) Use the process model 4, apply X as IV, Y2-Y1 as the DV, M as mediator and perform the mediation analysis.
2) Use the process model 4, apply Y2 as DV, X as IV, M as mediator and Y1 as covariate and perform the mediation analysis.
I'm wondering is there any other method that makes more sense? Thanks in advance!
Option 2 would produce estimates that some fields refer to as, "value added analyses". You would be looking at the ability of X to have a direct effect (and M to have an indirect effect) on Y2 that is above and beyond that from Y1.
You could always open AMOS in SPSS and just estimate all the pathways that you care to estimate. AMOS will calculate the indirect and direct effects for you very easily. Barbara Byrne's book comes highly recommended: https://www.amazon.co.uk/Structural-Equation-Modeling-AMOS-Applications/dp/0805863737
I would suggest estimating X and Y1 to have both direct effects on M, and direct effects upon Y2. That will give you two indirect effects: X>M>Y2 and Y1>M>Y2. Add some covariates of your choice to predict Y1 and Y2 and you'll be conducting a, "contextualized value-added analysis of the direct and indirect effrects of X on Y".
I am attaching articles that may help. The mediation model is included.
Baron and Kenny describe clearly please refer to their article
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS