Hi, I have quite a complex model that I'm trying to test. I've been using the Hayes PROCESS macro up to this point for my mediation/moderation testing, but I'm not sure if I can do the next test with it.
Attached is a diagram of the model I am trying to test. There is a serial mediation, possibly with a moderation (though my initial tests suggest it won't have a significant influence). Additionally, a second independent variable is expected to influence the last mediating variable before we get to the DV.
I have tried doing the test using covariation (models 6 and 84), and without the moderator, using model 80. Covariation doesn't seem to do what I'm trying to do given where the second IV drops in, and model 80 won't work because it will always assume the first IV as IV for the whole model.
Any suggestions on how to tackle this?
I recommend this youtube channel so you can better understand your analysis.
https://www.youtube.com/watch?v=8CjJosnbEU0&t=977s
I recommend this youtube channel so you can better understand your analysis. I think you can also separate your analysis into two blocks to accommodate the individual effects of mediation and moderation and adjust the model you propose.
Unsure I understand the details of your questions pertaining to specific "numbers/titles" of the models discussed. However, overall, I suggest simplifying your models into bite size pieces and running your mediation or moderation analyses separately using the PROCESS due to the limitation of the macro. I understand the macro can handle a limited number of mediators/moderators in one model. This might help specially since you have 2 exposures that work in separate serial and simple mediation paths as demonstrated in your conceptual framework.
Also, I suggest reviewing the book by Hayes (Introduction to mediation, moderation, and conditional process analysis: a regression-based approach) to develop a covariate matrix for running your adjusted models.
Dear Matt,
I think you can easily resolve this with a structural model approach (path analysis). In Mplus you can do that. You can easily specify the model that you refer to and obtain the estimated regression path, including the moderator´s effects (ore controlled paths for these variables), and several fit indexes.
Dear Pablo Ezequiel Flores Kanter
How can we report the analysis in the method section of a paper or proposal? What are the things that we need to report? Have you known any resources for reporting path analysis?
Dear Mohammad Mohsen Khoddami ,
As a general rule one should follow the standard guidelines and be transparent about the steps implemented in the procedure. The specifics of each case depend on the analytical procedure implemented. For example, if SEM is applied, a good resource regarding standard reporting can be found at Appelbaum et al., 2018, page 18, table 7 (Article Journal Article Reporting Standards for Quantitative Researc...
). Regarding transparency, it is important to add complementary materials as well as files in external open access repositories that maximise the possibility of replication-reproducibility of the study itself (e.g. open databases, open code).All the best!!
- Badges
- Science topic
- Similar topics
- Academic Journals