Moderated Mediation in R

[@mungamwangi_]

This is so detailed, hats off Tidy Data. Blessings bro.

[@tidydata]

My pleasure! Happy to be useful :)

[@qichunduan]

Thank you! Can I also know how to get change of F and change in adjusted R square?

[@stellahkilawe8208]

Thanks very much for detailed lecture, It will help me alot be blessed. You made it for stranger , the flood gate of God Blessing will fall upon your life.

[@tidydata]

You are very welcome! Glad to be useful :)

[@gonzalotorres8685]

Thank you a lot, this is really detailed and helpful. What about latent variables and the process function? I've been trying to figure it out on my own but it's not working.

[@tidydata]

Hi, thank for your kind words. Glad to be useful. I could be wrong, but as I know, PROCESS can not handle latent variables. PROCESS deals with variables with an aggregated format (e.g., the mean of a few items). You might need to use Mplus or some other R packages to deal with latent variables. Unfortunately, I have a limited understanding of latent variables and thus probably can not be helpful here. If the measure is just continuous (e.g., Likert scale measures), a simple mean of items can be used to test mediation model, IF you do NOT have specific reasons why you need to add latent structures to mediation models.

[@sharonmatzkin979]

Thank you very much for this wonderful tutorial! My IV has 5 categories and my W has 3 categories. How do I specify them as multicategorical in R? (I know how to do this in SPSS, but am trying to replicate my results in R). Thank you again!

[@tidydata]

If your IV has 5 categories and these categories cannot "approximately" be considered as "somewhat" linear, I personally do not know how to do it in SPSS PROCESS (I have never done it myself). When checking online, I found a paper by Hayes (the PROCESS author) on this: quantpsy.org/pubs/hayes_preacher_2014.pdf. The reason I focus on PROCESS is that, if PROCESS can be used to handle this situation, the PROCESS in R and SPSS should generate the same result. Hope my comment helps.

[@manjulaseneviratna4865]

When I do this process Model 7 with my SPSS, I got a positive sign for my moderator (w) and negative sign for Int_1 . I think this mean there is a positive moderation. I would like to know why I got negative sign for Int_1 ? Please help me

[@tidydata]

Hi, int_1 is the interaction, namely the moderation, not the moderator (i.e.,w), even though conceptually it is the moderator. You need to plot it, either via bar chart (if both X and w are categorical variables) orJohnson Neyman (for x and w, one categorical and one numerical; or even both are numerical). What you need to do is: (1) if int_1 is significant, reporting that the interaction effect is significant. (2) then, do the plot to tell your readers how the significant effect looks like. (Disclaimer: This is just my opinions and based on my best knowledge. Please verify by yourself. I do not take any responsibility if there is error, but I do my best already.)