Dear all,

I need to run two mediation analysis for a research project and after having studied the web and this forum for several hours I still have not come across a resource that could provide me with what exactly I need to do (I have already read Hayes book on it).
The first mediation has a binary independent variable (identifier), the mediator is not continuous (a frequency variable counting from zero to a maximum of 72 in my sample and only displays integers, hence not continuous, which as I understand, makes it problematic for the classic mediation analysis using SEM?!) and the dependent variabel is continuous. I need to include several control variables, which are partly binary and partly continuos.
The second mediation has a binary independent variable (identifier), the mediator is continuous and the dependent variabel is continuous. I need to include several control variables, which are partly binary and partly continuos.

When studying Hayes I thought that I would simple need to run three regressions, one for the direct and two for the indirect effect and then I could simply derive the indirect effect as the product of the two indirect regression coefficients and the total as the sum of the direct and indirect effect. Now however, I am confused about the test diagnostics and hence started to research further, only to realise that apparently the method I used is not correct.
So could anyone be so kind to explain to me what sort of mediation analysis I would need to run, and if I need to check some assumptions first? Also: What test statistics should I use in order to determine whether or not my mediator has a significant effect?
Thank you very much in advance!
Ps. While I could provide some exemplary data, I do not think it would add much value / clarity here. However, if it is required, I am happy to provide it.