SPSS 25 has no official commands for carrying out mediation analysis. Many SPSS users use the PROCESS macro. This YouTube video shows that it will automatically use logistic regression rather than linear regression when the outcome variable (Y) is binary.

• https://www.youtube.com/watch?v=bpaQ96Euafg

HTH.

The exposure is categorical variable, outcome binary and mediation variable is continuouns variable.

If exposure is binary (1=exposed, 0=not exposed), you can treat it as continuous.

No, the exposure is tertiles (1=low, 2=intermediary, anda 3= high), the reference being the low

Low, intermediate, and high what? "Tertiles" suggests you have a quantitative variable that was carved into 3 categories. What is that quantitative variable? And why can you not use it as the explanatory (exposure) variable in your model(s)? Thanks for clarifying.

Hi Bruce Weaver !

My exposure variable is the dietary intake tertile, and yes, it is a three-level quantitative variable... It is being used in the model as an exposure and gastric cancer as an outcome. So can I continue with the mediation analysis according to the example in the video even though my variables have these characteristics that I mentioned????

I am not a nutritionist, so pardon my ignorance. How is dietary intake measured? What are the units? Calories? Or something else?

Bruce Weaver, consumption tertiles are expressed in g/day. Calories is expressed in kcal/day. Calories is the possible mediating variable that I will use.

Thanks for clarifying. Why do you think you need to carve g/day into tertiles? Carving quantitative variables into categories can cause many problems. See this page for more info:

• https://discourse.datamethods.org/t/categorizing-continuous-variables/3402