How to do Interaction/Moderation in Logistic Regression?

#1
Hi guys, I need some help with my research.

I want to research about the relationship between Corporate Governance (X1) (consists of many proxies, but are turned into a score, using factor analysis) which is a continuous IV and Financial distress (Y), which is categorical binary DV (1 is for financially distressed firms, 0 is otherwise) and use a third variable, which is Intellectual Capital Score (Z), a continuous variable as a moderation/interaction variable in order to know whether Intellectual Capital strengthen this relationship or not.

I am told to do a Logistic Regression, but I don't know how to do it if I want to add a moderation/interaction variable into it.

Also, should I include a preliminary test, such as normality test or goodness-of-fit test? If so, which test that I would require to run?

I'm using SPSS.

Thank you!
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
Interaction terms are typically entered into the model as multiplicative interaction terms for the two variables, plus you keep their base terms in the model.

y-hat = X1 + Z + X1*Z
 
#3
Interaction terms are typically entered into the model as multiplicative interaction terms for the two variables, plus you keep their base terms in the model.

y-hat = X1 + Z + X1*Z
So it is possible to run moderation/interactions in logistic regression? If it is possible, is there any specific tutorial or steps to do it that you would recommend to me?
 
#4
As far as I know interaction is no different in logistic than linear regression. The interpretation would be different to some extent because it is in the context of the logit. I have not run across a tutorial for this (which I don't use often in honesty).
 

hlsmith

Less is more. Stay pure. Stay poor.
#5
Yeah as @noetsi mentioned - there really isn't anything addition to think about. Of note, you would no longer interpret the base terms, just the interaction term.

I would imagine there are countless tutorials on Youtube.
 
#6
Yeah as @noetsi mentioned - there really isn't anything addition to think about. Of note, you would no longer interpret the base terms, just the interaction term.

I would imagine there are countless tutorials on Youtube.
Well only interpreting the interaction term is true of linear regression as well. You can generate simple effects for the main effects.
 
#7
As far as I know interaction is no different in logistic than linear regression. The interpretation would be different to some extent because it is in the context of the logit. I have not run across a tutorial for this (which I don't use often in honesty).
Ohh I see. Thank you so much for the advice!

Yeah as @noetsi mentioned - there really isn't anything addition to think about. Of note, you would no longer interpret the base terms, just the interaction term.

I would imagine there are countless tutorials on Youtube.
Yes, I think I'll try to explore some tutorials in Youtube, I wasn't sure about whether it is possible or not, but now I think its worth to try! Thank you!