Interaction effects of nominal and ordinal variables on binary dependant variable

[knedlica]

Hi!

I'm doing an analysis of interaction effects of one binary nominal variable (medication used), one ordinal variable (duration of therapy ; 1-few weeks, 2-less than 3 months, 4-less than six months, 5-more than six months) and one interval variable (age) on binary dependant variable (0- does not have a symptom, 1-does have a symptom).

I know i can use logistic regression to get individual effects of each independent variable on propability of having a symptom but I'm confused about how to test the interaction effects.

From what I know the regression analysis can be used for this purpose when the predictor (independant) variable is interval but can it be used in this case?? From what I understand when doing so the predictor variables have to be standardized but it doesn't make much sense to me to standardize nominal and ordinal variables.

What should I do?

Thank you!

 

[hlsmith]

I believe you can introduce an interaction term into the model (nominal*ordinal) and look for significance in the beta coefficients. I have little experience standardizing, but I agree that standardizing is usually withheld for continuous variables and some feel it is problematic with other types of variables.

 

[knedlica]

Ok, thanks! I'll try just putting the interaction term into the logistic model without standardization.