# SPSS: Testing moderating effect in binary logistic regression

#### Melox

##### New Member
Hey, I presume the answer to my question is fairly straightforward but im really having a hard time understanding how to approach this.

I am conducting a binary logistic regression with dependent variable Conversion (0 = No, 1 = Yes) and three independent variables; Group (Control vs Experimental), Customer Type (New vs Existing Customer) and Culture (UK and US). As you can tell, all variables included are dichotomous binary variables. I am looking to test moderating effect of culture on the relationship between Group and Conversion.

I am aware SPSS allows to easily create interaction between two variables with the <a*b< function, however I am not so sure how to go about testing the interaction. Should I simply input the interaction variable with the remaining independent variables once and interpret the results as usual? Or does the fact that all variables being tested are dichotomous and dummy coded influence the correct procedure?

Any help is hugely appreciated!

Thanks )

#### GretaGarbo

##### Human
Should I simply input the interaction variable with the remaining independent variables once and interpret the results as usual?
Yes, you should!

#### Melox

##### New Member
Yes, you should!

Considering the variables being tested are both binary; Group (0 = Control, 1 = Experimental) and Country (0 = UK, 1 = USA), wont the results only be relevant for the interaction when both variables have a value of 1 (Experimental and USA)? Or will the results also indicate coefficient regarding observations within the experimental and UK group?

Thank you for the help

#### GretaGarbo

##### Human
That is called the corner point parametrication:
Group (0 = Control, 1 = Experimental) and Country (0 = UK, 1 = USA),

You can change it if you want to to the-classic-sum-to-zero-parametrication:

Group ( -1 = Control, +1 = Experimental) and Country ( -1 = UK, +1 = USA)

But they both will give the same result in that the the p-value from the test will be the same and the predicted values will be the same.