In accordance with the recommendations in the SPSS statistics book by Andy Field, I have tested the assumption of linearity of the logit assumption for my continuous variables by including interaction terms between explanatory variable and in natural log.
In each model, at least one of my models is showing a p value of less than .05. What I am struggling to work out is now what to do about it? I only have experience of working with SPSS, so don’t know how to address the problem. Do I just exclude the variable, or do I recode it somehow?
I have been searching on the internet for the last week or so for a solution. Although I know how to test for whether this assumption has been upheld…I don’t know what to do about the variables that are not fitting this assumption.