Logistic regression and likert scale

[david pesantes]

Hello guys!

I am estimating a logit model with "adoption" as the dependent variable, however as I am measuring risk perception some of my desciptive variables are ordinany (1-7 likert scale). My question is regarding the interpretation of these indepedent varibles in my model. Should I treat them first? (i.e. by using CPA). Or how should I proceed? Thank you in advance!

[Richard Williams]

I don't quite understand your question. But if you want to see if ordinal independent variables can be treated as continuous, see

https://www3.nd.edu/~rwilliam/xsoc73...ndependent.pdf

[david pesantes]

Thank you very much, Richard! I apologize that my inquiry was not quite clear, however the PDF that you shared with me has been very useful and it is exactly what I was looking for. Like if I can treat the ordinal variable as a continous - or as a categorical variable - and how to determine that.

Thanks!

[Nick Cox]

It's hard to see that the 7-level ordinal predictors are helping much at present, as many P-values do not fall below conventional levels and the patterns of coefficients suggest that many predictors are picking up noise rather than ordered effects. Treating such predictors as continuous would be more frugal with degrees of freedom and number of parameters estimated. Whether it is a good idea is more difficult to judge remotely. You will need to justify that for most readers. For only a moderate sample size, the model can't be called parsimonious.