Hello everyone,
I've got a question regarding how to interpret continious variables in logistic regression. In my model, I have continious variables for age and age squared. Also, I have a few ordinal variables that are "treated" as continious variables in the logistic regression, i.e. I don't have "i." before the variable. I was adviced to do this from another professional in order to save space in the table.
My command for the logistic regression:
logit VD i.VAA2 i.kön ålder ålder2 i.utb i.inkomst i.civilstatus i.bostadsort i.ff polint i.polid ideologi poleff partiskillnader käbbel i.övertygad2
"ålder", "ålder2" are the continious variables for age, and "polint", "poleff", "partiskillnader" and "käbbel" are ordinal variables with 4 response options ranging from e.g not interested to very interested. "ideologi" has 10 response options ranging from left to right. When I type margins, dydx (*) atmeans, I get the following result:
Array
But what does essentially "0,058" mean for the "polint" variable for example? I'm aware of that I need to type for continious variables margins, at (polint=(1 2 3 4)) atmeans to get the predicted probabilities for each response option with likert scale variables, but can I say anything about "0,058"? Below are the predicted probabilities for each response option for the "polint" variable. And what about the predicted probabilities of 0,0037 for the continious age variable "ålder"? (I know that I need to type e.g. margins, at (ålder=(20 (10) 84)) atmeans). What do that predicted probability tell?
Array
Thank you very much in advance.
Best regards,
Kajsa
0 Response to How to interpret predicted probabilities with continious variables in logistic regression?
Post a Comment