Dear Statalist,
I fitted a multiple logistic regression model where each of my independent variables have a Wald test p value < 0.05, but when tested for the linearity of the continuous variables with the log OR of the outcome they were not linear.
If the continuous independent variables are NOT linearly related with the logarithm of the odds ratio, can you include them in the analysis to estimate the OR for each variable?
I am wondering if the linearity of the odds has only to be assumed when you want to use your model for making predictions out of the equation or you just cannot include them in the logistic regression.
Thank you !!
I fitted a multiple logistic regression model where each of my independent variables have a Wald test p value < 0.05, but when tested for the linearity of the continuous variables with the log OR of the outcome they were not linear.
If the continuous independent variables are NOT linearly related with the logarithm of the odds ratio, can you include them in the analysis to estimate the OR for each variable?
I am wondering if the linearity of the odds has only to be assumed when you want to use your model for making predictions out of the equation or you just cannot include them in the logistic regression.
Thank you !!
