Log-transform IV that uses count data

by statsnewby   Last Updated November 17, 2017 17:19 PM

I am running a negative binomial regression model and one of my predictors is a count variable (no zeros are present). I am mainly interested in the betas of the regression function. Now I am wondering whether I have to log-transform that variable or whether I can continue with the un-transformed variable, which eases treatment considerably.



Related Questions