Negative Binomial Regression- General Question

#1
If I have an independent variable (X1) that only takes integer values, why do I log that in the regression model? (My dependent variable (Y) is also a count variable.

I don't know if it has to be done this way....but in many papers I see that people always log count variables

e.g. Y = a log(X1) +......


Thanks
 

TheEcologist

Global Moderator
#2
If I have an independent variable (X1) that only takes integer values, why do I log that in the regression model? (My dependent variable (Y) is also a count variable.

I don't know if it has to be done this way....but in many papers I see that people always log count variables

e.g. Y = a log(X1) +......


Thanks
A log transform does not have to be done in your situation [but will depend on e.g. the model fit results, the trend of the data ect].

Usually people do that to obtain normality, linearise an exponential trend or maybe even to reduce over dispersion.