Hi, is it possible to have a negative intercept estimate for count related data in a negative binomial regression? the p value is also greater than .05 while the other predictors are all significant.
Jochen Wilhelm
Usually, the negative binomial model is used with the log link, so the intercept may mean "log(counts)". A negative value then just means that the expected count (what all x are zero) is smaller than 1.
1 votes
1 thanks
Antonio Pedraza
Thanks, Ive since formed the belief that even if it is negative the amount on insight you can get in my model when all variables are zero is pretty irrelevant because that situation would never happen
0 votes
0 thanks
Mahmoud Elmorshedy
The negative values tends to the expected count is less than one.
0 votes
0 thanks
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression
More Antonio Pedraza's questions See All
Similar questions and discussions