Statistics newbie here. I have count data (the number of cherries infected with a certain pathogen out of all cherries counted from various orchards. The total number of cherries counted varies between orchards). I tried using poisson, but my data are over-dispersed. Quasi-poisson did not help. My research has indicated that I should use a negative binomial regression instead. How do I account for the fact that total cherries counted differed between orchards?
I was using:
glm(Infected_Count ~ Region + offset(log(Total_Count)), family="quasipoisson",data=countdata)
To use a negative binomial distribution, you could use glm.nb() from the MASS package instead of glm(). You've already accounted for the difference in campling effort between orchards with the offset. You can also include an offset with glm.nb().
Hi Mel,
How about analyzing the proportion of infected cherries (%) and total cherries in two different models? You can use the same or different distribution for two models.
Hope this helps!
Babu
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication