I'm using a 2-vector response = cbind(number.of.successes, number.of.failure) as recommended in CRAWLEY. In order to figure out variable importance in models including response = abundance we used the function hier.part (package hier.part) which calculates the independent contribution of the variables. Unfortunately, I can't find any equivalent function/package that can handle the 2-vector response.
Hi Gregg, thanks for your advice! I will have a look into the package.
Hi Gregg, I tried glm( successes/n ~ x, data = *, weights = n, ... ) on one example. The code works fine and model performance (explained deviance) is a little lower in comparisson with the 2-vector response (2-vector = 0.66, successes/n = 0.58). But, the usage of weights doesn't work with the hier.part function, which brings me back to the original problem...
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R Statistical Package