Timeline for Using Zero inflated GlMM when you have too many zeros
Current License: CC BY-SA 4.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Sep 25 at 12:07 | comment | added | Tammy | I also did the VIF as an additional check. Thanks for the latter part about the collinearity. This gives me a few possible ideas about what could be going on. Will check again. True, it is a lot of data and hard to explain without being able to share it. so I appreciate any advice I get. | |
| Sep 25 at 12:00 | comment | added | Peter Flom | First, correlation is not always a good measure of collinearity. My favorite measure is condition indexes, but VIF is also good. Collinearity can exist among 3 or more variables without very high correlation among any 2. But if collinearity is there, it is there. It can't be there for some species and not others Second, fully diagnosing a problem like this probably requires an expert to have full access to your data. | |
| Sep 25 at 11:55 | comment | added | Tammy | Hi Peter, Thanks for your response. When I use all 8 variables, the glmer seems to work for a few species, but for others it keeps giving me an error saying the model couldn't converge. Honestly, I don't understand why I get this error, because at the very start, I check for correlation and remove those correlated variables (or just use one of them). On the other hand, when I run the glmer with those species where it doesn't take 8 variables, it seems to work when there are less variables (mostly 3/4). I don't understand why this is so either, leading me to the zero inflation. | |
| Sep 25 at 11:44 | history | answered | Peter Flom | CC BY-SA 4.0 |