Timeline for Validity of AIC When Comparing Models with Varying Dispersion Parameters
Current License: CC BY-SA 4.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 15 at 8:17 | comment | added | Daniel | Hey, thanks for the response! I have a quick question about the overdispersion parameter. So in the formula, I'm guessing $\phi$ is equal to the true overdispersion of the model, and we mainly just substitute the approximated value based on the model we currently have. Usually that's good enough, but in the model I'm making, since I have hundreds of thousands of observations, a 0.001 change in the parameter is changing the qAIC by hundreds. Is that an expected result, and are these models comparable, or should I be holding the estimated phi constant? Thanks! | |
| Jul 15 at 8:02 | vote | accept | Daniel | ||
| Jul 14 at 17:36 | history | edited | Frans Rodenburg | CC BY-SA 4.0 |
deleted 1 character in body
|
| Jul 14 at 7:05 | history | answered | Frans Rodenburg | CC BY-SA 4.0 |