Timeline for Mixed model with 1 observation per level
Current License: CC BY-SA 3.0
3 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 25, 2013 at 7:24 | comment | added | Hong Ooi |
Thanks for the links! After reading those, and having a closer look at the fitted values from my model, I have a better idea of what's going on. I actually don't think what Ben is doing is appropriate for my analysis. He's using an observation-level variable to allow for overdispersion, so it's like a nuisance effect. For my analysis, distributor is an effect of interest: I want to see how distributors perform relative to each other when allowing for other variables. Thus it's more comparable to a conventional linear mixed model, where overfitting is a genuine concern.
|
|
| Jul 24, 2013 at 22:08 | comment | added | usεr11852 |
+1. I agree with what you say. As I mentioned in my original post : "over-dispersion (...) is how you "get around" the issue of having as many groups as observations." Thank you for making a better point of glmer in a conceptual manner.
|
|
| Jul 24, 2013 at 21:48 | history | answered | David J. Harris | CC BY-SA 3.0 |