Timeline for Accounting for time in repeated measures glmm, R
Current License: CC BY-SA 3.0
6 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 25, 2018 at 22:51 | comment | added | J.Con | Okay, thanks so much for your patience and perseverance. You earned the bounty. | |
| Feb 25, 2018 at 22:50 | history | bounty ended | J.Con | ||
| Feb 25, 2018 at 22:50 | vote | accept | J.Con | ||
| Feb 25, 2018 at 22:44 | comment | added | Frans Rodenburg |
An observation level random effect is the same as the residual term, you cannot add that to a model. However, negative binomial GLMM is implemented in lme4, see glmer.nb(). The issue with destructive sampling is that you can't really say what the effect of time is easily, since this is influenced by the sampling itself.
|
|
| Feb 25, 2018 at 22:34 | comment | added | J.Con | thank you so much, this is a very comprehensive answer. I feel much more confident in my model now. My only concern is when you say 'so long as the destructive sampling you mention does not affect the total number of snails too drastically', the sampling would have reduced the total population by approximately one fourteenth each time. Do you think this amount is small enough to not be of major concern? Also I see glmer no longer supports quasi- families, would an alternative to negative binomial distribution be an observation level random effect to account for over dispersion? | |
| Feb 25, 2018 at 0:22 | history | answered | Frans Rodenburg | CC BY-SA 3.0 |