Timeline for How to retrieve standard errors from random effects in nlme?
Current License: CC BY-SA 4.0
17 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Nov 20, 2023 at 0:15 | history | edited | kjetil b halvorsen♦ | CC BY-SA 4.0 |
added 27 characters in body
|
| Nov 20, 2023 at 0:07 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Jul 22, 2023 at 8:04 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Apr 15, 2023 at 23:54 | answer | added | DrJerryTAO | timeline score: 1 | |
| Nov 1, 2022 at 7:04 | comment | added | Thomas | Thanks a lot for your valuable input. It makes a lot of sense! | |
| Nov 1, 2022 at 5:59 | comment | added | Roland |
If you model states as a random effect, you assume they are a random sample from a normal distribution. This not compatible with testing for differences between them with a hypothesis test (because that implies that you do not believe the assumption to be valid). If you really think testing for pairwise comparisons is sensible, you should model states as a fixed effect. In my opinion, you should not do that many pairwise comparisons.
|
|
| Oct 31, 2022 at 16:42 | comment | added | Thomas | I am not sure I follow your logic. How can I do a pairwise comparison of e.g. the Intercept per state as I stated in the edit part? | |
| Oct 28, 2022 at 12:01 | comment | added | Roland | You can always do pairwise comparisons but doing so is not compatible with considering the effect random. | |
| Oct 28, 2022 at 11:58 | history | edited | Thomas | CC BY-SA 4.0 |
added 304 characters in body
|
| Oct 28, 2022 at 11:39 | comment | added | Thomas | Okay, thank you. Can I do a pairwise comparison of the intercept values related to my random effects (states) and not only the fixed effects? | |
| Oct 28, 2022 at 4:52 | comment | added | Roland | Yes, ecoregions were fixed effects. Comparing them with a post-hoc test is easy with the emmeans package. If you want to do pairwise comparisons, that is a good indicator that you are dealing with a fixed effect. | |
| Oct 27, 2022 at 19:17 | comment | added | Thomas | Oh, I can relay that information from the methods section: "To model above- and belowground C pools and C combustion as a function of ecoregion group (4 levels), we fitted generalized linear mixed-effects models with hierarchical random effects of projects (4 levels) and individual fires nested within projects (18 levels) using the package ‘nlme’41. These random effects allow for varying intercepts(...)" | |
| Oct 27, 2022 at 16:09 | comment | added | Roland | I strongly doubt that they modeled the ecoregions as random effects. I might have a look at the paper tomorrow. | |
| Oct 27, 2022 at 13:59 | history | edited | Thomas | CC BY-SA 4.0 |
added 363 characters in body
|
| Oct 27, 2022 at 13:56 | comment | added | Thomas | Hi @Roland. Sorry, I just want to test if these effect sizes are different from one another, see edit. Thank you for your response. | |
| Oct 26, 2022 at 7:03 | comment | added | Roland | Can you please explain why you believe you need standard errors for the BLUPs and why you believe such standard errors would be a sensible measure? I suspect we have an XY problem here. | |
| Oct 25, 2022 at 16:39 | history | asked | Thomas | CC BY-SA 4.0 |