Timeline for generalized linear model with log link using log transformed fixed/random effects?
Current License: CC BY-SA 4.0
12 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Sep 10, 2019 at 18:23 | vote | accept | user250071 | ||
| Sep 9, 2019 at 21:19 | answer | added | EdM | timeline score: 2 | |
| Sep 9, 2019 at 20:14 | comment | added | user250071 | In order to help narrow the focus, please just apply the question of how a log transformed fixed/random effect would impact a log-linked model to a model that meets your specifications. | |
| Sep 9, 2019 at 20:03 | history | edited | user250071 | CC BY-SA 4.0 |
added 7 characters in body
|
| Sep 8, 2019 at 7:21 | answer | added | Dimitris Rizopoulos | timeline score: 3 | |
| Sep 8, 2019 at 7:10 | history | edited | Dimitris Rizopoulos | CC BY-SA 4.0 |
added 35 characters in body
|
| Sep 7, 2019 at 1:41 | comment | added | Isabella Ghement | Let me read the article you suggested. | |
| Sep 7, 2019 at 0:21 | history | edited | user250071 | CC BY-SA 4.0 |
clarified question
|
| Sep 7, 2019 at 0:15 | comment | added | user250071 | Thank you for the link and info. It is my understanding that although using continuous random effects is not typical, it is not incorrect (link below). Regardless, assuming the model met your specifications, what about my question regarding log transformation? cell.com/trends/ecology-evolution/fulltext/… | |
| Sep 6, 2019 at 23:51 | comment | added | Isabella Ghement | The random grouping factors in a glmer model such as yours - namely, TimeLog and Sample - should both be categorical variables. However, TimeLog is a continuous variable so it is incorrect to use it as a random grouping factor as you have done through the (1|TimeLog) term. You may consider posting a separate question on this forum about the correct specification of your model given your study design and research questions. | |
| Sep 6, 2019 at 23:45 | comment | added | Isabella Ghement | If your outcome variable, Mutations, is a count, why not use a glmer with poisson family and log link? The Gamma distribution is typically used in situations where the outcome variable is continuous rather than a count. See this link for a nice discussion - albeit in the context of glm rather than glmer models - on when to use the Gamma distribution: stats.stackexchange.com/questions/67547/when-to-use-gamma-glms. | |
| Sep 6, 2019 at 21:59 | history | asked | user250071 | CC BY-SA 4.0 |