Log in
RegisterIt is the opposite way around for me [🇷 for BE/BA]
Hi again,
to me it is exactly the opposite way around:
A model with more than one variance component is a mixed model. I am of the impression that FDA and EMA have both been in a slight predicament with their guidances because they have both been "somewhat" aware of the specification/convergence issues with REML; hence both went in a direction with all effect s fixed as done explicitly by EMA and implicitly by FDA (when they use GLM for swr) in the guidance.
My contribution here is solely to open someone's eyes to the fact that we may not need to rely on a normal linear model for s2wr in a TRR/RTR/RRT design.
I would much more like to think of this thread as a contribution to regulators defining the next iteration of giudelines/guidances. Whether they will see it as an advantage to use REML to derive s2wr now it is shown to be doable (not causing convergence issues, because it is not involving the silly and unestimable s2wt component), is of course their decision solely. I have done my part.
For consideration further down the line, for those who enter discussions about estimator bias being catastrophic: Which estimator is more biased when handling RRT/RTR/TRR: the s2wr you get from the linear model or the one you get via REML? What would your answer imply for any future recommendations for guidelines/guidances?
(I am not a hardliner myself: I am usually thinking biased estimators are still useful estimators)
This thread already caught the attention of a number cruncher at an agency across the Atlantic unofficially, so I am OK with it all
to me it is exactly the opposite way around:
A model with more than one variance component is a mixed model. I am of the impression that FDA and EMA have both been in a slight predicament with their guidances because they have both been "somewhat" aware of the specification/convergence issues with REML; hence both went in a direction with all effect s fixed as done explicitly by EMA and implicitly by FDA (when they use GLM for swr) in the guidance.
My contribution here is solely to open someone's eyes to the fact that we may not need to rely on a normal linear model for s2wr in a TRR/RTR/RRT design.
I would much more like to think of this thread as a contribution to regulators defining the next iteration of giudelines/guidances. Whether they will see it as an advantage to use REML to derive s2wr now it is shown to be doable (not causing convergence issues, because it is not involving the silly and unestimable s2wt component), is of course their decision solely. I have done my part.
For consideration further down the line, for those who enter discussions about estimator bias being catastrophic: Which estimator is more biased when handling RRT/RTR/TRR: the s2wr you get from the linear model or the one you get via REML? What would your answer imply for any future recommendations for guidelines/guidances?
(I am not a hardliner myself: I am usually thinking biased estimators are still useful estimators)
This thread already caught the attention of a number cruncher at an agency across the Atlantic unofficially, so I am OK with it all
—
Pass or fail!
ElMaestro
Pass or fail!
ElMaestro
Complete thread:
- Semireplicated + REML in R ElMaestro 2020-07-10 13:19 [🇷 for BE/BA]
![[Mix]](https://static.bebac.at/img/mix.png "open in Mix view")
- Semireplicated + REML in R Helmut 2020-07-10 19:03
- Semireplicated + REML in R ElMaestro 2020-07-10 19:24
- Avoid partial replicate designs, pleeeze! Helmut 2020-07-11 11:57
- Avoid partial replicate designs, pleeeze! ElMaestro 2020-07-11 14:43
- Braveheart! Helmut 2020-07-11 15:38
- Who can help? ElMaestro 2020-07-12 12:28
- Who can help? ElMaestro 2020-07-12 13:26
- Update II ElMaestro 2020-07-12 21:36
- Update III ElMaestro 2020-07-12 21:46
- Final update today ElMaestro 2020-07-12 22:27
- Medium rare. Helmut 2020-07-13 13:52
- took just 52 hrs to do it :-) ElMaestro 2020-07-13 14:18
- Will take much more hours still… Helmut 2020-07-13 15:34
- Negative determinant ElMaestro 2020-07-14 03:22
- Are we loosers? Helmut 2020-07-14 13:58
- "we"? Loosers? ElMaestro 2020-07-14 15:07
- Misunderstanding? Helmut 2020-07-14 15:32
- "we"? Loosers? ElMaestro 2020-07-14 15:07
- Are we loosers? Helmut 2020-07-14 13:58
- took just 52 hrs to do it :-) ElMaestro 2020-07-13 14:18
- Medium rare. Helmut 2020-07-13 13:52
- Braveheart! ElMaestro 2020-07-13 10:13
- Braveheart! Helmut 2020-07-13 14:16
- Braveheart! PharmCat 2020-07-15 14:19
- Braveheart! ElMaestro 2020-08-02 17:39
- Who can help? ElMaestro 2020-07-12 12:28
- Braveheart! Helmut 2020-07-11 15:38
- Avoid partial replicate designs, pleeeze! ElMaestro 2020-07-11 14:43
- Avoid partial replicate designs, pleeeze! Helmut 2020-07-11 11:57
- Semireplicated + REML in R ElMaestro 2020-07-10 19:24
- We were all blind (except Detlew) Helmut 2020-07-15 14:27
- It is the opposite way around for meElMaestro 2020-07-15 16:25
- Desultory thoughts Helmut 2020-07-15 17:33
- FDA RSABE is ISC d_labes 2020-07-15 18:13
- FDA RSABE is ISC Helmut 2020-07-16 11:11
- Desultory thoughts ElMaestro 2020-07-15 23:06
- Desultory thoughts Helmut 2020-07-16 10:59
- FDA RSABE is ISC d_labes 2020-07-15 18:13
- Desultory thoughts Helmut 2020-07-15 17:33
- Phoenix - which template? mittyri 2020-07-19 00:42
- FDA RSABE Project template_ v1.4.phxproj Helmut 2020-07-19 01:45
- It is the opposite way around for meElMaestro 2020-07-15 16:25
- "By popular demand": likelihood ElMaestro 2020-07-24 10:07
- And by the way.... ElMaestro 2020-07-24 12:52
- And by the way.... PharmCat 2020-08-03 14:24
- Not understood ElMaestro 2020-08-03 22:55
- Not understood PharmCat 2020-08-05 01:41
- Not understood ElMaestro 2020-08-05 08:13
- Not understood PharmCat 2020-08-05 16:37
- Open issues ElMaestro 2020-08-06 21:11
- Open issues PharmCat 2020-08-07 00:02
- Open issues ElMaestro 2020-08-07 07:49
- Open issues PharmCat 2020-08-07 11:29
- Still can't make it work ElMaestro 2020-08-07 13:08
- Still can't make it work PharmCat 2020-08-07 15:42
- Still can't make it work ElMaestro 2020-08-07 16:44
- Still can't make it work PharmCat 2020-08-07 18:14
- Still can't make it work ElMaestro 2020-08-07 18:23
- And now it works ElMaestro 2020-08-07 21:31
- Still can't make it work PharmCat 2020-08-08 01:08
- Speed improvement ElMaestro 2020-08-08 12:25
- Speed improvement PharmCat 2020-08-08 17:27
- Speed improvement ElMaestro 2020-08-08 18:10
- Speed improvement PharmCat 2020-08-09 18:22
- Some tests... PharmCat 2020-08-10 11:48
- Speed improvement ElMaestro 2020-08-08 12:25
- Still can't make it work ElMaestro 2020-08-07 18:23
- Still can't make it work PharmCat 2020-08-07 18:14
- Still can't make it work ElMaestro 2020-08-07 16:44
- Still can't make it work PharmCat 2020-08-07 15:42
- Still can't make it work ElMaestro 2020-08-07 13:08
- Open issues PharmCat 2020-08-07 11:29
- Open issues ElMaestro 2020-08-07 07:49
- Open issues PharmCat 2020-08-07 00:02
- Open issues ElMaestro 2020-08-06 21:11
- Not understood PharmCat 2020-08-05 16:37
- Not understood ElMaestro 2020-08-05 08:13
- Not understood PharmCat 2020-08-05 01:41
- Not understood ElMaestro 2020-08-03 22:55
- And by the way.... PharmCat 2020-08-03 14:24
- And by the way.... ElMaestro 2020-07-24 12:52
- Semireplicated + REML in R Helmut 2020-07-10 19:03
Ing. Helmut Schütz