Log in
RegisterLess or More of an issue [Regulatives / Guidelines]
Dear ElMaestro,
I don't know if you are right here.
As far as I had understand it to now (but my brain is here some smaller than a walnut
) the natural way in an ANOVA model with all effects fixed is to use the error MS as denominator in the F-test of all of the effects.
This is why all statistical software, which I know off, come out with exactly these tests if no option is chosen.
I think you are mixing up estimation with test methods based on the ANOVA model. You are right in stating that Proc GLM fits the model as fixed, i.e. the the model parameters are estimated via least squares, which is identical to the likelihood estimates in that case.
But then after that it modifies the tests (called in a post hoc manner) according the model with subject as random by the RANDOM statement. Call it bogus if you like.
Or you make the appropriate test "by hand" as in the bear R-Code using lm() or in the former SAS code for cross-over studies we had here in the forum. Again you are estimating by least squares and adjust the tests appropriately.
I strongly assume (but had questioned because not sure) that the text of the guidance is aimed with respect to the estimation method by least squares. My reasoning is along your reasoning in respect to the 3x3 or 4x4 crossover study above: straight forward simple method, known and at hand by regulators, against 'complicated' iterative method in case of 'real' mixed model software, giving not always the same result, lacking also the well known ANOVA table of the least square solution.
❝ This I don't think is an issue at all. Even in spite of the bogus statement, Proc Glm fits the entire model as a fixed effects model, which is good enough for 2-seq 2-per 2-trt data. ...
I don't know if you are right here.
As far as I had understand it to now (but my brain is here some smaller than a walnut
) the natural way in an ANOVA model with all effects fixed is to use the error MS as denominator in the F-test of all of the effects.This is why all statistical software, which I know off, come out with exactly these tests if no option is chosen.
I think you are mixing up estimation with test methods based on the ANOVA model. You are right in stating that Proc GLM fits the model as fixed, i.e. the the model parameters are estimated via least squares, which is identical to the likelihood estimates in that case.
But then after that it modifies the tests (called in a post hoc manner) according the model with subject as random by the RANDOM statement. Call it bogus if you like.
Or you make the appropriate test "by hand" as in the bear R-Code using lm() or in the former SAS code for cross-over studies we had here in the forum. Again you are estimating by least squares and adjust the tests appropriately.
I strongly assume (but had questioned because not sure) that the text of the guidance is aimed with respect to the estimation method by least squares. My reasoning is along your reasoning in respect to the 3x3 or 4x4 crossover study above: straight forward simple method, known and at hand by regulators, against 'complicated' iterative method in case of 'real' mixed model software, giving not always the same result, lacking also the well known ANOVA table of the least square solution.
—
Regards,
Detlew
Regards,
Detlew
Complete thread:
- EMA BE guideline - final version Ohlbe 2010-01-28 18:31 [Regulatives / Guidelines]
![[Mix]](https://static.bebac.at/img/mix.png "open in Mix view")
- Final version published Helmut 2010-01-29 14:29
- Final version published ElMaestro 2010-01-29 17:02
- Statistix Helmut 2010-01-29 17:30
- Statistix ElMaestro 2010-01-29 18:11
- Calories ;-) Helmut 2010-01-29 19:22
- Calories ;-) ElMaestro 2010-01-29 20:43
- Calories ;-) Helmut 2010-01-29 21:08
- Variable calories d_labes 2010-02-01 12:05
- Calories ;-) ElMaestro 2010-01-29 20:43
- ANOVA only - no doubts Helmut 2010-02-10 23:41
- ANOVA only - no doubts ElMaestro 2010-02-12 21:09
- Statistix Helmut 2010-01-29 17:30
- Final EMA oracle d_labes 2010-02-01 11:53
- Final EMA oracle Helmut 2010-02-01 12:28
- Final EMA oracle ElMaestro 2010-02-01 12:50
- More then two EMA oracles d_labes 2010-02-01 13:30
- Less of an issue ElMaestro 2010-02-01 15:29
- Less or More of an issued_labes 2010-02-01 16:53
- Less or More of an issue ElMaestro 2010-02-01 17:22
- Less or More of an issued_labes 2010-02-01 16:53
- Effective with 1 Aug 2010 Helmut 2010-02-02 00:55
- Bias? ElMaestro 2010-02-04 11:27
- Less of an issue ElMaestro 2010-02-01 15:29
- Final EMA oracle Helmut 2010-02-01 13:43
- Final EMA oracle ElMaestro 2010-02-01 14:54
- Final EMA oracle d_labes 2010-02-01 15:20
- Teaching Helmut 2010-02-01 16:07
- Final EMA oracle d_labes 2010-02-01 15:20
- Final EMA oracle ElMaestro 2010-02-01 14:54
- More then two EMA oracles d_labes 2010-02-01 13:30
- Cmin really gone? tmax reappeared? d_labes 2010-02-01 14:02
- Cmin really gone. tmax reappeared - but how? Helmut 2010-02-01 23:30
- Interpol or not d_labes 2010-02-01 14:52
- Interpol! Helmut 2010-02-02 00:02
- Interpol! ray_be 2010-02-03 18:31
- WinNonlin/Phoenix extrapolation to t=tau Helmut 2010-02-03 20:26
- Interpol! ray_be 2010-02-03 18:31
- Interpol! Helmut 2010-02-02 00:02
- Meta-analysis Helmut 2010-02-07 17:41
- Final version published Panks.79 2010-03-08 07:22
- Cmin for MR-formulations? Helmut 2010-03-08 18:16
- Cmin for MR-formulations? Marcel 2010-04-23 10:09
- MR-Guideline? Helmut 2010-04-23 12:34
- Cmin for MR-formulations? Marcel 2010-04-23 10:09
- Cmin for MR-formulations? Helmut 2010-03-08 18:16
- Final version published ElMaestro 2010-01-29 17:02
- Overview of comments published Helmut 2010-02-10 18:33
- Comments commented d_labes 2010-02-11 08:19
- Definition of Cmin by EMEA Ravi 2010-03-13 12:05
- The EMA’s Cmin & WinNonlin Helmut 2010-03-13 12:52
- Final version published Helmut 2010-01-29 14:29
Ing. Helmut Schütz