Log in
RegisterPHX/WNL vs. SAS Proc Mixed [Regulatives / Guidelines]
Dear Helmut, dear All!
Here the results with real mixed model software:
(only the least square means part shown for the ln-transformed data)
Seems there is a perfect consistency of the results.
Remember (from our previous discussions about this topic) that SAS Proc GLM fits all effects as fixed effects. The random statement handles the named effects as random only in a "post-hoc manner", what ever this really means.
That's the reason why our captain calls this statement within Proc GLM bogus.
WNL on the other hand uses in his default evaluation obviously the real mixed model solution.
So far so good.
To be conform with the holy scripture (term invented by EM), page 15
"The terms to be used in the ANOVA model are usually sequence, subject within sequence, period and formulation. Fixed effects, rather than random effects, should be used for all terms."
the PHX/WNL user should leave the default and define all effects as fixed.
That answers Yicoating's question "Which is reliable?"
For the great oracle EMA: the SAS results using Proc GLM (random statement or not) or WNL results after setting all effects fixed
.
From a scientific point of view there are good reasons to use the real mixed effects solution.
Fortunately the outcome we are interested in at the end - the difference between formulations and their 90% CI - doesn't depend on the choice
. But only if we talk simple 2x2 crossover without missing values. See here.
Here the results with real mixed model software:
Proc mixed data=dose_equivalence;
class subject sequence period formulation;
model logAUC= formulation period sequence;
random subject(sequence);
lsmeans formulation / diff cl alpha=0.1;
run;(only the least square means part shown for the ln-transformed data)
The Mixed Procedure
Least Squares Means
Standard
Effect formulation Estimate Error DF t Value Pr > |t| Alpha
formulation 1 5.4356 0.05634 14 96.48 <.0001 0.1
formulation 2 5.4266 0.05634 14 96.32 <.0001 0.1
The Mixed Procedure
Least Squares Means
Effect formulation Lower Upper
formulation 1 5.3364 5.5349
formulation 2 5.3273 5.5258
Differences of Least Squares Means
Standard
Effect formulation _formulation Estimate Error DF t Value
formulation 1 2 0.009072 0.05134 14 0.18
Differences of Least Squares Means
Effect formulation _formulation Pr > |t| Alpha Lower Upper
formulation 1 2 0.8623 0.1 -0.08135 0.0995
Seems there is a perfect consistency of the results.
Remember (from our previous discussions about this topic) that SAS Proc GLM fits all effects as fixed effects. The random statement handles the named effects as random only in a "post-hoc manner", what ever this really means.
That's the reason why our captain calls this statement within Proc GLM bogus.
WNL on the other hand uses in his default evaluation obviously the real mixed model solution.
So far so good.
To be conform with the holy scripture (term invented by EM), page 15
"The terms to be used in the ANOVA model are usually sequence, subject within sequence, period and formulation. Fixed effects, rather than random effects, should be used for all terms."
the PHX/WNL user should leave the default and define all effects as fixed.
That answers Yicoating's question "Which is reliable?"
For the great oracle EMA: the SAS results using Proc GLM (random statement or not) or WNL results after setting all effects fixed
.From a scientific point of view there are good reasons to use the real mixed effects solution.
Fortunately the outcome we are interested in at the end - the difference between formulations and their 90% CI - doesn't depend on the choice
. But only if we talk simple 2x2 crossover without missing values. See here.—
Regards,
Detlew
Regards,
Detlew
Complete thread:
- Least Square Means (LSM) for unequal sequence yicaoting 2011-10-03 14:37 [Regulatives / Guidelines]
![[Mix]](https://static.bebac.at/img/mix.png "open in Mix view")
- Least Square Means (LSM) for unequal sequence ElMaestro 2011-10-03 14:51
- Least Square Means (LSM) for unequal sequence yicaoting 2011-10-03 15:28
- Least Square Means (LSM) for unequal sequence ElMaestro 2011-10-03 15:36
- Least Square Means (LSM) for unequal sequence Helmut 2011-10-03 16:14
- Least Square Means (LSM) for unequal sequence ElMaestro 2011-10-03 20:10
- Least Square Means (LSM) for unequal sequence Pankaj Bhangale 2011-10-04 14:52
- Weighted Means? Helmut 2011-10-05 01:53
- Least Square Means (LSM) for unequal sequence yicaoting 2011-10-03 15:28
- LSM for unbalanced sequences -SAS- d_labes 2011-10-04 09:39
- LSM for unbalanced sequences -SAS- yicaoting 2011-10-04 19:37
- SE for unbalanced sequences SAS vs. WinNonlin Helmut 2011-10-05 01:01
- LS Means mean? d_labes 2011-10-05 12:45
- LS Means mean? ElMaestro 2011-10-05 13:23
- PHX/WNL vs. SAS Helmut 2011-10-08 02:21
- PHX/WNL vs. SAS Proc Mixedd_labes 2011-10-10 15:46
- PHX/WNL vs. SAS Proc Mixed Helmut 2011-10-10 16:30
- PHX/WNL vs. SAS Proc Mixed ElMaestro 2011-10-10 19:53
- PHX/WNL vs. SAS Proc Mixed Helmut 2011-10-10 16:30
- Real mixed effects in WNL, SAS, GLM or Mixed (1) yicaoting 2011-10-13 18:39
- Real mixed effects in WNL, SAS, GLM or Mixed (2) yicaoting 2011-10-13 18:52
- Real mixed effects in WNL, SAS, GLM or Mixed (3) yicaoting 2011-10-13 19:05
- Must admit I am lost ElMaestro 2011-10-14 00:02
- Must admit I am lost yicaoting 2011-10-14 06:30
- Must admit I am lost - edited ElMaestro 2011-10-14 13:23
- Must admit I am lost yicaoting 2011-10-14 06:30
- Must admit I am lost ElMaestro 2011-10-14 00:02
- PHX/WNL vs. SAS Proc Mixedd_labes 2011-10-10 15:46
- LSM for unbalanced sequences -SAS- yicaoting 2011-10-04 19:37
- Least Square Means (LSM) for unequal sequence ElMaestro 2011-10-03 14:51
Ing. Helmut Schütz