Peculiar partial replicate [🇷 for BE/BA]
Dear Öberster Größter Meister,
Well noticed . We had discussed this peculiarity of the partial replicate design also here in the forum if you remember. See f.i. this post and followings. But be warned: Rather lengthy and nitpickin' .
As you can see there we get s2WT, but rather different values between different software. And, as I have noticed, you get also different values if you change optimization parameters in SAS Proc MIXED. Seems the REML optimisation stops by chance. Thus that value is not real, the model is over-specified, but the REML or ML estimation can't of course decide how the model has to be formulated without an explicit value for it.
It's the same as if we attempt to fit a 2x2 design with distinct intra-subject error terms for Test and Reference. We could not be successful because only (s2WT+s2WR)*0.5 is estimable.
Meanwhile I have the opinion that we can't handle this evaluation via such a detailed model of the different sources of variation. I at any rate was not successful to implement a reasonable simpler model in Proc MIXED.
Maybe this is one reason why the FDA code given in the progesterone guidance resorts to evaluation via intra-subject contrast?
BTW: Link corrected.
❝ With the intra-subject sWR and sWT only on the diagonal of V my alarm bells are ringing. Imagine a ref-replicated design; following your thinking we could fit an sWT on the diagonal of V even though Test isn't replicated. This sounds very wrong. Apparently others have got fits and sWT values for 2-trt, 3-seq, 3-per designs (see e.g. here). This is fishy business.
Well noticed . We had discussed this peculiarity of the partial replicate design also here in the forum if you remember. See f.i. this post and followings. But be warned: Rather lengthy and nitpickin' .
As you can see there we get s2WT, but rather different values between different software. And, as I have noticed, you get also different values if you change optimization parameters in SAS Proc MIXED. Seems the REML optimisation stops by chance. Thus that value is not real, the model is over-specified, but the REML or ML estimation can't of course decide how the model has to be formulated without an explicit value for it.
It's the same as if we attempt to fit a 2x2 design with distinct intra-subject error terms for Test and Reference. We could not be successful because only (s2WT+s2WR)*0.5 is estimable.
Meanwhile I have the opinion that we can't handle this evaluation via such a detailed model of the different sources of variation. I at any rate was not successful to implement a reasonable simpler model in Proc MIXED.
Maybe this is one reason why the FDA code given in the progesterone guidance resorts to evaluation via intra-subject contrast?
BTW: Link corrected.
—
Regards,
Detlew
Regards,
Detlew
Complete thread:
- Freedom for replicate bears d_labes 2009-04-17 12:06 [🇷 for BE/BA]
- Freedom for replicate bears ElMaestro 2009-04-17 14:50
- Contain does not contain d_labes 2009-04-17 15:07
- Freedom for replicate bears ElMaestro 2009-04-18 03:45
- Freedom incompatible with baseball bats d_labes 2009-04-20 08:42
- Freedom for replicate bears yjlee168 2009-04-19 21:38
- Freedom to some degree d_labes 2009-04-20 08:30
- Freedom to some degree yjlee168 2009-04-21 14:03
- FDA loves replicate bears? d_labes 2009-04-21 16:11
- Cholesky factor ?! Helmut 2009-04-22 00:46
- Cholesky factor ?! d_labes 2009-04-22 11:20
- FDA loves replicate bears? yjlee168 2009-04-23 08:08
- FDA loves SAS code? d_labes 2009-04-23 09:48
- FDA loves SAS code? definitely. yjlee168 2009-04-23 10:46
- SAS? or R? or what? d_labes 2009-04-23 15:07
- SAS? or R? or what? Helmut 2009-04-23 15:25
- SAS? or R? or what? d_labes 2009-04-23 16:06
- SAS? or R? or what? Helmut 2009-04-23 16:30
- SAS? or R? or what? d_labes 2009-04-23 16:06
- SAS? YES; R? why not? as long as they are useful... yjlee168 2009-04-23 19:56
- New Forum-Category 'bear'? Helmut 2009-04-23 20:03
- New Forum-Category 'bear'? yjlee168 2009-04-23 20:19
- New Forum-Category 'R for BE/BA' Helmut 2009-04-23 20:37
- New Forum-Category 'R for BE/BA' yjlee168 2009-04-24 09:19
- New Forum-Category 'R for BE/BA' Helmut 2009-04-23 20:37
- New Forum-Category 'bear'? yjlee168 2009-04-23 20:19
- Replicated again d_labes 2009-04-24 08:27
- Replicated again yjlee168 2009-04-24 09:26
- New Forum-Category 'bear'? Helmut 2009-04-23 20:03
- SAS? or R? or what? Helmut 2009-04-23 15:25
- SAS? or R? or what? d_labes 2009-04-23 15:07
- FDA loves SAS code? definitely. yjlee168 2009-04-23 10:46
- FDA loves SAS code? d_labes 2009-04-23 09:48
- Cholesky factor ?! Helmut 2009-04-22 00:46
- FDA loves replicate bears? d_labes 2009-04-21 16:11
- Freedom to some degree yjlee168 2009-04-21 14:03
- Freedom to some degree d_labes 2009-04-20 08:30
- getting decimals ElMaestro 2010-06-03 05:21
- getting decimals yjlee168 2010-06-03 08:37
- getting decimals ElMaestro 2010-06-04 01:59
- getting decimals yjlee168 2010-06-03 08:37
- Freedom for replicate bears ElMaestro 2011-01-27 21:50
- lmer vs. lme d_labes 2011-01-28 09:02
- lmer vs. lme ElMaestro 2011-01-28 09:42
- lmer vs. lme d_labes 2011-01-28 11:37
- V ElMaestro 2011-01-28 12:14
- V corrected d_labes 2011-01-28 14:12
- V corrected ElMaestro 2011-01-28 15:20
- Peculiar partial replicated_labes 2011-01-28 16:10
- V corrected ElMaestro 2011-01-28 15:20
- V corrected d_labes 2011-01-28 14:12
- V ElMaestro 2011-01-28 12:14
- lmer vs. lme d_labes 2011-01-28 11:37
- lmer vs. lme ElMaestro 2011-01-28 09:42
- lmer vs. lme d_labes 2011-01-28 09:02
- Freedom for replicate bears ElMaestro 2009-04-17 14:50