CV limbo [Regulatives / Guidelines]
this question boils down to "something" which is not so easy and straightforward.
First of all, when going full mixed there is to the best of my knowledge no particular relevance of a CI2CV kind of function. It provides an answer to something which is not really asked, but which may be identical to the answer truly sought in case of balance (not even sure about this last part, need to read up on things).
Now for the remainder, and this is everything based on the linear model as applied to replicated BE data, please bear in mind:
1. When we evaluate a subject, the guideline says the subject only contributes if she/he provides at least one T and one R measurement.
2. When we estimate swr, there is no equivalent wording. Look for example at subject 24 of EMA dataset I. The subject contributes with only one ref measurement. But in the grand scheme of things such a subject does not contribute to the within-ref (try and fit dataset I for swr with and without that subject). The subject's data is estimated with no residual because there is only one measurement. So, as someone wrote in another thread in another galaxy, it is all in the df's. The estimate of swr (if we go the anova way as per EMA) has 71 dfs, and 69 for swt.
But plugging the MSEs and df's into the equation for pooled variances still does not get you where you want to be (a pooled MSE reflecting the overall anova MSE if I got you right). And that's because of the fixed effect estimates that are allowed vary between the two anova's. Think about it.
So, if you are looking for 'manual' and simple equations that allow you to swap back and forth between swr, swt and pooled variances, then I am afraid you are out of luck and you may be asking too much. You can easier get what you are looking for by working backwards from the matrix forms of the linear model.
Pass or fail!
- FDA's HVD SAS Code / Standard Error / CV j_kevin 2022-12-08 09:41 [Regulatives / Guidelines]
- FDA's HVD SAS Code / Standard Error / CV Helmut 2022-12-08 12:09