V corrected [R for BE/BA]

posted by d_labes  – Berlin, Germany, 2011-01-28 14:12 (3918 d 10:02 ago) – Posting: # 6508
Views: 62,670

(edited by d_labes on 2011-01-28 15:32)

Dear ElMaestro,

I think your V needs to be corrected to reflect the FDA model for replicate cross-over designs.
See Patterson et al
"REPLICATE DESIGNS AND AVERAGE, INDIVIDUAL, AND POPULATION BIOEQUIVALENCE"
GSK BDS Technical Report 2002 – 01
online resource

It has to be composed from these two parts:
(sequence TRTR, hope I got all these nasty subscripts right :cool:)
between-subject:


 vBT   covTR  vBT  covTR
covTR   vBR  covTR  vBR
 vBT   covTR  vBT  covTR
covTR   vBR  covTR  vBR


and within-subject diagonal matrix:

  vWT       
      vWR
          vWT
              vWR


I'm not so familiar with the so-called V matrix but I guess its the sum of both?*see PS
Thus you have 5 variance-covariance parameters to estimate:In your model/abbreviated output I only identify 4 covariance parameters:
vBT, vBT and covTR, latter here given as Corr(elation) and one residual error term which is presumable the mean of vWT and vWR or other spoken common within-subject error (assuming vWT = vWR=ve).

Have a look at the very beginning of this thread to see that FDA's Proc MIXED and my lme() attempt are giving the same covariance parameter values (except that Proc MIXED gives variances/covariances and lme() stddev's for the between part and an error term and factors for T/R for the within-subject part, i.e. you have to calculate f.i. sWR as error(residual) term multiplied by the factor for R.

*PS: Regarding the matratzes we had it already here.
[edit]----------------------------
link to technical report corrected

Regards,

Detlew

Complete thread:

Activity
 Admin contact
21,749 posts in 4,547 threads, 1,545 registered users;
online 11 (0 registered, 11 guests [including 2 identified bots]).
Forum time: Thursday 01:14 CEST (Europe/Vienna)

Imagine if every Thursday your shoes exploded
if you tied them the usual way.
This happens to us all the time with computers,
and nobody thinks of complaining.    Jef Raskin

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5