Bioequivalence and Bioavailability Forum

Dear Pingouinche!

❝ Indeed, there is probably the SAS-code on the book,

                   ^^^^^^
The probability is close to 1 (unless you buy a misprint or somebody has ripped these pages out).

❝ Unfortunately, I don't have the book and the code is not included in the extracts available on Google Books...

No? Would you mind looking at page 560? OK, the interesting ones are 561-562.

❝ Precisely, I'd like to know their code, because whatever I do, I don't manage to obtain the same ANOVA table as them !

In your first post you said you read the book. Fine. But f…g Google Books is not a source to gain knowlege from - at least nobody can ‘read a book’ there. The third edition was published this year. If you are with a company which can’t afford a book and give you enough time to study it, change the company. Quickly! If you are a poor fellow, maybe you can get the second edition cheaper or even a used copy. If you are a student, go to your library. Or get a copy of pages 559-576 (2nd ed, 691-714 3rd ed.) from a remote reference provider. It’s even possible to get their ANOVA-table up and running in M$-Excel (see the screenshot in this post). Or search this forum (codes for both SAS and are available).
Sorry for the harsh reply, but please do you homework first.

See also this post for something you may check.

Dear Helmut,

❝ Yes, but have you noticed that results for period and treatment are numerically identical?

You're right! I really did NOT notice that part. A typical anova output obtained from bear is shown as follows.
  Dependent Variable: lnAUC0t                                               

Type I SS
Analysis of Variance Table

Response: lnAUC0t
          Df  Sum Sq Mean Sq F value Pr(>F)
seq        1 0.01876 0.01876  0.5138 0.4872
prd        1 0.00789 0.00789  0.2162 0.6503
drug       1 0.04216 0.04216  1.1546 0.3037
seq:subj  12 0.29718 0.02477  0.6782 0.7443
Residuals 12 0.43818 0.03652               

Type III SS
Single term deletions

Model:
lnAUC0t ~ seq + subj:seq + prd + drug
         Df Sum of Sq     RSS     AIC F value  Pr(F)
<none>                  0.438 -84.405               
prd       1     0.008   0.446 -85.905  0.2162 0.6503
drug      1     0.042   0.480 -83.833  1.1546 0.3037
seq:subj 12     0.297   0.735 -93.909  0.6782 0.7443

Tests of Hypothesis for SUBJECT(SEQUENCE) as an error term

Error: subj
          Df   Sum Sq  Mean Sq F value Pr(>F)
prd:drug   1 0.018762 0.018762  0.7576 0.4012
Residuals 12 0.297182 0.024765               

Error: Within
          Df  Sum Sq Mean Sq F value Pr(>F)
prd        1 0.00789 0.00789  0.2162 0.6503
drug       1 0.04216 0.04216  1.1546 0.3037
Residuals 12 0.43818 0.03652               

Intra_subj. CV = 100*sqrt(abs(exp(MSResidual)-1)) = 19.2847 %
Inter_subj. CV = 100*sqrt(abs(exp((MSSubject(seq)-MSResidual)/2)-1)) = 7.6536 %
    MSResidual = 0.03651522
MSSubject(seq) = 0.02476518

❝ Maybe there’s a mistake in the coding of the dataset (levels?).

The period and drug (treatment) could be coded incorrectly in the dataset.