Bioequivalence and Bioavailability Forum

Dear shekarsunkoju!

❝ […] iam getting problem in calculating the probabilities, which are given in the formulas.

Not even with the formulas, but also the program's result:

Open 'Data22.PWO' in folder 'Examples', and start the BE Wizard
Set the 'Reference value' to 'Capsule', 'Next >'
Drag 'AUClast' to 'Dependent Variables', 'Next >'
Keep defaults in the 'Variance Structure' Window, 'Next >'
Keep defaults in the 'Bioequivalence Options' Window, 'Next >'
In the 'Bioequivalence General Options' Window:
Check the 'ASCII Output' box and change 'Degrees of Freedom' to 'Residual', 'Calculate'

ASCII-Output (shortened):
Final variance parameter estimates:
    Var(Sequence*Subject)    0.0979228
            Var(Residual)    0.0532774
          Intersubject CV    0.320745
          Intrasubject CV    0.233928

    Ratio(%Ref) =   104.6588

               Power of ANOVA for Confidence Level 90.00
    Power at 20% = 0.898818

But I get (for 104.6588%, 0.233928) values above WinNonlin’s:
0.9582836 (R-code)
0.95828  (StudySize2.01)
0.9582    (nQuery Advisor 5.0)
For a ratio of 100% I get 0.9906747 – and WinNonlin’s value for a ratio of 93.2701%.
So what the h… is WinNonlin actually calculating here?


BTW, a posteriori power does not exist; see:

Hoenig JM, Heisey DM. The Abuse of Power. The Pervasive Fallacy of Power Calculations for Data Analysis. Am Stat. 2001;55(1):19–24. doi:10.1198/000313001300339897. free resource.