Bioequivalence and Bioavailability Forum • Satterthwaite’s df in Phoenix/WinNonlin

Satterthwaite’s df in Phoenix/WinNonlin [Software]

posted by ElMaestro – Denmark, 2013-07-05 19:08 (4386 d 11:38 ago) – Posting: # 10946
Views: 12,167

Hi Helmut,

❝ If I trust Wikipedia

❝ (s²₁+s²₂)²/[s⁴₁/(n₁–1)+s⁴₂/(n₂–1)] = 11.64240178 (≠ 10.75546079) :confused:

10.76 appears correct.
Quickie:

WelchDF=function(V1, V2, N1, N2)

{

 Num = ((V1/N1)+(V2/N2))^2

 Denom = (V1/N1)^2/(N1-1) + (V2/N2)^2/(N2-1) 

 return(Num/Denom)

}

But I have to admit I am still one big question mark as to why all this works correctly. All the matrix explanation and stuff is rather gibberish to me. I looked up what SAS writes about DDFM=SATT and I felt like hitting myself with a hammer after reading two lines of it.

—
Pass or fail!
ElMaestro

Complete thread:

Parallel design [unequal variances] in Phoenix/WinNonlin Helmut 2013-07-04 20:22
- Parallel design [unequal variances] in Phoenix/WinNonlin ElMaestro 2013-07-04 21:16
  - Workaround Helmut 2013-07-04 22:13
    - 3-trt study? ElMaestro 2013-07-04 22:37
      - ؟ῳ ♠♣♥♦♪♫ѳ٭ Helmut 2013-07-04 23:08
  - Satterthwaite’s df in Phoenix/WinNonlin Helmut 2013-07-05 15:13
    - Satterthwaite’s df in Phoenix/WinNonlinElMaestro 2013-07-05 17:08
      - Wikipedia, the free encyclopedia that anyone can edit. Helmut 2013-07-05 17:15
        
        Wikipedia, the free encyclopedia that anyone can edit. ElMaestro 2013-07-05 18:25
        
        Improbable Helmut 2013-07-05 18:40