Bioequivalence and Bioavailability Forum • Imbalance or incomplete data, WinNonlin5.2.1

Imbalance or incomplete data, WinNonlin5.2.1 [🇷 for BE/BA]

posted by Helmut – Vienna, Austria, 2008-10-29 13:12 (6075 d 21:36 ago) – Posting: # 2601
Views: 34,159

Dear all,

let’s continue the example in WinNonlin 5.2.1 (which uses also linear mixed effects modeling). Identical results to SAS are obtained both for the complete data set, and for the data set with #24 missing in both periods.
If data of #24 are missing in period 2 only, I obtained:

Denominator df option : residual

Partial Tests of Model Effects

Hypothesis  Numer_DF  Denom_DF     F_stat   P_value

---------------------------------------------------

  Sequence         1        21  0.988858     0.3313

    Formul         1        21  0.383815     0.5422

    Period         1        21  0.0564614    0.8145

90% Confidence intervals

point est.  lower     upper

0.963755    0.869808  1.067849

Denominator df option : satterthwaite

Partial Tests of Model Effects

Hypothesis  Numer_DF  Denom_DF     F_stat   P_value

---------------------------------------------------

  Sequence         1      21.8  0.988858     0.3309

    Formul         1      21.0  0.383815     0.5422

    Period         1      21.1  0.0564614    0.8145

90% Confidence intervals

point est.  lower     upper

0.963755    0.869820  1.067835

The second table is in agreement with SAS’ Proc MIXED.

Cave: although Satterthwaite degrees of freedom are default in WinNonlin, it’s possible to change this behaviour in the program’s preferences. This doesn’t make a difference if complete subjects are missing, but should be avoided in this case.

@DLabes: I gave the PE/CI to 6 decimal places in order to see a difference (if any); Satterthwaite dfs were 21.760741 (Sequence), 21.049560 (Formul), and 21.059500 (Period); can you give us your results also?

—
Dif-tor heh smusma 🖖🏼 Довге життя Україна!
Helmut Schütz

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes

Complete thread:

bear v2.0.1 (upwards) cont'd yjlee168 2008-10-26 20:46 [🇷 for BE/BA]
- bear v2.0.1 (upwards) cont'd ElMaestro 2008-10-27 20:36
  - bear v2.0.1 (upwards) cont'd yjlee168 2008-10-28 08:59
    - bear v2.0.1 (upwards) cont'd ElMaestro 2008-10-28 09:49
      - Imbalance or incomplete data, SASophylistic view d_labes 2008-10-28 17:14
        
        Imbalance or incomplete data, SASophylistic view yjlee168 2008-10-29 07:36
        
        Incomplete cases d_labes 2008-10-29 09:24
        
        Incomplete cases yjlee168 2008-10-30 07:12
        
        Imbalance or incomplete data, SASophylistic view ElMaestro 2008-10-29 09:27
        
        Gold in discussion d_labes 2008-10-29 09:56
        
        Gold? Helmut 2008-10-29 13:03
        
        Imbalance or incomplete data, WinNonlin5.2.1Helmut 2008-10-29 12:12
        
        Incomplete data, SAS9.2 d_labes 2008-10-29 16:38
        
        Missing values, imputation; separate study ElMaestro 2008-10-30 08:57
        Incomplete data, SAS9.2, DDF option residual d_labes 2008-10-30 15:54
        
        Never belief in any software you haven't written yourself Helmut 2008-10-30 16:30
        
        Imbalance or incomplete data, SASophylistic view yjlee168 2008-11-07 05:12
        
        Imbalance or incomplete data, SASophylistic view d_labes 2008-11-11 08:35
        
        Imbalance or incomplete data, SASophylistic view yjlee168 2008-11-11 08:54
        
        Imbalance or incomplete data, SASophylistic view d_labes 2008-11-11 13:13
        
        Imbalance or incomplete data, SASophylistic view yjlee168 2008-11-11 13:35
        
        Imbalance or incomplete data, SASophylistic view d_labes 2008-11-11 14:02
        
        Notification about new posts Helmut 2008-11-11 12:24
        
        Notification about new posts d_labes 2008-11-11 13:24
        
        3 Prerequisites for biostatistics Helmut 2008-11-11 13:39
        
        3 Prerequisites for biostatistics d_labes 2008-11-11 14:09