Bioequivalence and Bioavailability Forum • Reparameterization

Reparameterization [🇷 for BE/BA]

posted by yjlee168 – Kaohsiung, Taiwan, 2009-05-25 16:17 (5867 d 20:52 ago) – Posting: # 3747
Views: 11,661

Dear D. Labes and ElMaestro,

Thank you for your reminding. I just checked mt SAS output as the follows.


                                The SAS System

                               The GLM Procedure

Dependent Variable: y

Tests of Hypotheses Using the Type III MS for subject(group) as an Error Term



Source      DF    Type III SS    Mean Square    F Value Pr > F

group       1    21.42857143    21.42857143    7.65    0.0395



                                         Standard 

Parameter                   Estimate       Error       t Value Pr > |t| 

Intercept               14.00000000 B   1.22474487      11.43   <.0001 

group          1         0.00000000 B   1.54919334       0.00   1.0000 

group          2         0.00000000 B    .                .      . 

subject(group) 1 1      -3.00000000 B   1.54919334      -1.94   0.1106 

subject(group) 2 1      -2.00000000 B   1.54919334      -1.29   0.2532 

subject(group) 3 1      -1.00000000 B   1.54919334      -0.65   0.5471 

subject(group) 4 1       0.00000000 B    .                .      . 

subject(group) 5 2       1.00000000 B   1.54919334       0.65   0.5471 

subject(group) 6 2       2.00000000 B   1.54919334       1.29   0.2532 

subject(group) 7 2       0.00000000 B    .                .      . 

period         1        -2.00000000 B   0.83666003      -2.39   0.0624 

period         2         0.00000000 B    .                .      . 

Tmt            1         1.00000000 B   0.83666003       1.20   0.2856 

Tmt            2         0.00000000 B    .                .      . 



NOTE: The X'X matrix has been found to be singular, and a generalized inverse was used tosolve the normal equations. Terms whose estimates are followed by the letter 'B'are not uniquely estimable.

And output from R:

Call:

lm(formula = lnY ~ Trt + Subj + Seq + Per)



Coefficients:

(Intercept)         Trt1        Subj1        Subj2        Subj3  

  1.400e+01    1.000e+00   -3.000e+00   -2.000e+00   -1.000e+00  

      Subj4        Subj5        Subj6         Seq1         Per1  

 -6.571e-17    1.000e+00    2.000e+00           NA   -2.000e+00

Yes, you're correct about this. The results are the same from R and SAS. However, it's not the overall mean for lnY. Sorry about this.

❝ ❝ We have tested R codes of your examples with SAS. And SAS comes out the

❝ ❝ correct answer with intercept equal to the mean. [...]

❝

❝ Therefore, Yung-jin, check your SAS results.

—
All the best,
-- Yung-jin Lee
bear v2.9.2:- created by Hsin-ya Lee & Yung-jin Lee
Kaohsiung, Taiwan https://www.pkpd168.com/bear
Download link (updated) -> here

Complete thread:

Mean as intercept; model matrices ElMaestro 2009-05-20 20:35
- Mean as intercept; model matrices yjlee168 2009-05-21 13:35
  - Mean as intercept; model matrices ElMaestro 2009-05-21 14:30
- Mean as intercept; model matrices ElMaestro 2009-05-22 14:39
  - Mean as intercept; model matrices yjlee168 2009-05-23 20:06
  - Mean as intercept; model matrices Aceto81 2009-05-26 10:28
- Mean as intercept; model matrices yjlee168 2009-05-23 19:30
  - Mean as intercept; model matrices ElMaestro 2009-05-23 19:53
- Reparameterization d_labes 2009-05-25 11:47
  - Reparameterizationyjlee168 2009-05-25 14:17
  - Reparameterization ElMaestro 2009-05-25 20:29
    - Reparameterized brain d_labes 2009-05-26 08:01