Bioequivalence and Bioavailability Forum

jag009 ★★★ NJ, 2013-05-03 18:45 (4448 d 19:43 ago) Posting: # 10537 Views: 6,760	Reverse Engineering → Back calculate the intra CV [Power / Sample Size] Post reply
	Hi Helmut! How does one back calculate the intrasubject CV (and MSE?) from the 90% confidence interval obtained for a three way partial replicate study? Can it be done using the same suggestion that you presented in your previous slides? Thanks John Edit: Started a new thread & changed category. [Helmut]

Helmut
★★★

Vienna, Austria,
2013-05-03 19:09
(4448 d 19:19 ago)

@ jag009
Posting: # 10538
Views: 5,567

Why not?

Post reply

Hi John!

❝ How does one back calculate the intrasubject CV (and MSE?) from the 90% confidence interval obtained for a three way partial replicate study? Can it be done using the same suggestion that you presented in your previous slides?

Essentially yes. You have to get the degrees of freedom right. See Detlew’s post. For your log AUCt data and my CI (ignoring the fact that you had 16, 17, and 18 subjects in the three sequences) I got:

require(PowerTOST) CVfromCI(lower=0.874898, upper=1.02586, n=51, design="2x3x3", alpha=0.05) [1] 0.2850511

…or better?*

require(PowerTOST) CVfromCI(lower=0.874898, upper=1.02586, n=51, design="2x3x3", alpha=0.05, robust=TRUE) [1] 0.2820788

But: According to this discussion I duno whether the CI is correct. :confused:

Maybe it’s better to go with the linearized CI from the RSABE-code instead (no mixed-effects limbo). With the 90% CI of 0.874723 – 1.02766 I got:

require(PowerTOST) CVfromCI(lower=0.874723, upper=1.02766, n=51, design="2x3x3", alpha=0.05) [1] 0.2886897

Documentation of PowerTOST/CVfromCI.

P.S.: Any news from Donald Schuirmann? ;-)

—
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

jag009 ★★★ NJ, 2013-05-09 22:12 (4442 d 16:16 ago) @ Helmut Posting: # 10563 Views: 5,524	Why not? Post reply
	Hi Helmut, ❝ Essentially yes. You have to get the degrees of freedom right. See Detlew’s post. Question (obvious answer I am sure). The degrees of freedom for a 3-period 3-sequence 2-treatment partial replicate should be similar to that of a 3-period 3-sequence 3-treatment crossover right? ❝ P.S.: Any news from Donald Schuirmann? Waiting... Thx John

d_labes
★★★

Berlin, Germany,
2013-05-10 10:12
(4442 d 04:16 ago)

@ jag009
Posting: # 10564
Views: 5,845

degrees of freedom

Post reply

Dear John,

❝ Question (obvious answer I am sure). The degrees of freedom for a 3-period 3-sequence 2-treatment partial replicate should be similar to that of a 3-period 3-sequence 3-treatment crossover right?

Define similar :-D

.

If you need the exact ones:
Have a look at known.designs():

    no   design    df df2 steps  bk bknif       bkni                name    

     0 parallel   n-2 n-2     2 4.0   1/1 1.00000000         2 parallel groups

...

4    2      3x3 2*n-4 n-3     3 2.0   2/9 0.22222222             3x3 crossover

5    3    3x6x3 2*n-4 n-6     6 2.0  1/18 0.05555556           3x6x3 crossover

...

10   9    2x3x3 2*n-3 n-3     3 1.5   1/6 0.16666667 partial replicate (2x3x3)

...

df are the usual degrees of freedom if you use Proc GLM or equivalent.
df2 are the so-called robust degrees of freedom obtained if you analyze via appropriate intra-subject contrasts for T-R averaged over sequence groups.

—
Regards,

Detlew

BEQool
★

2025-07-04 09:20
(4 d 05:08 ago)

@ d_labes
Posting: # 24406
Views: 172

degrees of freedom of partial replicate 2x3x3

Post reply

❝ Have a look at known.designs():

❝ no design df df2 steps bk bknif bkni name

❝ 0 parallel n-2 n-2 2 4.0 1/1 1.00000000 2 parallel groups

❝ ...

❝ 4 2 3x3 2*n-4 n-3 3 2.0 2/9 0.22222222 3x3 crossover

❝ 5 3 3x6x3 2*n-4 n-6 6 2.0 1/18 0.05555556 3x6x3 crossover

❝ ...

❝ 10 9 2x3x3 2*n-3 n-3 3 1.5 1/6 0.16666667 partial replicate (2x3x3)

❝ ...

❝

❝ df are the usual degrees of freedom if you use Proc GLM or equivalent.

Hello all,

firstly, it is good to have BEBAC forum up and running again :ok:

I have a question regarding df in partial replicate 2x3x3 design. 44 subjects were included in the PK/STAT analysis - 43 of them had data for all periods while 1 subject had data only for 2 periods (the missing period was with reference treatment B so the subject had data for both treatment A and treatment B). Both in SAS and Phoenix WNL the df of the design (Error) is 84. Why is this so? According to the formula 2*n-3 with 44 subjects included in the analysis, the df should be 2*44-3= 85. Or is this discrepancy (84 vs 85 df) a conseqeunce of one missing period for one subject (although the subject has data for both A and B and is included in the analysis)?

Best regards
BEQool

Reverse Engineering → Back calculate the intra CV [Power / Sample Size]

Why not?

Why not?

degrees of freedom

degrees of freedom of partial replicate 2x3x3