Loky do
★

Egypt,
2020-07-19 01:33
(673 d 08:04 ago)

Posting: # 21746
Views: 1,775

## Sequence assignments in partially replicate design [Design Issues]

Dears

In partially replicate 3-way design, FDA recommends using the following example for sequence assignments 1,2,3 --> (TRR, RTR, RRT), if we used different sequence assignments 1,2,3 --> (TRR, RRT, RTR), is this change significant? or is it ok as long as it's considered in statistical analysis?

Helmut
★★★

Vienna, Austria,
2020-07-19 11:17
(672 d 22:21 ago)

@ Loky do
Posting: # 21748
Views: 1,494

## Sequence codes irrelevant

No Loky do,

it is not relevant how you code sequences and in which order. You could call them trousers.shirt.shirt, shirt.trousers.shirt, shirt.shirt.trousers as well. You have only to adapt the SAS-code at the end of the guidance’s page 4 accordingly.

Why on earth do you want a partial replicate design?

Dif-tor heh smusma 🖖
Helmut Schütz

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

Egypt,
2020-07-20 00:10
(672 d 09:28 ago)

@ Helmut
Posting: # 21752
Views: 1,454

## Sequence codes irrelevant

» Why on earth do you want a partial replicate design?

I think some countries like ours are very traditional in following guidelines, but excuse my asking  why using partially replicate design is not favorable nowadays?

Thanks Helmut
Helmut
★★★

Vienna, Austria,
2020-07-20 10:04
(671 d 23:33 ago)

@ Loky do
Posting: # 21757
Views: 1,452

## 3-period full replicates

Hi Loky do,

» I think some countries like ours are very traditional in following guidelines,…

Agree. But the Egyptian GL of 2017 in Section 3.2 recommends the EMA’s approach and not the FDA’s. That’s not a different league but a different sport. For the SAS-codes recommended by the EMA see the Q&A document. Alternatively* you can use Phoenix/WinNonlin, STATISTICA, SPSS, Stata, JPM, or the -pack­age replicateBE.

» why using partially replicate design is not favorable nowadays?

As a starter see this post and that one. I suggest also to search the forum with the keywords lousy and design. Most important reasons:
• FDA model
If swR <0.294 you are not allowed to employ reference-scaling. The model for ABE might not converge in any software. Study done, no result. Bad luck.
• All models
A three period full replicate design (TRT|RTR or TRR|RTT) is more informative than partial replicates because you can estimate the within-subject variance of T as well.

Dif-tor heh smusma 🖖
Helmut Schütz

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

Egypt,
2020-07-22 10:17
(669 d 23:21 ago)

@ Helmut
Posting: # 21772
Views: 1,350

## 3-period full replicates

Thanks a lot, Helmut