Williams design 3-way [Design Issues]

posted by Brus – Spain, 2021-06-11 12:19 (473 d 03:20 ago) – Posting: # 22408
Views: 5,283

Dear Vezz and Helmut,

I think I understood Helmut. With my design (3x4x3), the carryover effect would be covered, since it generates balanced sequences for two-to-time comparisons, and any formulation is preceded equally often by each of the other formulations, but it does not cover the period effect, since that not all treatments are the same number of times in all periods (for example, in my design treatment "A" is administered 1 time in period 1, twice in period 2 and 1 time in period 3. In this way, in the event that there was a period effect, it would not be offset by the design and it would affect the results. Right?

As Vezz says, if you consider period effect in your model and the software itself already incorporates that correction in case there is period effect, you could have unbiased results, in fact, considering the period effect and the rest of possible effects (sequence, treatment,...) in the model, any design would be acceptable, right ?

But at this point, a question comes to mind, if our software is capable of calculating the period, sequence, group, and treatment effect, what need is there to make balanced designs?

In this way IMHO, and if I'm not mistaken, after kindly discussing it, I think the ideal option for a 3 treatment crossover study is 3x6x3 williams design.

Best regards,

Complete thread:

UA Flag
 Admin contact
22,391 posts in 4,685 threads, 1,594 registered users;
online 14 (1 registered, 13 guests [including 3 identified bots]).
Forum time: Tuesday 15:39 CEST (Europe/Vienna)

Statistics. A sort of elementary form of mathematics which consists of
adding things together and occasionally squaring them.    Stephen Senn

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz