Bioequivalence and Bioavailability Forum

Hi Ahmed,

❝ can someone please refer me to statistical analysis of a 3 period 6 sequence William's design BE study, 2 test products, one reference. I use WinNonlin, do not use SAS

See this article.

Say, your products are coded T1, T2, and R. In Phoenix WinNonlin set up two Data Wizards / Built In Filters. In the first exclude T2 and in the second T1. By this you get two Incomplete Block Designs. Perform analyses separately on each of these data sets. Whether or not you have to adjust \(\small{\alpha}\) (the level of the test), i.e., may use the conventional 90% or have to use a 95% confidence interval, depends on the purpose of the study (see also there).