1 reply to this topic

[Veronika Trushko]

Hello! Dear all, I need some help with replicate design.

Is it common practice to iteratively adjust α (Type I Error)?

is there an information about this in the guidelines?

And can I adjust α using Phoenix winnonlin?
 

 

[Helmut Schütz]

Hi Veronika,

Is it common practice to iteratively adjust α (Type I Error)?

No. In the meantime a few companies do it regularly. If your estimated CV_wR is outside the “critical range” of ~22–43% you don’t have to worry about an inflated TIE anyhow.

 

is there an information about this in the guidelines?

No.

 

And can I adjust α using Phoenix winnonlin?

Theoretically yes. Though I never tried it and would not recommend it. I guess it would be as least as time-consuming than in SAS – which is easily >1000times slower than by means of the R-package PowerTOST, function scABEL.ad().