No – futile! – test for G×T, pleeze… [Study Performance]
❝ If this is a dossier for EU:
❝ 1. If you don't add a group effect, the regulator is likely to ask for it after submission.
Hundreds of studies accepted in the past without one. Seems to become increasingly fashionable recently.
❝ 2. If you do add a group effect, regulators are likely to ask you to submit a secondary analysis without it.
Xactly. See there.
❝ At the end of the day Group is a between-factor, like sequence. Thus, the discussion is quite academic if we look at the confidence interval.
❝ Another matter of course is if we, for some reason or other, take an interest in the p-value of the group effect itself or group x treatment. The latter sends shivers down my spine.
Not even the highest dose of Schützomycin would help.
BTW, last month I endured a ‘Type A Meeting’ for an IND with the FDA.
I proposed to use ‘Model II’ without the stupid pre-test for a G × T interaction in ‘Model I’ because it inflates the Type I Error. Since writing a paper is on my todo-list for more than five years ( Nastia; I’m still collecting data), I was referring to my presentation at BioBridges in Prague 2018 (see esp. the backup-slides). Was accepted.
Dif-tor heh smusma 🖖🏼 Довге життя Україна!
The quality of responses received is directly proportional to the quality of the question asked. 🚮
- Additional Group in Bioequivalence Study PVRC 2021-04-23 10:35