Posting: # 19169
I am trying to spell out the underlying model of FDA's pop BE analysis used for in vitro comparison of signle actuator content and more. A link is here.
The players are:
The formulations (F, two levels)
The stages (S, three levels in FDA's example, but may be anything from one upwards)
The cans ("Cans in batch", C, at least ten levels)
The batches ("batch within formulation", B, at least three levels)
I was initially thinking
Y=F+S+C+B+ek where ek is N(0, MSWk), and where Y is the in vitro metric we are sampling, like SAC, and where we's consider C, possibly B and S random (?).
But since MSWk comes from sampling across stages, I am inclined to work without S, so
where C and possibly B are random. When there is only a single sampled stage I guess this would then be
b. Y=F+B+ek where ek is then N(0, MSBk)
What's your opinion on a. and b.?
How would you formulate the models yourself?
"(...) targeted cancer therapies will benefit fewer than 2 percent of the cancer patients they’re aimed at. That reality is often lost on consumers, who are being fed a steady diet of winning anecdotes about miracle cures." New York Times (ed.), June 9, 2018.