Skip to content
Prev 35160 / 398502 Next

inverse prediction and Poisson regression

Hello again, sorry for the notation. Again, I'm just a biologist!!!

;-)

But I'm enjoying this problem quite a bit! I'm very grateful for all the input. This is great.
On 2003-07-25 08:38:00 -0400 Prof Brian Ripley <ripley at stats.ox.ac.uk> wrote:
Answers:
I will clean up my notation!
OK, I want x for mu(x) = mu(0)/2.
Correct, This is part of the problem! The model does not "hold" for X = 0.
Yes, you are right, but then there is no growth, nad no LD50 value, so we reject this sample...
Then you suggest using another model that will account for zero dose, OK. I think I saw something similar in another reply. I need to read it more carefully.