One of the requirements during method validation is to test parallelism between the standar curve and the sample curve when they are not the same product. I'm trying to find a way to measure this parallelism on Minitab and to get a proper test that provides a single p-value that states whether two curves are parallel or not.
Are the curves linear ?
If they are linear and parallel, the difference between their equations should only be a b0 value, i don't see how you can get a p-value on it.
Maybe if you perform the experiment a few times, you might get different results with the equation substraction and run a two sided test to determine whether thoose difference are the same or not ?
Thank you.
The curves are linear. I read that there is a way in the regression method in Minitab. If you cross the factors in which column are the two curves you want to test parallelism on (asuming both have the same values on the X axis) then the p-value test if both curves are parallel.
I'm doing some more research on it but that's what I've found out.
It sounds like you are trying to compare two slopes (m1 and m2), each of which has an error value (e1, e2) associated with it. This is an extension of Jean-Valery Guillaubez 's answer which assumed m1 and m2 were accurately quantified.
I don't have a copy of Zar's Biostatistical Analysis with me today, but that was a (relatively) simple test as far as I remember. Most stats textbooks would cover it in some form or another. It might even be as simple as a t-test.
How it is achieved with Minitab shouldn't take too long to work out either, but again, I don't have access to a copy right now.
The answer can be found at https://blog.minitab.com/en/adventures-in-statistics-2/how-to-compare-regression-lines-between-different-models
Te deseo éxitos Laura!
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling