I am so pleased today to be able to point to this publication detailing our use of a versatile differential equation comprising a normal-distribution-modified line for modeling certain analytical data of interest to pharmaceutical development. My sincere thanks to my co-workers Susannah Gaye and Morgan Crawford, professor Scoville of Ursinus College, and lastly to Dennis Peters (1937-2020) to whom this work is dedicated.
https://lnkd.in/e9SVydg3
If you've read this post this far, may I ask you to consider the following?
Have you ever collected or analyzed any measurement data which were ostensibly linear but sometimes had one or more data points (often at one extreme or another of a given range) which seemed to deviate variably and uncomfortably far away from a linear trend?
Be honest--have you ever excluded one or more of those data points so you could do a simple linear regression through only the best looking values, even if a sub-set of the data really does cover the range you need?
Certainly expediency has its place. After all, it is undeniably true that "the perfect is the enemy of the good". Still, there are times when a deeper examination of raw data is warranted. If and when such a time arrives for you, please look through this paper and remember that I, like Dr. Peters before me, would be delighted if you reached out.
Be well,
Lee