Friday, July 8, 2011

Pitfalls in Scientific Research

I'm currently reading "The Art and Science of Low Carbohydrate Living" by Drs. Phinney and Volek. Here's Dr. Steve Parker's review. As one who dabbles in the "art and science" of friction materials, I greatly appreciate the following quote from the book:

"...None of them alone can fully explain the phenomenon of the body ignoring insulin's signal. Part of this problem lies with how we do science. This is the reductionist approach to discovering scientific truth. This is straight forward and relatively easy to do, so lots of scientists swear by it. But what if a clinical problem like insulin resistance is not due to a single domino, but rather a number of dysfunctional proteins or other structural materials in combination? The answer - the reductionist approach can't deliver an answer in this situation. If multiple steps in a pathway, working in varying combinations, eventually compromise that pathway's action, the reductionist paradigm fails. But if one takes a more holistic or cosmopolitan approach to assessing the problem, the cause of the problem might be better appreciated."

Simply said, studying a system one factor at a time doesn't necessarily tell the whole story. To really get the big picture, you have to do several trials where several factors are changed at one time, in a designed way, and analysis the average effects of those changes. This method, referred to as design of experiments (DOE), allows for determining the effects of several factors in a set of trials and the interactions between them.

No comments:

Post a Comment