>This is a rather interesting read that showed up in my inbox this morning:
>http://www.newyorker.com/reporting/2010/12/13/101213fa_fact_lehrer?currentPage=all
>...talking about the "decline effect" (not a short read, but worthwhile).
I have to say that this article reveals and explains everything I have been seeing in studies I read and in popular science articles and saying over and over
Specifically, it seems that unconscious bias and social pressures have a huge influence on what we believe and the proofs offered. Selection -- deliberate or unconscious -- takes place in almost every study, I dare to say, and also determines which studies get completed or which are dropped or morph into something else along the way.
Widespread self-deception and fuzzy thinking are reasons that a clear and well-discussed statement of purpose and proposed methods up front is so essential to integrity of any undertaking. We know that nothing is ever going to proceed exactly as planned, but it is essential to track where deviations occur.
It is easy to look back in time a few centuries or millennia and say, "What were they thinking? How could they possibly believe that?", but examples in our own time are invisible to most of us, and the people who can see through the shared delusions are not especially popular.
>this also makes me think of the recent discussions on doing bee research and controls. i'm no math or statistics whiz by any "mean" (sorry, couldn't resist), but to me it seems that the sledgehammer approach would be to run multiple control groups (ie, instead of 50% control and 50% experimental using 25% control 25% control 25% control 25% experimental).
I can't see how more than 50% controls could improve accuracy, and also, I think a lot depends on what is being examined. If there are several treatments being run consecutively, the size of each group, IMO should be the same and large enough that inevitable losses of hives do not reduce the sample size below the minimum to extract meaning from the noise.
Additionally, it is the very times when things do not work as planned or where losses occur, apparently for reasons other than the effects sought that the greatest opportunity to learn and be honest or to go wrong and filter data to suit preconceptions are found.
Excellent article.
***********************************************
The BEE-L mailing list is powered by L-Soft's renowned
LISTSERV(R) list management software. For more information, go to:
http://www.lsoft.com/LISTSERV-powered.html
Guidelines for posting to BEE-L can be found at:
http://honeybeeworld.com/bee-l/guidelines.htm
|
