Let's assume the drug works. Is there anything in the underlying definitions, assumptions or clinical trial design that will hide that fact?
An allergy to imprecision helps in this business.
Be skeptical of treatments for phenotypic or symptoms-based diagnoses (irritable bowel syndrome, chronic fatigue, Alzheimer's) that likely lump several underlying pathologies together, each of which may react differently to a given treatment and make the treatment and control groups undetectably different.
Distrust imprecise endpoints. Overall survival is good: dead or not dead is pretty definitive. Composite, observer-dependent composite scores are not.
"Significant" p values, based on incorrect assumptions of bell-curved data distribution, are evidence of incompetence -- or are deliberately inaccurate -- until proven otherwise.
Answers to these questions disqualify a lot of stocks very quickly. More on this in future posts.
Every September -- well some of them anyway-- after the kids are back in school, after the plane and hotel reservations for the planned trips are credit card reserved and the decision not to bother with frequent-flyer and hotel chain loyalty programs has been reconfirmed...
I start a blog page.
Buried in the web dust are Wordpress, Tumblr, Blogger, Blog.com, and Weebly v1 pages, none bookmarked or favorited or part of anyone's morning must-read list.
Hopefully this will be more than the internet version of the old television test patterns that those of us who remember rabbit-ear antennas remember.