willbryant.net

Survivor bias and simple probability

Published Sun 23 August 2009 21:50 (+1200)
Tagged
    (no tags)

This great post on the survivor bias and related effects like the ‘file-drawer effect’ is well worth reading. Most pharmaceutical, management and business success studies are not only based on selective data, but there isn't any attempt made to count the failures too, so most of the conclusions reached aren't actually supported by any kind of meaningful data; there's no reason to believe the stated cases aren't just the ones that happened to succeed rather than the ones that succeeded because of the author's theory. Great examples in the post. via