Monday, March 29, 2010

B - Challenging conventions on sample size

Bacchetti P. Current sample size conventions: Flaws, harms, and alternatives. BMC Medicine 2010;8:17.
http://www.biomedcentral.com/1741-7015/8/17/abstract

The widespread belief that medical research studies need a statistical power of at least 80% to be scientifically sound is a seriously flawed requirement. Standard calculations are unreliable, and move focus away from studies' more important results: estimates and confidence intervals. Current conventions may harm the research process in many way, including promoting misinterpretation, giving reviewers inappropriate powers, and inhibiting innovation. Medical research would benefit from alternative approaches. Peer reviewers should consider whether or not to raise issues of "inadequate" sample size, and reports of completed studies should not discuss power.

No comments: