1222 Multiplicities and False-Positive Rates in Science: Overview

Saturday, February 20, 2010: 3:50 PM
Room 11A (San Diego Convention Center)
Juliet P. Shaffer , University of California, Berkeley, CA
Although there are a number of explanations for the failure of apparent scientific findings to replicate, one of the major ones is undoubtedly the ailure to allow for the effects of multiplicity on the probability of false results.  Although statistical methods for controlling error rates in ultiple testing have been available for some time, they have been gnored in many scientific areas.  When the number of hypotheses is relatively mall, or the tests relate to different aspects of the study,  it is easy,
although unwise, to ignore the problem. ith the increase in the simultaneous testing of massive
numbers of hypotheses in recent years, as in, for example, genomics, europhysiology, and astronomy, the problems have become impossible to ignore. This area of statistics has become more prominent, and many new approaches o error control are being introduced. The talk will give a brief historical background and a broad overview of the current situation in this aea.