Tuesday, August 07, 2007

Statistics and subjectivity in science

There seems to be a general misunderstanding about statistical significance versus real significance, particularly when it comes to health related issues. In this paper Facts versus Factions: the use and abuse of subjectivity in scientific research, Robert Matthews discusses the problem from an historical and statistical perspective.

Of particular interest is the discussion of Baysian analysis, which implicitly recognizes scientific subjectivity in its characterization of the probability that a particular hypothesis is correct.

A more recent discussion of this phenomena is found in Why Most Published Research Findings Are False by John P. A. Ioannidis.

No comments: