Dear Michael,
There is a mathematical critique of both the papers by a Czek physicist Lubos Motl:
http://motls.blogspot.co.uk/2013/01/medical-literature-do-wrong-results.html
Regards,
Ash
------------------------------
On Fri, Feb 8, 2013 14:29 GMT Michael Power wrote:
>Mohammad, thanks for the pointer to a most interesting paper
>
>
>
>It finds that the number of false positive statistical tests in abstracts is
>about 14% - the expected number would be 5% (as P<5% is used as the
>threshold for statistical significance).
>
>
>
>What would be really useful is a study of the level of pollution in the
>research literature. The numerator would be the number of papers with
>informative results; the denominator would be the number of papers with
>informative results + the number of reports of studies that should not have
>been done, e.g. because they used a patient unimportant outcome, or the
>comparator was a placebo instead of current best practice, or .
>
>
>
>Another interesting study would be to look at all Cochrane systematic
>reviews with more than x studies, and compare the results of the first study
>with the overall summary result. This would give us an indication of bias in
>first publications
>
>
>
>michael
>
>
>
>
>
>
>
>From: Evidence based health (EBH)
>[mailto:[log in to unmask]] On Behalf Of Mohammad Zakaria
>Pezeshki
>Sent: 28 January 2013 18:14
>To: [log in to unmask]
>Subject: The claim that most biomedical research is wrong is being
>challenged by a new result suggesting that only 14 per cent is wrong
>
>
>
>http://arxiv.org/abs/1301.3718
>
>http://www.technologyreview.in/blog/guest/28927/
>
>http://simplystatistics.org/2013/01/24/why-i-disagree-with-andrew-gelmans-cr
>itique-of-my-paper-about-the-rate-of-false-discoveries-in-the-medical-litera
>ture/
>
>Mohammad Zakaria Pezeshki, M.D.
>Associate Professor
>Program for Estimation of Pretest Probability
>Department of Community Medicine,
>Tabriz Medical School, Golgasht Avenue, Tabriz, Iran,
>Tel: 0098 411 336 46 73
>Fax: 0098 411 336 46 68
>
|