Subject: | |
From: | |
Reply To: | |
Date: | Tue, 12 Nov 2013 19:35:20 -0500 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
As one of the few who takes the time to sweat the statistics, I can say with
exceedingly high confience that this post is very probably correct. :)
There's been a spate of attempts to reproduce results that failed to
reproduce the results.
http://www.nature.com/news/replication-studies-bad-copy-1.10634
http://tinyurl.com/bs4l697
The current gold standard for certainly, a "P value" of 0.05 or less just
does not mean what it used to, as researchers have been using "frequentist"
statistical methods, which are "weak-kneed", there's also been some outlier
data dropped on the floor more than once:
http://www.nature.com/news/weak-statistical-standards-implicated-in-scientif
ic-irreproducibility-1.14131
http://tinyurl.com/p586v3m
When one uses "Bayesian" tests ("Bayesian" being the reason why you don't
see much spam any more) one finds that 17-25% of "frequenist" findings of "P
< .05" are probably misleadingly labeled "conclusive". This means even more
turgid statistical work!
www.pnas.org/content/early/2013/10/28/1313476110.full.pdf
http://tinyurl.com/ktwpn6t
"...recent concerns over the lack of reproducibility of scientific studies
can be attributed largely to the conduct of significance tests at
unjustifiably high levels of significance. To correct this problem, evidence
thresholds required for the declaration of a significant finding should be
increased to 25-50:1, and to 100-200:1 for the declaration of a highly
significant finding. In terms of classical hypothesis tests, these evidence
standards mandate the conduct of tests at the 0.005 or 0.001 level of
significance."
This is not the first voice in the wilderness on this issue. In 2011, this
very similar plea was generally ignored:
"The false-positive to false-negative ratio in epidemiologic studies."
Epidemiology. 2011 Jul 22
doi: http://dx.doi.org/10.1097/EDE.0b013e31821b506e
http://tinyurl.com/n3q3qc7
***********************************************
The BEE-L mailing list is powered by L-Soft's renowned
LISTSERV(R) list management software. For more information, go to:
http://www.lsoft.com/LISTSERV-powered.html
|
|
|