News for hackers. Stuff that matters.

A comment on How to make fake data look meaningful.

Re:How to make fake data look meaningful by tokenadult on Wednesday November 20, 2013 @03:33PM (#6770965)
Some more sophisticated (but very readable and sometimes laugh-out-loud funny) articles about detecting fake data can be found at the faculty website[1] of Uri Simonsohn, who is known as the "data detective"[2] for his ability to detect research fraud just from reading published, peer-reviewed papers and thinking about the reported statistics in the papers. Some of the techniques for cheating on statistics that he has discovered and named include "p-hacking,"[3] and he has published a checklist of procedures to follow to ensure more honest and replicable results.[4]

From Jelte Wicherts writing in Frontiers of Computational Neuroscience (an open-access journal) comes a set of general suggestions[5] on how to make the peer-review process in scientific publishing more reliable. Wicherts does a lot of research on this issue to try to reduce the number of dubious publications in his main discipline, the psychology of human intelligence.

"With the emergence of online publishing, opportunities to maximize transparency of scientific research have grown considerably. However, these possibilities are still only marginally used. We argue for the implementation of (1) peer-reviewed peer review, (2) transparent editorial hierarchies, and (3) online data publication. First, peer-reviewed peer review entails a community-wide review system in which reviews are published online and rated by peers. This ensures accountability of reviewers, thereby increasing academic quality of reviews. Second, reviewers who write many highly regarded reviews may move to higher editorial positions. Third, online publication of data ensures the possibility of independent verification of inferential claims in published papers. This counters statistical errors and overly positive reporting of statistical results. We illustrate the benefits of these strategies by discussing an example in which the classical publication system has gone awry, namely controversial IQ research. We argue that this case would have likely been avoided using more transparent publication practices. We argue that the proposed system leads to better reviews, meritocratic editorial hierarchies, and a higher degree of replicability of statistical analyses."

[1] http://opim.wharton.upenn.edu/~uws/

[2] http://www.nature.com/news/the-data-detective-1.10937

[3] http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2160588

(this abstract link leads to a free download of the article)

[4] http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1850704

(this abstract link also leads to a free download of the full paper)

[5] http://www.frontiersin.org/Computational_Neuroscience/10.338...

Jelte M. Wicherts, Rogier A. Kievit, Marjan Bakker and Denny Borsboom. Letting the daylight in: reviewing the reviewers and other ways to maximize transparency in science. Front. Comput. Neurosci., 03 April 2012 doi: 10.3389/fncom.2012.00020

[ Parent | on HN ]
Re:How to make fake data look meaningful by dwaltrip on Wednesday November 20, 2013 @07:14PM (#6772243)
Start-up idea: Review site that generates scores for published scientific papers using an open methodology. Maybe even allow community to generate their own metrics!

Edit: looks like the domains "sciencetomatoes.com" and "rottenpvalues.com" are available ;) http://www.internic.net/whois.html

[ Parent | on HN ]
Re:How to make fake data look meaningful by Houshalter on Wednesday November 20, 2013 @11:03PM (#6772983)
This is perhaps a ridiculous idea but I'll post it anyways. There are a lot of features of a paper that might correlate with it's truthfulness. For example surface level stuff like what journal published it or the authors, what subject it's on, how controversial/interesting it is. Then the data itself.

Then get a large sample of papers which are known to be true or false (or at least were able to be replicated or not) and use some machine learning or statistical method to make decent predictions how likely new papers are to be true or false.

[ Parent | on HN ]
Re:How to make fake data look meaningful by yen223 on Thursday November 21, 2013 @07:53AM (#6774428)
How are you going to prove that your system isn't pulling random ratings out of thin air? ;)
[ Parent | on HN ]