In Volume 12, Issue 2 of SIGKDD, Giraud-Carrier and Dunham emphasized on the importance of sharing negative results. I think that this goes well in hand with Poppers falsifiability criterion for scientific discovery. Do you think that showing positive results is a phenomena in hands of science? If so, how can we change it?
I think most editors accept positive reports for publication and tend to reject negative ones, particularly if the negative result fails to replicate the work of reputable scientists. There may be pockets of science, but scientific practices of dissemination of negative outcomes may just cut across many of them. At least the Association for Psychological Science has started to address this in a significant way (see a variety of such efforts here: http://www.psychologicalscience.org/?s=replication+failures
Any result ( either positive or negative) is a contribution. I agree that people like positivist approach then merely negative results. But, science like poetry is not disabled and there are many pockets of science. Sharing scientific result (truth) is more important than anything.
I think most editors accept positive reports for publication and tend to reject negative ones, particularly if the negative result fails to replicate the work of reputable scientists. There may be pockets of science, but scientific practices of dissemination of negative outcomes may just cut across many of them. At least the Association for Psychological Science has started to address this in a significant way (see a variety of such efforts here: http://www.psychologicalscience.org/?s=replication+failures
Publication bias is the bias resulting from where published results differ systematically from unpublished results. There are quite a few reasons that contribute to this and the term dissemination bias covers a larger range of causes.
The most well studied is due to the publication of significant results versus the studies supporting the null hypothesis. This is a well known phenomenon in the medical sciences, well studied in epidemiological methods and many efforts have been put in place to try to reduce this problem which results in incorrect clinical effects and associations eg new drugs often show a higher efficacy in trials prior to release than following release to market (although there are other issues that contribute to this difference).
In the investigation of publication bias in clinical trials, funnel plots can be used to show that the distribution of published results from a series of related studies does not show a predictable normal distribution around a true mean, but that they are skewed towards positive results and a gap will exists which represents the studies which were not published due to small or non- significant results.
One way to try to reduce this is through registering of all clinical trials prior to execution to ensure that all results are recorded in a clinical trials registry, but apparently this hasn't gone a long way to improving the problem.
For a nice open source paper reviewing this subject see:
Publication bias: what is it? How do we measure it? How do we avoid it?
Source: Open Access Journal of Clinical Trials . 2013, Vol. 5, p71-80. 10p.
Author(s): Fujian Song; Hooper, Lee; Yoon K. Loke
http://www.dovepress.com/publication-bias-what-is-it-how-do-we-measure-it-how-do-we-avoid-it-peer-reviewed-article-OAJCT
Scientists need to move fast and break things!
I define science in real time as always sharing what you can now. Results, methods, questions, failures and everything in between should be published immediately with unique identifiers and time stamps to make clear who discovered what and when, alleviating the fear of being scooped. This would provide a window into scientists’ work as it unfolds and open new opportunities for collaboration, because researchers could see what peers are currently working on...
This is what helps developers develop products faster, too. In most tech companies I know, everyone can see what everyone else is working on all the time. Engineers and other teams collaborate and benefit from one another’s unique skill sets and perspectives. They release their results and iterate on alpha and beta versions together with their users. There’s a sense of urgency to launch new features quickly because tech companies depend on their products being used.
It’s puzzling that whereas the steps scientists and techies take are so similar, scientists wait so long to share results. In science we often only see results after they have been published, following years of tests, negative and positive results, peer review and scrupulous editing...
https://www.researchgate.net/blog/post/scientists-need-to-move-fast-and-break-things
- Badges
- Science topic
- Similar topics
- Philosophy
- Philosophy Of Science