6 ms·
One of the many challenges in science is that there is no publication outlet for experiments that just didn't pan out. If you do an experiment and don't find st
by jimmar 13y ago
One of the many challenges in science is that there is no publication outlet for experiments that just didn't pan out. If you do an experiment and don't find statistical significance, there aren't many journals that want to publish your work. That alone helps contribute to a bias toward publishing results that might have been found by chance. If 20 independent researchers test the same hypothesis, and there is no real effect, 1 might find statistical significance. That 1 researcher will get published. The 19 just move on.
- essayist 13y agoThey are working on it. There was an effort at Oxford to track perinatal trials -- started in the 1980's. It looks like it hasn't happened yet, but that various major players (PLOS, Center for Evidence-based Medicine) want to expand the brief to cover all clinical trials: http://www.alltrials.net/about/ http://www.alltrials.net/about/ http://www.cochrane.org/about-us/history http://www.cochrane.org/about-us/history
- capnrefsmmat 13y agoFor many types of clinical trials, pre-registration and publication of results through ClinicalTrials.gov is required by the FDA. I think it's been five or ten years now. Unfortunately, compliance isn't great -- something like 80% of studies registered on ClinicalTrials.gov never have results published there. 20% of registered studies never have results published anywhere.
- RogerL 13y agoThis is called the "file drawer effect".
- r0muald 13y agoThere's Figshare <http://figshare.com/> http://figshare.com/> to publish negative data.