Ill-defined chance that more than half of scientific papers are wrong

[nancylebov]

http://www.newscientist.com/article.ns?id=dn7915

Comments

[zenlizard]

"We should accept that most research findings will be refuted. Some will be replicated and validated. The replication process is more important than the first discovery," Ioannidis says.

Tempest in a teapot. This is why the reports of new findings have always been accepted with skepticism by the scientific community, and why there are such things as peer-review, and critique by the comminuty as a whole.

Only thing that may be truly notable about this is if the whacko, mind-control creation "scientists" use this as an excuse to further thier political agendas.

[madfilkentist]

This is the first time I've seen a mention of the law of large numbers as a way of creating false statistical significance. If 100,000 studies are done in a year, it's likely that just by chance one will produce a result that has a one in 100,000 probability of occurring due to chance. If it says something alarming, the probability is close to 1.0 that the news media will pick it up.

Even scientifically-minded people are going to be more skeptical about results which disagree with their beliefs than about those which agree with them, so this is a useful cautionary note for everyone, not just for alleged loonies.

[zenlizard]

>Even scientifically-minded people are going to be more skeptical about results which disagree with their beliefs than about those which agree with them


Yep, this is one of the most-often critiques levelled during peer-review.

[louiseroho.livejournal.com]

95% of all statistics are crap.

[zenlizard]

95% of all statistics are meaningless. The other 10% are erroneous.