Skimmed around the site for all of 30 seconds and saw this [1]:<p>> Only 0.1% of published exploratory research is estimated to be reproducible.<p>Which sounds downright <i>absurd</i>. Only one in one thousand studies are <i>even in principle</i> able to be performed again? So I followed the reference.<p>As far as I can tell, it doesn't say that. Based on skimming the reference[2], I think the 0.1% number comes from the section "An Example: Science at Low Pre-Study Odds".<p>It is not talking about reproducibility, it is talking about p-values. It is not talking about studies in general, it is talking about a specific type of study (whole genome association). It assumes statistical mistakes are made on purpose, as opposed to estimating how often this is done in reality, and that the prior probability of a given association being true is 10^-4. The example-of-a-bad-study study barely moves that probability to 1.5 * 10^-4 (that's why it's a good example of a bad study), which the quote rounds back down to 0.1%.<p>The quote is an egregious exaggerations, especially for a site criticizing a method of gaining knowledge. Hopefully the whole site is not like that.<p>1: <a href="http://www.scienceisbroken.org/posts/30" rel="nofollow">http://www.scienceisbroken.org/posts/30</a> <i>edit</i>: it appears that the fact changes every time. I don't know how to link it.<p>2: <a href="http://www.plosmedicine.org/article/info:doi/10.1371/journal.pmed.0020124" rel="nofollow">http://www.plosmedicine.org/article/info:doi/10.1371/journal...</a>