According to this piece approximately 50% of published studies from academic labs are not replicable. Not too surprising given the complexity of the variables involved, but not often acknowledged... (and how poor would the rate be among NON-published studies, which are the vast majority?):
http://marginalrevolution.com/marginalrevolution/2011/09/how-good-is-published-academic-research.html
In a slightly related matter, Ben Goldacre points to a widespread statistical error, dealing with "the difference in differences," that is commonplace in published (neuroscience) studies:
http://www.guardian.co.uk/commentisfree/2011/sep/09/bad-science-research-error
No comments:
Post a Comment