Skip to content

Scientific discoveries based on probability analysis are biased and probably wrong

This well researched Guardian article, published Nov. 9, 2009, on the use of statistics in scientific research is good for scientists and non scientists to read.  It discusses how a particular study, just by random chance, can get a scientifically interesting result and then get that result published in a prestigious scientific journal.  It also discusses how the current model of scientific research can bias the data and methods such that interesting results are more likely to be claimed.
The authors give a specific example of a study about the link for a serotonin enzyme and depression.  Initially, one study published a strong link between a mutant enzyme and depression.  That study received a lot of attention and was published in Neuron, a very well-know journal. Subsequent studies did not support that link but received much less attention and were published in less prestigious journals.
Based on the publication bias, the authors warn the readers that many published studies are probably wrong.  They also discuss some social reasons for why scientists over reach in their claims.  For example, one social cause of problems is that funding and tenure is strongly based on how good a discovery is so scientists have a strong incentive to find support for an important discovery.

Reblog this post [with Zemanta]
Related Posts with Thumbnails

Posted in Web Log.

Tagged with , , , , .

0 Responses

Stay in touch with the conversation, subscribe to the RSS feed for comments on this post.

Some HTML is OK

or, reply to this post via trackback.