Posted: Feb 09, 2013 9:23 am
by Pebble
Transilvanian wrote:So do you think that ESP corresponds to the scientific criteria and is not accepted just because is "paranormal"?
@[color=#CC0000][b]tolman:[/b][/color] Both.



The problem with a data driven approach to proving something is that the quality and completeness of the data is paramount. If negative results are less likely to be published, if negative trials are less likely to be published in searchable journals, if the differences in trial construction and reporting make meta-analysis unreliable, if the data do not follow the usual funnel plot spread etc then there are valid worries about the data.

In most scientific fields we do not use a purely data driven approach. We have good mechanistic reasons for considering a relationship possible. We have initial pilot trials demonstrating the magnitude of effect. Then we devise confirmatory independently performed trials of adequate size and power.

If someone then goes back and decides to check whether the data in isolation would prove the association, not all the above criteria would necessarily be met. So one can claim that we have a 'scientifically accepted' position based on incomplete data! In reality this is ignoring the other supportive data and hence comparing apples and oranges. Where the implications of the association are important or challenged then further studies are performed and the data requirements will then be met - think for example about MMR and 'autism' where because of fradulent activity and reporting huge amounts of work had to be done to prove the absence of any association.