Naked Science Forum

General Science => General Science => Topic started by: krool1969 on 26/03/2012 06:03:20

Title: Why publish data you know is probobly wrong?
Post by: krool1969 on 26/03/2012 06:03:20

When I first heard of faster than light neutrinos, the first thing I thought was, "OK what did they do wrong."
This same thought MUST have occurred to the researchers who found and then published the data. Why didn't they go back and check their data again?

Title: Re: Why publish data you know is probobly wrong?
Post by: Geezer on 26/03/2012 07:04:34

They did. Lots of times, but they could not find a problem, so they published it to see if anyone else could.
 
I'm pretty sure they were just as skeptical as you are, but data is data, and data should not be suppressed even if there is a good chance the conclusions might be suspect.

Title: Re: Why publish data you know is probobly wrong?
Post by: imatfaal on 26/03/2012 09:48:28

You must always publish wrong data!  That's the whole point of experimentation; in some fields the results are so smeared out that if you do enough runs of an experiment and only publish the stuff that is "right" you gain false credence for a theory that is completely incorrect.  Medical testing by the pharmacological companies is a prime example of this problem - Ben Goldacre has written extensively on it. 

For data where you believe the methodology is correct but you have a unknown bug - then you publish as Geezer mentioned to get thousands of extra minds on the case.  Additionally - it also stops others making similar mistakes

Tevatron:  My dog has got fleas - wish I knew how to get rid of them
LHC:  when my dog got fleas I gave him arsenic
Tevatron:  arsenic...hmmm

.....a few days later

Tevatron: I gave my dog arsenic and it just died
LHC: yep - that's what happened to mine