Re: Official Trek XI Website Has Teaser Trailer

Christopher C. Stacy wrote:
> "Jaxtraw" <jax@[EMAIL PROTECTED]
> writes:
>>
>> And it's worth remembering that if I show that eating more than 4
>> carrots per day causes gout in women over 65, and you show that
>> eating more than 7 carrots per day causes gout in West African men
>> between 40 and 60, that second experiment is not a replication of
>> the first one. They're entirely different experiments. In the ideal
>> world, if you've found a statistical correlation of that type, the
>> next step should be to use it as a prediction and repeat the
>> experiment to see if you get a similar result.
>
>> But that's hardly ever done with that kind of epidemiology, because
>> doing that would show that the overwhelming majority of such results
>> are just statistical noise, and nobody wants to discover that, do
>> they?
>
> I think a literature search (let me suggest just in the last 2 years
> in the area of bioinformatics) will suggest that people are very
> eager to discover exactly that.  (Umm, because they are trying to
> look for true answers?)

Umm, I think if you do that you'll find a lot of papers with weak
correlations between behaviour X and disease Y, each of which produces a
different RR somewhere between 1 and around 1.6, which naively appear to
be
mutually supporting but aren't, based on various data mining techniques
and
ignoring results that don't fit the paradigm. Have you looked at the
science
behind the current obesity hysteria? It's a disgrace. There seems to be
far
less interest in "true answers" than in expanding the funding circle jerk.