More than the studies being bogus, the
statistical machinery in common use is bedeviled
with problems. p-values are problematic; they
don't mean what you would like them to mean.
Significance testing is on shaky grounds. Why do
papers have a discussion section that discusses
the plausibilty of a p-value or a significance
test? After all if the test was doing its job,
the conclusion sould be a slam dunk with no need
to second guess it or bolster it.
- Shyam
On 9/27/07, Rishab Aiyer Ghosh <[EMAIL PROTECTED]>
wrote:
>
> > statistics, though, when one of the most read
> > papers in Plos medicine
> > recently was a study showing that over 70% of
> > medical studies were
> > bogus, simply due to poor statistics.
>
--- Charles Haynes <[EMAIL PROTECTED]>
wrote:
> Then they'd get mad when I'd say "results not
> significantly different
> from expected by the null hypothesis." Hey,
> negative results
> contribute to human knowledge too, right?
____________________________________________________________________________________
Fussy? Opinionated? Impossible to please? Perfect. Join Yahoo!'s user panel
and lay it on us. http://surveylink.yahoo.com/gmrs/yahoo_panel_invite.asp?a=7
