(Fleecing thread): Fake Your Data

[Alexander Berezin]

On 25 Mar 1996, Leonard P. Paplauskas wrote:

> hrubin at b.stat.purdue.edu (Herman Rubin) wrote:
> 
> >So investigators keep results back so that they can use them 
> on a new >grant, and will often fake or select data to make thir 
> progress look >good. > 

>  If you had any credibility before you made the statement above about 
> faking data, it's completely gone now!
>
> Leonard P. Paplauskas
> Assistant Vice President for Research 
> University of Connecticut Health Center 
> Farmington, CT   06030-5355 
> 203-679-3173                         
> FAX 679-2670                                            
>                                                                 
> paplauskas at adp.uchc.edu
> 

Any-one is, of course, free to withdraw his/her credibility 
from anyone theey (dis)like but the issue of faking and/or
manupulating data to improve 'fundability' is not going to 
disapper. You can't solve the problem by putting it under 
the carpet and pretending it does not exist.

There are 2 (opposite) hypotheses on the frequency of 
data faking:

(A) Data faking is very rare indeed and all (but
perhaps few cases of it) are revealed to the public

(B) Data faking is very common and what is publicly
reported is just a tip of an iceberg. Liones share
of it goes undetected and unpunished.

As any-one learned in the art of science discourse 
can see, BOTH of these hypothesis satisfy the 
available 'experimental data' [ and that is the 
relatively rare happening of PUBLICLY REPORTED cases 
of research misconduct ].

For as long that selecitivity/grantsmanship/
percentile/, (etc) model works, it makes a good 
pragmatic sense to fake (trim, manipulate) data
(to get grants), and, correspondingly, the 
hypothesis (B) is likely MORE plausible than (A).

Alex Berezin