Xref: utzoo sci.research:702 talk.politics.misc:23418 sci.bio:1927
Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!iuvax!silver!drsmith
From: drsmith@silver.bacs.indiana.edu (drew smith)
Newsgroups: sci.research,talk.politics.misc,sci.bio
Subject: Re: replicating experiments (was animal research)
Summary: it all comes out in the wash
Message-ID: <3442@silver.bacs.indiana.edu>
Date: 13 Mar 89 22:48:21 GMT
References: <726@orbit.UUCP> <3254@ttrdc.UUCP> <3437@silver.bacs.indiana.edu> <37128@bbn.COM>
Reply-To: drsmith@silver.UUCP (drew smith)
Distribution: na
Organization: Indiana University BACS, Bloomington
Lines: 24

In article <37128@bbn.COM> presnik@labs-n.bbn.com (Philip Resnik) writes:
>...it is
 >*absolutely necessary* to replicate experiments, not only because 
>replications confirm results, but also because experimenters have
>their own goals in mind, and this tends to influence the experiment.
>... *But* the replication
>of experimental results is absolutely critical -- if more "competent"
>scientists took the time replicating experiments, I believe science in
>general would benefit.  

Scientific knowledge is based upon a chain of inferences. If one or
more of those inferences is incorrect it will, sooner or later, lead
to the formulation of an experimentally testable hypothesis whose
predictions will be disproved. It is at this stage that it is most
worthwhile to the average competent scientist to repeat someone else's
experiment, in order to see if his/her reasoning, or the assumptions
it was based on was incorrect. This, to me, seems the most efficient
way of validating reported results. If a result is truly important, it
will lead to further experimentation, which will necessarily test its
accuracy. A result which does not stimulate further experimentation
will soon be deservedly forgotten, without any additional waste of
time and money.

- Drew Smith Dept. of Biology, Indiana University