[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Response to "Position Statement to the Dunc-Tanc "experiment""

"Roberto C. Sanchez" <roberto@connexer.com> wrote:
> On Fri, Oct 27, 2006 at 03:21:10PM +0200, Thibaut VARENE wrote:
> > On 10/27/06, Anthony Towns <aj@azure.humbug.org.au> wrote:
> > >An "experiment" is successful as long as it provides useful information.
> > 
> > What the **** is this definition of "successful"??!
> > 
> First, foul language is not necessary.  Second, it is the academic
> definition.  See, an experiment is performed to confirm or dispel a
> hypothesis.  If it does either, and you can explain way or draw some
> other useful conclusion from it, then it is a success. [...]

References, please!  The dictionaries cited so far have been oblique 
justification at best.

Personally, I'd call an experiment that provides worthwhile information 
"useful", "worthwhile" or "valuable", but I'd not describe an experiment 
that fails to achieve a desired outcome as a successful experiment - 
that would be unnecessarily confusing.

I feel that that the problem at the core of this subthread remains: what 
hypothesis can Dunc-Tank confirm or dispel?  What other useful 
conclusion could be drawn?

Only one measurement for success/failure is obvious - whether etch 
releases on the date forecast at the start - but Dunc-Tank is not the 
only influence on that, and Dunc-Tank's fillers have not agreed any 
measurements, or ways to measure them.  I predict that we are going to 
get to the end of this trial and everyone is going to put forward 
personal opinions and anecdotes to justify whether this trial succeeded 
or failed, according to their prejudices.  Dunc-Tank will provide little 
useful information.

The structure of experiments is taught to teenagers as, roughly:
  1. phrase your research question;
  2. pick your outcome measure(s);
  3. determine/select your resources and design the trial;
  4. take your measurements while running the trial;
  5. analyse the measurements;
  6. suggest conclusions and/or further research.

As far as I can tell, Dunc-Tank now is no experiment worth the name.

"What is critical /is/ that the design be described in sufficient detail 
that it can be properly evaluated.  [...] Any study that is deficient in 
its design will rarely be able to settle the question that prompted the 
research, but it may be able to provide valuable information 
nonetheless." -- Gerard E. Dallal, Some Aspects of Study Design, in The 
Little Handbook of Statistical Practice, www.statisticalpractice.com

Does anyone care enough to rescue the deficient Dunc-Tank design enough 
to provide valuable information?  Can people even agree what information 
would be valuable?

MJ Ray - see/vidu http://mjr.towers.org.uk/email.html
Somerset, England. Work/Laborejo: http://www.ttllp.co.uk/
IRC/Jabber/SIP: on request/peteble

Reply to: