REPLY Structural changes and validity (SD6660)

[SDMAIL ybarlas]

Posted by  ybarlas at boun.edu.tr

Quoting SDMAIL Monte Kietpawpan <kietpawpan at yahoo.com>:
> > 
> > No existing validity 
> >   tests serve to check whether the relationships in the real system 
> >   will not significantly change in the future--within the time horizon. 
> >   ...

Dear Monte,

Your questions are important. It is  true (and essential) that any given system
dynamics model  assumes that the corresponding system  structure does not
significantly change in the time horizon. Indeed, in testing the validity of a
model, we are testing if we can come up with a simplified, constant
approximation to the real structure (that is for sure changing in reality) that
can reproduce the essential dynamical behavior patterns of the real problem. 
Our model, our dynamic hypothesis is in a sense  a constant structural
approximation to very messy, changing structures. This is why there is no
'valid' or invalid models; there is a continuum of model validity: terrible,
bad, OK, good, great, superb!... 

About your assertion that I quoted above: It is still possible to model and
test problems that involve structures that change significantly in time. 
We then build SD models that try to represent the way the structures change. 
This 'meta' model will still have a constant structure, but this 'meta' constancy
is now about the rules that are assumed to govern the structural changes in the
real problem. And when we test this type of model, we would be testing your
question quoted above. Not many of us build such models, but it is perfectly
possible within the realm of system dynamics modeling. 

thanks and all the mest,

Yaman Barlas
Posted by  ybarlas at boun.edu.tr
posting date  Tue, 9 Oct 2007 20:18:12 +0300