
191Analyses
and there is no a priori knowledge to help the selection. Having noted this,
the modeler can then implement both forms of the equation and dene a
ag that tells the simulation which equation to use in any given model run
(Figure5.6). During execution, the ag is rst set one way for one run and
then reset the other way for another run. The results are then compared to
determine the signicance of the uncertainty around that particular struc-
tural factor. If it turns out that it matters (e.g., it changes the direction of
the HTA decision), then at a minimum, the decision maker must be given
this information. Ideally, further work ...