Assessing the model
How can you can tell what is good enough? Even Fossil Creek Software can't
tell you that. Although various statistical techniques can go part of the way,
all fall short of providing any single, reliable measure of goodness.
Your goal will not be to prove how valid your model is, but instead, how invalid
it is. Put your model at risk. See if next year's incoming data
can throw the whole simulation into disarray. If it does, you should question
everything and start over. If it doesn't, you are on the way to building some
confidence in your model.
What about the predictive simulations you might make? There is certainly no way to
predict what the winters are going to be like or whether there will be a change in
the reproductive rate. What you can do is to create plausible scenarios
to get a feel for what a range of conditions might bring. In this way, you can be
better prepared for the unexpected. You will get a feel for the risks you assume
when you must make a harvest recommendation. With a little imagination, you can put
together a worstcase scenario to test in the model.
Finally, show your simulation to others who are in a position to be
critical of your model. You can learn a lot by trying out their ideas about how the
population might be working. Compare and contrast to see if you can come up with a
reasonable compromise, or present your respective alternatives to the next highest
Previous (2 of 3)