The process can inform the choice of the appropriate model. For example, in the above, Model 2 may be the favoured approach if itis believed that the final analysis used in decision support will require running a sensitivity to market size or share (since these are input variables in Model 2, but not in Model 1 nor Model 3).