A necessary part of any system dynamics analysis is the estimation of parameter values which best correspond to the real system. The method of estimating parameters for a typical system dynamic model usually involves two steps (with potentially multiple iterations in the second) 1. Make an a priori estimate based on direct observation, educated guesses by managers, or similar parameters in other settings (Graham (Graham [1980] refers to this as data {at or below the level of aggregation); and 2. Revise those estimates in the process of calibrating the model to aggregate data.