Page 285 - Mechanical Engineers' Handbook (Volume 2)

276 Analysis, Design, and Information Processing

There exists a strong need in the forecasting and assessment community to recognize and
ameliorate, by appropriate procedures, the effects of cognitive bias and value incoherence in
expert-opinion-modeling efforts. Expert-opinion methods are often appropriate for the ‘‘ask-
ing’’ approach to issue formulation. They may be of considerably less value, especially when
used as stand-alone approaches, for impact assessment and forecasting.
Simulation and modeling methods are based on the conceptualization and use of an
abstraction or model of the real world intended to behave in a similar way to the real system.
Impacts of policy alternatives are studied in the model, which will, it is hoped, lead to
increased insight with respect to the actual situation.
Most simulation and modeling methods use the power of mathematical formulations
and computers to keep track of many pieces of information at the same time. Two methods
in which the power of the computer is combined with subjective expert judgments are cross-
impact analysis and workshop dynamic models. Typically, experts provide subjective esti-
mates of event probabilities and event interactions. These are processed by a computer to
explore their consequences and fed back to the analysts and thereafter to the experts for
further study. The computer derives the resulting behavior of various model elements over
time, giving rise to renewed discussion and revision of assumptions.
Expert judgment is virtually always included in all modeling methods. Scenario writing
can be an expert-opinion-modeling method, but typically this is done in a less direct and
explicit way than in Delphi, survey, ISM, cross-impact, or workshop dynamic models. As a
result, internal inconsistency problems are reduced with those methods based on mathemat-
ical modeling. The following list describes six additional forecasting methods based on
mathematical modeling and simulation. In these methods, a structural model is generally
formed on the basis of expert opinion and physical or social laws. Available data are then
processed to determine parameters within the structure. Unfortunately, these methods are
sometimes very data intensive and, therefore, expensive and time consuming to implement.

• Trend extrapolation/time-series forecasting is particularly useful when sufficient data
about past and present developments are available, but there is little theory about
underlying mechanisms causing change. The method is based on the identification of
a mathematical description or structure that will be capable of reproducing the data
into the future, typically over the short to medium term.
• Continuous-time dynamic simulation is based on postulation and qualification of a
causal structure underlying change over time. A computer is used to explore long-
range behavior as it follows from the postulated causal structure. The method can be
very useful as a learning and qualitative forecasting device, but its application may
be rather costly and time consuming.
• Discrete-event digital simulation models are based on applications of queuing theory
to determine the conditions under which system outputs or states will switch from
one condition to another.
• Input–output analysis has been specially designed for study of equilibrium situations
and requirements in economic systems in which many industries are interdependent.
Many economic data fit in directly to the method, which mathematically is relatively
simple and can handle many details.
• Econometrics is a method mainly applied to economic description and forecasting
problems. It is based on both theory and data, with, usually, the main emphasis on
specification of structural relations based on macroeconomic theory and the derivation
of unknown parameters in behavioral equations from available economic data.