Risk, Uncertainty and Ambiguity in Engineering Design Decision Making

Robert P. Smith
Industrial Engineering
University of Washington

The decision-based design philosophy intends to analyze how designers frame and make decisions. This field of design research can and should profit from established results on decision making in the organizational community with which it is closely related. The organizational work concerns how individuals make decisions, and how organizations and managers create environments that support or discourage types of decision making [Weick 1995]. The body of work on organizational decision making recognizes three distinct types of decision making: decision making under risk, decision making under uncertainty, and decision making under ambiguity (also known as equivocality) [Einhorn and Hogarth 1986, Daft and Macintosh 1981, Daft and Lengel 1986]. This position paper is intended to describe these three types of decision making, and to show how they are relevant to engineering design. Decision making under risk constitutes the condition where there is information unavailable, but a probabilistic description of the missing information is available. A technical decision of this type might be that a manufacturing engineer knows the probability distribution of manufacturing process outputs, and is trying to determine how to set an inspection policy. The design response might be to construct a stochastic program and find a minimum cost solution for a known defect rate. Decision making under uncertainty, by contrast, involves distributions are unknown. This situation involves less knowledge than decision making under risk. A situation that involves decision making under uncertainty might be that a communications design engineer knows that transmission quality is a function of the antenna design, the frequency, and the background radiation, but is unsure of what the distribution of background radiation will be in the user environment. In this situation the design response might be to collect field data in the user environment to characterize the radiation, so that antenna design and frequency could be chosen. Decision making under ambiguity involves a still more profound lack of knowledge. Decision making under ambiguity means that the functional form is completely unknown, and often that the relevant input and output variables are unknown. An example of decision making under ambiguity is a design engineer who is considering building airplane wing panels out of composite materials, but is uncertain of the ability of the new materials to withstand shock loads, and indeed which design variables might affect shock loads. The engineering design response to this situation might be to start a research project that will vary possible input variables (panel thickness, bond angle, securement method, loading and others), and determine which, if any, of these variables has significant effect on shock resistance.

Problem definition is an important part of the problem solving process [Smith 1993, Smith 1989]. It has been noted that technical problem solvers exert control over the content of ambiguity in decision making [Schrader and others 1993]. The same engineering design problem can often be expressed as a situation involving risk, uncertainty or ambiguity. A manufacturing engineer could decide that a stochastic description of defect formation is available and accurate, and use that to design an inspection policy (decision making under risk), or alternatively could believe that defects are arising from an unknown source and attempt to trace down the source and quantity of the defects through a thorough investigation of the production environment (decision making under ambiguity). This choice of decision making mode has important implications for the amount of resources needed to solve the problem, and the technical value of the produced design [Smith and Hausjah 1997].

Problems framed under ambiguity typically require greater resources to find a solution than do the same problems framed under uncertainty. Solutions to problems framed under ambiguity generally are more well understood and transferable than solutions to problems framed under uncertainty. Designers are aware of these effects as they choose how to structure their design problem. There are several significant outcomes of the distinction between risk, uncertainty, and ambiguity for the decision-based design approach. First, we must recognize that there are many situations in engineering decision making where no stochastic description exists, or is possible. Understanding how the human designer makes the decision of what to consider the design problem is often more interesting than understanding how the problem is solved once it has been defined. Second, whether a stochastic characterization of a situation exists is not a decision that is extrinisic to the decision, but is one that is under the explicit control of the decision maker. If a stochastic description is posited it should be recognized that this implies an underlying assumption of decision making under risk (or uncertainty, or ambiguity, if the framework is structured appropriately). Because of the effects that the choice between risk, uncertainty and ambiguity implies it is a determination that should be made carefully.

References

Daft, Richard L., and Norman B. Macintosh, "A Tentative Explanation into the Amount and Equivocality of Information Processing in Organizational Work Units," Administrative Science Quarterly, Vol. 26, pp. 207-224, 1981.

Daft, Richard L., and Robert H. Lengel, "Organizational Information Requirements, Media Richness and Structural Design," Management Science, Vol. 32, No. 5, pp. 554-570, 1986.

Einhorn, Hillel J., and Robin M. Hogarth, "Decision Making under Ambiguity," Journal of Business, Vol. 59, No. 4, pp. S225-S250, 1986.

Schrader, Stephan, William M. Riggs and Robert P. Smith, "Choice over Uncertainty and Ambiguity in Technical Problem Solving," Journal of Engineering and Technology Management, Vol. 10, pp. 73-99, 1993.

Smith, Gerald F., "Defining Managerial Problems: a Framework for Prescriptive Theorizing," Management Science, Vol. 35, No. 8, pp. 963-981, 1989.

Smith, Gerald F., "Defining Real World Problems: a Conceptual Language," IEEE Transactions on Systems, Man, and Cybernetics, Vol. 23, No. 5, pp.1220-1234, 1993.

Smith, Robert P., and Carolina M. Hausjah, "Implications of Choice over the Level of Ambiguity in Design Problem Framing," Industrial Engineering working paper, University of Washington, revised July 1997.

Weick, Karl E., Sensemaking in Organizations, Sage Publications, Thousand Oaks, Calif., 1995.