AC 2012-5240: THE INFLUENCE OF THE ”DECOY EFFECT” ON THE ENGINEERING DESIGN PROCESS Dr. Joseph C. Musto, Milwaukee School of Engineering Joe Musto is a professor of mechanical engineering and Director of the Mechanical Engineering Program at Milwaukee School of Engineering, He holds a B.S. from Clarkson University (Potsdam, N.Y.), and both a M.Eng. and Ph.D. from Rensselaer Polytechnic Institute (Troy, N.Y.). Dr. Alicia Domack, Milwaukee School of Engineering Page 25.1316.1 Page c American Society for Engineering Education, 2012 The Influence of the “Decoy Effect” on the Engineering Design Process Abstract Engineering students are educated in formal design methodologies to aid in the decision-making processes involved in the creation of new products and systems. These formal methodologies make use of such design tools as decision matrices to aid engineering teams in the evaluation and selection of a design solution from the various alternatives considered for development. The Decoy Effect (or Asymmetrically Dominated Alternative Effect), is a well-studied phenomenon that affects decision-making. In essence, it describes the experimentally-verified effect that occurs when an inferior choice is introduced to the available alternatives. This inferior choice is said to be “dominated” by one of the original alternatives, which then prompts individuals to choose the dominating alternative 1,2 . The effect of this so-called “decoy” choice often leads the decision-maker to the selection of a suboptimal alternative. The decoy effect has been shown to adversely affect decision-making capabilities involving consumer product purchases, gambling, and job offers 3,4,5 . While the impact of this effect would certainly extend to decision-making among technical professionals, there has been little attempt to address the impact on the decision-making process used by engineers in design. This paper introduces the decoy effect and describes the implications it can have on the engineering design process. Initial small-scale experiments to validate and quantify the effect as it relates to engineering design will be described. Development and modification of design tools to mitigate the impact of the decoy effect in the engineering design process will be described. The Decoy Effect Individuals are often faced with problems that have uncertain solutions. In such cases, the context of the situation becomes influential on the decision outcome 6. Such context effects could be the number of choices available, the quality of the choices, or the value that the individual places on the qualities of the choices. One may assume that individuals make choices based upon how they can best maximize value, but it has been found that is often not the case. Many studies have found that individual choice can easily be manipulated by the inclusion of a “decoy” alternative 3,5,7 . The Decoy Effect (or Asymmetrically Dominated Alternative Effect), has been found to influence the decision making process when choice is uncertain. If two options are considered equally viable and a third option is introduced (the decoy), which is asymmetrically dominated 25.1316.2 Page by only one of the original options, individuals will likely choose the option that is easily compared to the decoy. Therefore, it is important to assess the context within which the decision is being made in order to understand the rationale behind the decision making process. For example, imagine you are buying a car and you have decided there are two qualities that are of equal importance to you, cost and fuel economy. You have narrowed down your search to the following vehicles: Car Fuel Economy Price A 28mpg $17,500 B 38mpg $24,500 It is expected that either alternative could be chosen because there is a balance between cost and fuel economy. Now imagine another alternative thrown into the selection pool: Car Fuel Economy Price A 28mpg $17,500 B (target) 38mpg $24,500 C (decoy) 38mpg $26,500 The choice is now made easier because the decoy is asymmetrically dominating one alternative, in this case, choice B. Individuals are now more likely to choose car B, because it is clearly superior to car C and this reduces the cognitive strain of having to compare equally viable options. The Decoy Effect has been found to affect decisions in a variety of situations but has not been investigated in engineering decision making 2,3 4,5 . The Decoy Effect can be seen as a way for individuals to minimize cognitive strain. The brain has limited resources and decision making requires the use of memory, attention, and prior knowledge, which could be inaccurate or inaccessible. Therefore, individuals often rely on simple and easy to access information, which leads to biases and the use of heuristics 8. Heuristics and biases, which can be seen as “short-cuts”, may have led us to appropriate decisions in the past, but they do not guarantee a successful outcome. On the other hand, when using an algorithmic approach, an individual will use a step-by-step, often complex, set of rules to come to a conclusion. If the set of rules are followed correctly, the individual will be lead to the correct result. Several studies have found that when an algorithm is taught, this process will be used more often than a heuristic when individuals are faced with complex decisions 9,10 . It is hypothesized that individuals enrolled in an academic engineering program will learn appropriate algorithmic ways of dealing with uncertain decisions (i.e. decision matrices or morphological charts) and use the algorithm instead of relying upon the heuristics and biases that untrained individuals may use. 25.1316.3 Page A Small Scale Study to Quantify the Decoy Effect in Engineering Design The engineering design process is essentially a process of structured decision-making under uncertainty; the engineer seeks to select a design solution from a set of alternative designs that meets a set of performance specifications and is optimal in some respect. 11 While many versions of the design process have been published, each includes some version of the following basic steps 12,13,14 : 1. A design problem is identified 2. The problem is defined and quantified using formal specifications 3. Many alternative design solutions are synthesized in response to the specifications 4. One “best” solution is selected from the various alternative solutions 5. Detailed design and optimization is performed to fully develop the selected design alternative 6. The completed design solution is evaluated, to ensure that specifications are met 7. The design is communicated to the customer An important aspect of the formal design process is that while many alternative designs may be synthesized, there is generally only sufficient economic resources to fully detail one candidate design solution. Therefore, Step 5 indicates that only one “best” design can be selected from the list of design alternatives. The selection of a single “best” design involves comparison and evaluation of the alternatives; since the Decoy Effect has been shown to influence such decisions in a variety of other topical areas, it is reasonable to believe that design engineers may be swayed in their evaluation of candidate designs by the presence of asymmetrically dominated alternatives. A reliable engineering design process would be one that would minimize the impact of the decoy effect on engineering decision-making, and allow for justifiable selection of an optimal alternative. In order to quantify the impact of the Decoy Effect on the selection of an alternative design, the results of a design exercise used in two engineering design courses were evaluated. In each of these courses, the students were provided with the design case study shown in Figure 1. The case study was designed such that while all three designs meet the specifications, Design A is asymmetrically dominated by the similar Design Alternative B in all categories related to the decision criteria (cost, size, weight, and reliability). Design Alternative C was chosen such that there was no clear choice between Alternatives B and C; tradeoffs between cost, size, and weight would need to be weighed to determine the “best” choice from these two alternatives. In order to validate the case study, the three design alternatives were presented to four mechanical engineering faculty members, two at a time. In all four cases, the faculty members judged Alternative B to be preferable to Alternative A, Alternative C to be preferable to A, and Alternatives B and C to be equally viable solutions. 25.1316.4 Page _________________________________________________________________________________ Assume the role of a Project Engineer working on the development of a new high-speed printing press. Three design alternatives have been proposed by the project team to act as the primary power transmission mechanism to drive the main line shaft with a 0.5 horsepower electric motor. All three designs meet the requirements for the system. You must select the best design alternative from the three proposals, based on a balance of cost, weight, size and reliability (all are equally important in this application). The three designs proposals are summarized on the following page. Select one and only one of the three design alternatives as the choice for the machine. Write a brief (one-sentence ) rationale for your decision Alternative A: Beltronics Sealed Belt Drive (uses toothed belt and pulleys) Manufacturer: Beltronics Power Transmissions, Missasagua, Ontario Cost: $475 Weight: 41 lb. Size: 27” x 19” x 9” Reliability: 92% at 5,000 hrs. of operation Power Rating: .55 horsepower Alternative B: TransPower Sealed Belt Drive (uses toothed belt and pulleys) Manufacturer: TransPower Drive Systems, Cupertino, CA Cost: $300 Weight: 30 lb. Size: 22.0” x 16.0” x 7.1” Reliability: 97% at 10,000 hrs. of operation Power Rating: .75 horsepower Alternative C: GearMax Sealed Gear Drive (uses helical gears) Manufacturer: GearMax Transmission Components, Davis, CA Cost: $310 Weight: 29 lb.