Properties of the Decision Environment and Display Organization in Choice

This paper describes two experiments and a simulation, aimed at investigating the impact of specific features of the decision environment and of the organization of the information display on choice processes and outcomes. The first experiment compared a matrix organization with a Cartesian graph, while also manipulating the distribution of the decision weights and the dominance structure of the problems. It was conducted to test the predictions of models derived from different choice frameworks, but none of the models were able to provide a full account of the results. Therefore, an alternative explanation was put forward, relying on the simulation of a variant of the lexicographic heuristic (LEX-2). The second experiment supported the existence of LEX-2 and explained a counter-intuitive display-related effect brought out by the first study. Theories and Models of Choice The first experiment presented in this paper takes into account both the influence of the display and the role of two significant properties of the decision environment: the distribution of importance weights and the dominance structure of the decision problems. Two prominent choice frameworks were selected and their predictions are clearly set out. The theories will be described in general terms in the following sections, while their specific predictions will be presented further on in the paper. The Adaptive Decision Maker Framework The adaptive decision maker framework (Payne, Bettman, & Johnson, 1988, 1993) regards choice as a constructive process. According to this theory, the decision maker (henceforth DM) is sensitive to the specific properties of the decision environment and is able to construct a choice strategy or to select it from memory in an adaptive way. The basic assumption of the framework is that the DM will use a heuristic that assures a satisfying trade-off between the anticipated accuracy of the decision and the anticipated effort required in the choice process, given the specific features of the decision environment. This framework has recently included in the main decision goals the need to avoid negative emotions and the desire to justify the decision (Bettman, Luce, & Payne, 1998). The properties and the performance profiles of different choice strategies have been analyzed through simulations, and the predictions of the framework have received a good degree of experimental support (Payne et al., 1993). This research effort has provided us with knowledge about the effects of different task and context variables on strategy selection. The dominance structure of the decision problem is one of the variables that has been taken into account. It was shown that the presence of dominated options in the choice set allows noncompensatory heuristics to perform reasonably well (Payne et al., 1993). Another significant variable is the distribution of the DM’s importance weights. These weights represent the importance that the DM gives to the various features of the available options. The weight distribution, together with the value functions describing the DM’s subjective evaluation of the attribute values, defines the DM’s subjective utility function. Kleinmuntz and Schkade (1990, 1993) analyzed the impact of the information display variables on choice tasks. Their cost-benefit proposal complements the adaptive decision maker framework. According to them, strategy selection is also sensitive to the features of the display: “This cost-benefit perspective implies that information displays define a cognitive incentive system for decision makers, whereby displays influence the effort and accuracy of each available strategy and, therefore, induce decision makers to use different strategies” (Kleinmuntz & Schkade, 1993, p. 224). Kleinmuntz and Schkade (1994) highlighted the role of three features of the display: the organization of the information items, their form, and their sequence. The information display may affect choice strategies through the reduction of the cognitive effort required. For instance, if comparisons are simplified by the display, the effort involved in comparison-based heuristics will be reduced, increasing the likelihood that a (sufficiently accurate) procedure of this kind will eventually be adopted. Alternatively, some strategies could be more prone to error with some specific displays. In these cases, accuracy concerns may lead the DM to avoid potentially unsound procedures. The effects of the display variables on strategy selection are mediated by the anticipation of the effort required and the accuracy of the different strategies. It has been shown that these estimates (in particular the accuracy anticipation) may be incorrect (Fennema & Kleinmuntz, 1995). The cost-benefit approach predicts that the DM will learn to select the best strategy from experience, using the feedback on the experienced effort and accuracy collected during previous choices. However, despite the evidence for adaptive decision behavior, there are also indications that the DMs are not very sensitive to accuracy feedback (Brehmer, 1980). The Search for a Dominance Structure Theory Montgomery (1983, 1989) proposed the Search for a Dominance Structure theory (SDS), postulating a tight linkage between decision and action. A decision is seen in the context of an intention to act, which is defended and supported by the DM. Therefore, it is important that the DM is able to find good reasons for his/her choices. Being able to identify a dominance structure provides a particularly strong justification for choice: if an alternative dominates the other options in the set, this choice will be very easy to defend. Therefore, the DM will be motivated to find a dominance structure and, if this does not exist, he/she will restructure the problem, in order to bring about such a structure out. According to the SDS theory, the dominance structure found by the DM may be close or far from a strict dominance relationship: this distance will depend on the type of problem presented and on the specific type and sequence of restructuring operations applied by the DM. In the process that leads to the definition of a dominance structure the DM will go through the following stages (Montgomery, 1983): (a) pre-editing, (b) search for a promising alternative, (c) dominance testing, and (d) dominance structuring. Different known decision rules may have a local function in this process, depending on the stage in which they are used. In the pre-editing stage, the DM selects alternatives and attributes to set up the problem representation. The less important attributes and the less valuable options may be discarded. During the search for a promising alternative, the DM identifies an option that could have some chance of dominating the others (or to be better). This promising option is then evaluated through an absolute or relative test. The absolute test may be carried out using a procedure similar to a satisficing heuristic (Simon, 1955). The relative test is based on pairwise comparisons of the attribute values of different alternatives, and it can be considered as a real test of dominance. If the promising alternative receives a good evaluation (i.e. the dominance test succeeds), the DM will choose the current option. If the dominance test fails, the DM will enter the dominance structuring stage. In this stage a decision problem can be restructured in many ways. The de-emphasizing operation is used to reduce the importance of a disadvantage or of a difference between alternatives concerning the promising option. The bolstering operation increases the advantages of the promising option or decreases the advantages of the other options. The cancellation operation rules out a disadvantage through the association with a (logically connected) advantage. Finally, the collapsing operation ties together different attributes in a single dimension (for instance, using a time or money scale). If the dominance structuring process is not able to find a satisfying solution, the DM starts again from the pre-editing stage or searches for a new promising alternative.