The Effect of Choice Environment and Task Demands on Consumer Behavior: Discriminating Between Contribution and Confusion

  • Author(s) / Creator(s)
  • Choices, whether they are made in actual markets (revealed preferences or RP) or in surveys (stated preference of SP), provide information about the preferences of individuals. These choices also contain what researchers interpret as \"noise\" or unexplained variation; a variety of techniques (e.g. statistical design theory, econometric specifications, and the combination of stated and revealed preference data) have been used to understand preferences and separate the signal from noise. However, in applying these tools we have tended to focus on the information provided by the choices themselves, to the detriment of understanding the effect of the choice environment or the task demands on the observed choice behavior, the quality of the information provided, and noise levels. Task demands (or the choice environment) can be characterized by such factors as choice set size (the number of alternatives the individual is choosing from), the number of attributes under consideration within alternatives in a choice set, the correlation structure of the attributes of the alternatives in the choice set, the number of and sequence of choices the individual is asked to make during the \"task,\" as well as a variety of other factors. While the definition of the choice set, and the implications of choice set definitions on empirical results (Swait and Ben-Akiva, 1987a,b), has received some attention in the literature, there has been little attention paid to the characterization of the choice set or other dimensions of task demands. This paper presents an analysis of context effects on choice in the same tradition as Heiner (1983), who investigated the processing capacity of consumers. We characterize task demands and incorporate them into random utility models of choice. We describe task demands using entropy, a well-known information theoretic mesure which provides a summary measure of the uncertainty inherent in the choice environment. We model choice as a function of attributes in a traditional compensatory model but include our summary measure of task complexity in the variance term of the model. Both entropy and cumulative entropy are included in the model to account for the fact that task demands are partly defined by the current choice set, and partly by prior effect expended (which we term \"cumulative cognitive burden\"). We use this model to examine choice data from a number of contexts. It is worth noting that in such a model elasticities are now made up of two components: a direct effect through the impact on utility and an indirect effect through the impact on complexity. Our results indicate that task complexity significantly affects the variance of choice in a fashion that is consistent with notions of limited consumer processing capacity and cognitive budgets. Employing six case studies that examine choice within very different product classes, we find that complexity affects variance in a non-liner fashion with both very difficult and very easy choices resulting in near random choice behavior. Furthermore, we find that pooling across the six cases, we cannot reject the null hypothesis that complexity affects these very different choice processes in a similar fashion. This is an interesting finding suggesting a generalizability of our formulation over a variety of product classes. We also find some evidence of fatigue and learning effects in the different case studies, although the cumulative effects are not as systematic as the direct impacts of complexity on choice. We begin the paper with a brief review of the literature on task complexity and choice, then outline several possible measures of complexity. We next expound on our chosen measure of complexity, entropy, as a form of characterizing task demands. A statistical modeling approach, that includes entropy in the variance function, is presented in Section 3. Examples using choice data from different situations are then examined in Section 4. The paper concludes with a discussion of the implications of our findings and future research topics.

  • Date created
  • Subjects / Keywords
  • Type of Item
  • DOI
  • License
    Attribution-NonCommercial-NoDerivatives 3.0 International