Stated preference techniques and consumer choice behaviour

Abstract

This PhD thesis examines the way in which individuals make choices during stated preference experiments (commonly referred subsets of which are called stated choice methods, conjoint analysis and trade-off analysis). Stated preference experiments ask respondents to rank, rate or choose between different product/service options, which are made up of a number of attribute mixes. The responses made by individuals within these experiments allow researchers to estimate consumer preferences. This thesis traces the historical background of stated preference experiments, from the field of utility theory and experimental economics. An understanding of this historical background explains the reliance by practitioners on the assumption that respondents make rational choices during the stated preference experiment (where all the information presented to them within the experiment is traded off in order to come to an overall preference). In light of considerable research evidence within the field of psychology that consumers do not do not conform to this economic concept of rational choice, and recent criticisms within recent stated preference literature, this thesis identifies the choice strategies employed by respondents during three stated preference experiments, where attributes were represented in different ways. Choice based stated preference experiments designed as the context for this research, measure consumers preferences for a newly developed fuel-efficient vehicle, with attributes currently unavailable in the marketplace. The experiments were presented to respondents as a series of choices between the newly developed vehicle and another currently available in the marketplace, described in terms of a number of attributes. The experiments were implemented using `think-aloud' protocol to allow the identification of respondent's choice strategies. The research successfully identifies the choice strategies employed by respondents during the stated preference experiments, and in support of recent criticisms within stated preference literature, finds significant deviations from the economic concept of rational choice. Furthermore, significant differences between the choice strategies employed by respondents are identified between the experiments where the appearance of the vehicles is represented in different ways. Using response data that is simulated to mirror the respondent choice strategies identified in each of the three stated preference experiments, the research tests the implications of these choice strategies on the estimation of consumer utility models. The research identifies significant differences between the parameter estimates derived from responses simulated assuming different choice strategy profiles. The research also identifies significant improvements in the estimated parameter values when the identified choice strategies are used in the analysis of the response data, rather than using the assumption of rational choice as an approximation. This suggests that stated prelcrence practitioners might improve model estimation by identifying the choice strategies used by respondents to inform the analysis of stated preference response data.