Attribute Processing in Choice Experiments and Implications on Willingness to Pay

Abstract

There is a growing interest in understanding the rules respondents invoke to deal with the information presented in stated choice (SC) studies. Although the impetus for this focus appears to have been motivated by an interest in cognitive burden, research by Hensher and DeShazo (amongst others) has found that the real issue is not the amount of information respondents are expected to process, otherwise known as ‘complexity’, but rather the relevance of the information. This discovery opened up the possibility that a study of the implications on choice response of the amount of information provided in a choice experiment should be investigated in the context of the broader theme of what rules individuals bring to bear when assessing the information. These rules may be embedded in prejudices that have little to do with the amount of information in the experiment; rather they may be rational coping strategies that are used in everyday decision making. There is an extensive literature on information processing, which includes prospect theory (Kahnemann and Tversky, 1979), case-based decision theory (Gilboa and Schmeidler, 2001), and nonexpected utility theory (Starmer, 2000). This literature has not yet been integrated into the modeling of the SC process. This chapter promotes the case for increasing our knowledge of the roles played by (i) the dimensionality of a SC experiment, (ii) the framing of SC design profiles relative to an experience profile (a reference base) and (iii) aggregating attributes. If these factors influence choices made, then it is important to understand how. It is also important to account for these influences, both with respect to the resulting willingness to pay (WTP) estimates and in any use of model outputs in prediction on hold out samples.