Revealing the extent of process heterogeneity in choice analysis: An empirical assessment

Choice analysts increasingly use a mix of revealed preference and stated choice data paradigms to identify preferences of samples of individuals that are used to infer behavioural response and willingness to pay for specific attributes. These data are in a sense artificial constructs that are developed to approximate real choice settings of the way that individuals process relevant information in making choices. As such, all data designs formalized through a survey instrument seek information through questions that become descriptions of events and as such the probabilities of choice that are of interest are strictly probabilities attached to event descriptions and not choice probabilities of events per se. The recognition of this distinction, initially noted by Kahneman et al. [Kahneman, D., Slovic, P., Tversky, A., 1982. Judgement under uncertainty: Heuristics and biases. Cambridge University Press, New York], can be captured, at least in part, through the idea of process heterogeneity, as a way of recognizing and accounting for the many ways in which individuals process information, and in part is influenced by the way the analyst describes the context in which preference data is sought. Building on previous contributions on attribute processing, this paper draws on recent empirical evidence to further reinforce the importance of joint modelling of process and outcome in choice analysis. This study adds to the evidence of a trend emerging on the upward bias of mean estimates of marginal willingness to pay when ignoring process heterogeneity.