Working Paper No. 06-06
Preference Uncertainty, Preference Refinement and Paired Comparison Choice Experiments
David C. Kingsley
Paired comparison choice experiments allow researchers to measure within individual choice consistency and to retest originally inconsistent choices. Results from paired comparison choice experiments suggest that as respondents progress through a random sequence of paired choices they become more consistent, apparently fine-tuning their preferences (Brown et al., 2006). This paper investigates the implications of these results within a model of preference uncertainty allowing for preference refinement. Preference uncertainty implies that choices on each choice occasion are driven by realizations from an underlying valuation distribution (Thurstone, 1927; Li & Mattsson, 1995). Preference refinement is defined as a significant reduction in the standard deviation, referred to as the scale, of the valuation distribution. Results support the asserted inverse relationship between choice consistency and the scale of the model (Deshazo & Fermo, 2002). Additionally, the probability of observing an inconsistent choice decreases the greater the utility difference between the involved pair and the narrower the scale becomes. Finally, preference reversals are investigated using originally inconsistent choices and a subset of consistent choices which were retested within the experiment.