Violations of scale compatibility: results from pricing tasks and choice tasks in choice experiments

According to rational choice theory, preference orderings should be invariant with respect to the elicitation procedure. The contingent trade-off model (Tversky et al., 1988), however, argues that if an attribute of decision alterna-tives is also used as response mode (scale compatibility), then the attribute will be weighted more heavily. We propose a two-stage design for a systemat-ic assessment of how scale compatibility impacts willingness to pay (WTP) es-timates in choice experiments. At the first stage, a pricing task is implemented. Respondents face two alternative goods. For one good, the price is given. For the alternative good, respondents state the price (WTP) that makes them indifferent between the alternatives. In the second stage, a choice task is implemented: another group of respondents makes pairwise choices between two alternative goods with the price being one of several attributes. Our empirical findings support the contingent trade-off model, as pricing tasks yield systematically lower WTP estimates than choice tasks. While the trade-offs between the other attributes do not differ between the pricing and the choice task. Thus, in the choice task the weight shifts away from the price attribute but does not change the relative weight of the other attributes.