Choke price bias in choice experiments

This paper reports on the results from a split sample choice experiment, where the effect of changing the maximum level of the cost attribute is investigated. The hypothesis was that changing the maximum price level would either give rise to an income effect or have no effect on consumers' preferences. This was tested in the framework of a valuation study of different quality characteristics of minced pork. The data was analysed using a mixed logit error component model, which accounts for correlation in the unobserved part of the utility. Results suggest that the size of the maximum price level does matter, and that changing the maximum price level has a statistically significant effect on both the general preferences structure and the WTP estimates. Hence, researchers should be very careful when defining not only the range of the price attribute but also the maximum level of the price attribute.