Does it matter how we produce ambiguity in experiments?

The Ellsberg urn is conventionally used in experiments to measure ambiguity attitudes, yet there is no uniformity in the method for producing Ellsberg urns, which complicates the comparability of results across studies. By surveying 69 experimental studies, we distill four different methods of ambiguity production—Ellsberg urns that are produced by (i) the experimenter, (ii) another random participant, (iii) compound risk lotteries, and (iv) compound risk derived from random numbers in nature. In an experiment we then assess participants’ ambiguity attitudes concerning each production method and detect no statistically significant differences among them. However, a notable proportion of preference inconsistency is observed when utilizing compound risk lotteries for ambiguity generation. Generally, our findings suggest interchangeability among the four production methods in future laboratory experiments. Nevertheless, we suggest employing method (i) as it is the most uncomplicated and straightforward production method.