Survey_BGJ_2013

Abstract

The experiment was conducted using the on-line labor market Mechanical Turk. Participation was restricted to Mechanical Turk workers residing in the United States. In total about 300 individuals participated in the experiment. The experimental design involves two phases and two treatments. Only the first phase—the reliance on intuition manipulation phase—varies across treatments. The second phase, in which we elicit ambiguity and risk preferences, is identical across treatments. Our two treatments are referred to as “High Intuition” and “Low Intuition.” In the first phase of the High Intuition (Low Intuition) treatment, participants are asked to briefly describe two (ten) situations in which they relied on their intuition to make a decision and it turned out to be the correct thing to do. The reasoning behind these treatments is that listing two situations where intuition provided correct advice should be relatively easy for most people, so that the High Intuition treatment should enhance participants’ willingness to rely on their intuition in subsequent decisions. On the other hand, coming up with ten such situations will be difficult for many people, bringing to mind some cases where intuition failed, so that this treatment should not enhance, or may even undermine, our participants’ willingness to rely on their intuition. The second (common across treatments) phase involves three questions. Participants know that only one of these three questions will be chosen to determine their experimental earnings and that ten percent of participants will be randomly chosen to be paid their experimental earnings. With the first two questions we construct a three-category ambiguity preference measure—ambiguity averse, ambiguity neutral or ambiguity seeking—and with the third question we elicit a continuous risk preference measure. All three questions as well as our method for selecting which participants get paid involve publicly verifiable sources of risk and ambiguity to allay concerns about trusting the experimenter.