The validity of risk estimates elicited through the Exchangeability Method (EM) has been theoretically questioned because the use of chained questions may undermine the incentive compatibility of the game even when subjects are rewarded with real monetary incentives. In this paper, we examine the validity of stated risks elicited via the EM by using a laboratory experiment. The risk under study is the presence of pesticide residues in apples. Taking inspiration from the de Finetti’s notion of coherence, we consider risk measures as valid if and only if they obey all axioms and theorems of probability theory. Our experiment consists of four treatments: in the first, subjects are provided with real monetary incentives, but in the second, they are not. Each experimental group is further sub-divided in two groups, in the first, the chained structure of the experimental design made quite clear to the subjects, while, in the second, the chained structure is hidden by resorting the elicitation questions. We found that the beneficial effect of real monetary incentives on the validity of stated risk estimates is completely vanished when people are presented with chained experimental design.