Testing hypothetical bias with a real choice experiment using respondents' own money

Investigation of hypothetical bias in choice experiments has been done mainly in laboratory and only few in the field. Moreover, it is common practice to provide an initial endowment (money or coupons) to respondents. We carried out a between sample choice experiment in the field (in supermarkets) with three different treatments (hypothetical, hypothetical with cheap talk and real). We asked respondents to use their own money in the case of actual payment. This incentive-compatible mechanism proves to be useful for mimicking the everyday purchasing situation. Results confirm the presence of hypothetical bias and the mixed effectiveness of the cheap talk.