SUMMARYIn this study, respondents were randomly allocated to three variants of the payment card format and an open‐ended format in order to test for convergent validity. The aim was to test whether preferences (as measured by willingness to pay additional tax) would be affected by framing the willingness‐to‐pay question differently. Results demonstrated that valuations were highly sensitive to whether respondents were asked to express their maximum willingness to pay per month or per year. Another important finding is that the introduction of a binary response filter prior to the payment card follow‐up tends to eliminate the positive aspects of introducing a payment card and produces response patterns that are much in line with those of the open‐ended contingent valuation format. However, although a filter will impact on the distribution of willingness‐to‐pay bids and on the rate of zero and protest bids, the overall impact on the welfare estimate is minor. The outcomes of this study indicate that valuations in the stated preference literature may be, at least in part, a function of the instrument designed to obtain the valuations. Copyright © 2013 John Wiley & Sons, Ltd.