A contingent valuation study is conducted to estimate willingness to pay (WTP) for reducing the risks of skin cancer arising from exposure to solar UV light in New Zealand, where skin cancer incidence rates are among the highest in the world. A split sample design contrasts dichotomous choice (DC) with open ended (OE) methods for eliciting WTP responses. We attempt to distinguish theoretically consistent preferences from those which are procedurally variant and constructed. This is achieved both through explicit testing of the influence of available heuristics upon responses and a novel scope sensitivity test. The latter test addresses concerns that survey respondents sometimes fail to understand a specified change in provision. This is achieved by holding the good constant but making its remit vary from just the individual respondent to their entire household. A key feature of this latter test is whether the observed degree of scope sensitivity is not only statistically significant but also conforms to prior expectations. While our OE responses show clear evidence of preference construction, the DC responses pass both forms of testing. We conclude by arguing that the degree of scope sensitivity and its conformity with prior expectations should form a focal criterion for future validity testing.